Q3 2023 Netflix Inc Earnings Call

Joining me today are co Ceos head Saran dosing, Greg Peters and CFO Spence Neumann our interviewer. This quarter is Jessica Reif Ehrlich from Bank of America. As a reminder, we will be making forward looking statements and actual results may vary Jessica let me turn it over to you now for your first question.

Yeah. So let's start with you Ted now that one strike itself are the writers Guild and what are the implications for your business.

Oh, Hey, Thanks, Jessica let me first say, we want nothing more than to resolve this and get everyone back to work.

That's true for Netflix that's true for every member of the N. P. T. P. It's why our member Ceos and prioritize these negotiations above everything else. We're doing we spent hours and hours with Sag aftra over the last few weeks and we were actually very optimistic that we are making progress.

But then at the very end of our last session together. The Guild presented this new demand that kind of on top of everything for a person subscribers levy unrelated to viewing or success in this really broke our momentum. Unfortunately.

But you should know we are incredibly and totally committed to ending this strike you know the industry our communities and the economy are all hurting so we need to get a deal done that respects all sides as soon as we possibly can.

In terms of the impact you know these are the times that I'm glad we have such a rich and deep and broad programming selection.

Uh huh.

We are programming cost himself a rise nearly every year, primarily driven by competition.

Competition for talent.

Competition for shows and films and you can see we've managed successfully through that year on year on year.

And the same is true for during Covid, when we were able to manage the slate through a prolonged and pretty unpredictable production interruptions.

But I really think we're you know we're not really that focus right there, but how this impacts much except for our biggest opportunity which is to continue improving the quality of the site are we're focused on that day in day out year in year out and I'm incredibly pleased with Bella and the team and the progress that they're making so if you'll indulge me for just a SEC.

I would draw your attention to the Q4 slate is an example of that are you know headlined by the return of the Crown for its final season. You know this is one of the most ambitious TV shows in the history of television we have a new season of big mouth history, a new season of a leap day are the larger Berlin, which is a spin off from our like I said, hey, propel her money heist.

<unk> and New limited series like all the light you cannot see from Shawn Levy, that's incredible and bodies from the U K and that's just on the TV side and the film side, one of our strongest quarters ever.

Enormous.

Science SIFI spectacular from Zack Snyder rebel Moon, a new film from David Fincher Killer and these films that just lit up the fall film film festivals recently like May December from Todd Haynes and Bradley Cooper is maestro the Doc feature American Symphony, that's all coming in Q4.

For family viewing to you know we've got a new animated feature from Adam Sandler got Leo that's hysterical.

Chicken run two which is the sequel to the most successful stop motion animation.

Film ever.

And a new series from the cocoa Melon World called Coco minimum Lane, a family switched from director Mcgee are with Jennifer Garner and at home. So it's an incredible slate something new and exciting for all tastes all moves all ages and we're just super proud of the team that they've been able to manage through this and and still deliver so much joy for our members.

One more on the strike related like the just the aftermath and you discussed at a recent conference getting talent more transparency.

Could you talk about what that looks like what are the new metrics talent will be paid on and is it even standardized across the industry. Yeah look what I've talked about there was heading towards a world worth that we're streaming data will be much more readily available member stream data streaming itself is not that exotic anymore, we've been doing it for <unk>.

18 years. So are we at the beginning we thought there was a hard kind of apples and oranges comparison to ratings and streaming and I think we've gotten to a place where it's mostly about engagement and that does capture the value of watching and that things will become much more transparent the way T. V's always had ratings in music has always had billboard and let's see.

<unk> always had box office, so it'll be much more common for the data to be fully transparent what I didnt mention though is that part of that of our reason for not publishing early it was part of our promise with creators at the time we started.

The original program.

Create a creator's, where it felt like they were pretty trapped in this kind of overnight ratings world and weekend box office world defining their success and failures and as we all know show might have enormous success down the road and it wasn't captured in that opening box office. So so part of this was the relationship with talent not just the business.

Aspects of it and I do think that over time people are much more interested in this where on the continuum today of how much data do we publish I think we've been leading the charge starting everyone down the path of a top 10 publishing our top 10 list and our annual wrap up list and everything that give a lot of transparency to the viewing and I just.

It will be more and more transparent.

Great, let's move on to page sharing have you identified most of the borrowers borrowers and can can you provide any help in how much more is left to go in the challenging completing the crackdown.

Sure I'll take that one and I'll start by saying, we're just incredibly pleased with how it's been going and you can see the progress from our membership growth in Q2 now in Q3, you can see it embedded in the revenue outlook for Q4, I think paid sharing represents the kind of difficult challenge, where we needed to balance both important relevant.

Consumer considerations with the importance of ensuring that our business got reasonably paid when we deliver entertainment.

And it's an example, where we leverage core execution of the capabilities that we've been building for over a decade sort of how you develop a good product experiences how do you solve hard problems through them. How do you have an iterative model, where you listen to consumers to tell us what's working and what's not so we've been excited about that but because it's such a challenging problems, that's where we're shifting essentially consumers expectation.

And what they what they expect from US we've always thought that making this change should be done in this steady considered way and so our plan has been to stage out this rollout.

Been delivering our product experience to different borrower cohorts. According to that plan and as a result, I think as you're alluding to there are a number of borrower cohorts, which has as of today have not received part of that experience and just explain that a bit I mean part of the motivation to stage. It out is based on technical considerations. So this is our ability to build feature.

Errors and improve model accuracy over time in a way that allows us to ensure that were accurately developing and applying our innovation and our interventions.

And as effective and as positive a way for consumers as possible part of that has been just the stage things out based on borrower behavior. So we want to show up with the right product experience at the right moment, that's more likely.

Likely to convert a borrower over rather than have them spin off. So we wanted to think about that for maximizing long term revenue. So we're going to continue the rollout for the next couple of quarters. I think you know folks are trying to figure out how much how much juice is left there and I would say, we anticipate that we will have incremental acquisition incremental.

<unk> for the next several quarters, we've seen that in the last couple of quarters. I think also worth noting that that was on top of also very healthy organic meaning not driven by paid sharing growth. So we anticipate seeing.

Seeing that for the next several quarters to come and then just stepping back you know there's a set of borrowers.

We're not going to convert we haven't converted yet we're not going to convert over the next couple of quarters, but that really represents how we think about paid sharing going forward, which has now become part of just our standard way of operating and we have many hundreds of millions of qualified households out there theyre smart TV household that we want to win over over.

The next several years and those borrowers we're not going to convert in the next couple of quarters represent that same group. So we gotta go after them. The same way we're going after people who'd never sign up for Netflix, which is having an incredible content offering an incredible value and get them. So excited that they just have to sign up.

Right.

Moving on to the recent advertising restructuring can you talk about why you made the management change and what you would like to accomplish.

First I'd say, Jeremy has done a great job getting us essentially from zero to where we are today. She has laid the foundation for the ads business She's hired and built a burgeoning team of leaders who in turn now are hiring the teams and people who are going to take the business forward, but it's an important time in I think a great time for aiming to come in.

And extend that great work to build on that foundation and drive our ads business to the next level and you know why.

Why am I, so specifically excited about Ami and the role first of all she's a high Netflix tenure employee she has been with the company for over seven years. She has demonstrated a really positive impact in great results in several different roles, but most recently as part of the studio and leading a big global team that is scaling very very very quickly which sounds familiar.

Sure you know when you think about what where we want to take our ads business second she's got broad entertainment.

Experience ranging from content licensing and distribution she's got business development finance strategy at Netflix and in prior roles. So I think when you think about that assemblage of skills and you think about the existing ads leadership team that we have that has got a rich rich history and adds in general and connected TV, especially you think about somebody like Peter Naylor, who.

Started selling connected TV at Hulu, that's a strong team to take our ads business to the next level.

And maybe I'll just I.

I wanted to maybe just restate, what we think the promise and the opportunity and sort of where we're at on ads business is and so first of all just starting off with you know this is $180 billion opportunity. When you think about linear TV and think about connected TV, not including Youtube, not including China, and Russia, and we think we're in a great position to win some of those dollars we've got great content.

The brands want to be next to where a safe place for brands to exist. We got great engagement from our members. That's a really strong foundation to work with but we got a lot of work and we know we have a lot of work to fulfill that potential among that work. We've said it many times I'll say it probably many times going forward, but scale is the number one priority we're making.

Good progress there this quarter, we grew our AD plans my membership, 70% sequentially quarter to quarter. That's on top of the last quarter, where we grew at a 100% quarter to quarter. We now have 30% of our new sign ups choosing our ads plan and our ads countries and we've done it by making the ads are offering more competitive we've gone to over.

<unk> 95 per cent content parity with our non as planned we've improved features like number of streams. The video resolution, we're going to keep doing that we're adding downloads now so we'll keep that good trajectory going and keep focusing on it second big priority for us as delivering features and products that advertisers want we've heard again and again I've heard it this week.

Adweek from advertisers top of that list is measurement, we've launched our measurement partnership with Nielsen in the United States. This month in October . So we're excited about that we've got a long list of other partners across other countries that we've got to deliver that same capability and so we're excited about getting that out.

We're also excited about new products. So we've rolled out our top 10 media buy were going to rollout our binge AD product later this year, we're launching more ways to buy programmatically through Microsoft that gives more buyers more ways to access our inventory. So we got a lot of work to do here on all those fronts, but we've always said this is.

The multi year build a multi year progress we got a lot that we've got going on and we're excited about the future to come.

So now that you've phased out basis for new subs and you're getting extra members are paid more precise from Katherine Shan crackdown and you've introduced advertising in 12 countries could you talk about the outlook for on in 'twenty four and beyond.

So let me take that one Ben.

Alright, you wind it up I mean, thanks, Jessica So I would say just generally when we think about 'twenty four and beyond I think about it as our revenue growth profile in general and you talked about this recently, but we expect a more balanced mix of membership and arm growth in 'twenty four and beyond 24, So just looking at 'twenty four specifically.

As Ted talked about we expect to have a great slate to drive the business forward and we expect to continue to do things like add extra members grow our advertising revenue.

As Greg discussed and in addition have some pricing adjustments you saw that in our letter all of those things will drive arm. So 23 was a pretty unusual year, where essentially all of our growth came from member growth and going forward are more broadly not just 'twenty four and beyond we will grow our business by continuing to continuously improve our service increasing engagement increasingly.

Satisfying current and future members and now that you know as Greg discussed and we've got an account sharing solution, we have a more clear path.

More deeply penetrate that big addressable market of.

Half a billion connected TV households, and growing and with our continued plan evolution in pricing sophistication and all that hard work on our ads business will keep getting better monetizing that big and growing reach and engagement. So we believe.

We've got a long runway for growth in both kind of more membership in higher arm overtime.

A more balanced way than what you saw this year, which was again a pretty unusual year.

And then you you touched on it Greg touched on on scale and advertising, how do you get to scale as it all through pricing.

Pricing changes and what would you consider scale.

Yeah, I think it's important to note that scale isn't it's not a binary a condition right. So it's not like you are suddenly I had one more member and you become a must buy situations. So we become increasingly competitive with increasing reach.

It's also I think worth noting that it's different in different countries and is largely based on whats the competitive channels and what's that competitive dynamic.

So having said that though we carry several relevancy targets are on a per country basis think about this as essentially percentage of market penetration that helps us focus and drive their rate of growth that we desire.

And we've got more work to do to get them. So I mean like we're not satisfied with the scale that we're at in any country that we're in we want to be bigger and we know we can be bigger I think theres a variety of techniques that we can employ to do that our pricing and thinking about how do we factor in what's the optimal pricing for ads no add that's part of what we're doing and thinking about planned evolution part of it.

What I mentioned before which is feature set right.

The things that.

<unk> want to sign up for part of it too is actually just educating consumers I think what we are seeing is in some of our countries consumers think about an ads experienced mostly anchored in linear and what their expectation around AD load our frequency rates are and to some degree actually some of our streaming competitors haven't done maybe as.

Great job in building and adds experience, which informs that expectation as well the part of it is just educating consumers about what the actual Netflix ads experiences. So that they can think about what's the right choice for them do they want a lower price with ads and what we think is a great ads experience for consumers really or do they want to pay more and.

Skip ads. So it's all those things coming together that ultimately drive us to the several multiples of scale that we're at today that will be satisfied with.

One last one maybe on advertising before I move on to margins.

And you don't you you mentioned a lot of the innovative offerings that you plan on an expansion you just it.

It's very unique it's different.

When do we get to a point of Lindley Youll have a point, where it's targeted addressable so it's really relevant for consumers and so they would want to see the ads.

Yes, well, so we're working with Microsoft right now on targeting so you'll see that rollout in the near future.

And that I think is the first step is how we think about increasing targeting relevance through both a combination of product set. So what are the types of AD products that brands can buy that yield increasing relevance as well as improving our sort of sophistication on what we might call targeting from a digital perspective, which is basically matching.

Consumers, who are most interested in that particular brand's message.

I spent I guess this one's for you on on margin, but can you elaborate on areas like AD Tech content spend but you you did talk about content spending you know a lot of them, but any other meaningful investment areas something that maybe we're not thinking about.

Sure. So let me step back a bit with some quick context. So first.

Jessica we set margin targets there, they're our best judgment of how best to grow the long term value of Netflix and we're trying to balance investment for future growth with near term profits. So for instance.

After investing heavily to launched global in 2016 global Netflix we wanted to take a disciplined approach to building profitability has as we grew revenue because.

We felt one he was a good way to build that profit muscle across the company and two we understood that investors, where they've been pretty patient with us. So we wanted to demonstrate the scalability and the health of the business model and so that took us from.

It was like 4% Oi margin and operating income margin business in 2016 to our current roughly 20% margin. So we think a pretty good indicator that at scale streaming can be a quite good business.

Now stepping back there's there's no change in our financial objectives and also no change in our long term margin expectations, including the fact that we see a we don't think we're anywhere near a margin ceiling. We've got a long runway of margin growth. So again no change in our objective of no change in our long term margin expectations.

But at our current profitability and scale, we think it's prudent to balance that historical pace of margin improvement with.

With growth investments. So you asked about growth investments, we think we've got a lot of places where it can continue to invest plenty of room to invest further in our existing content kind of categories, where a small share of viewing in every country in which we operate plus building out those adds capabilities that Greg talked about are alive, offering and new content categories.

Games, so theres plenty to do but all of that said, we will continue to drive healthy margin expansion, we expect roughly <unk>, 22% to 23% operating margin in 'twenty four assuming no material swings in FX. So that's up from our current expectation of 20%. This year, which is at the high end of the range that we targeted in the beginning of the year.

So so again just like we did in the past going forward, we will take a disciplined approach to balancing margin improvement with investing into our growth, we actually put a chart and at the end of the letter that shows how we've managed that balance historically growing content investment profit margins and cash flow and you should expect that we'll catch.

That same discipline going forward as we invest and grow into that big opportunity.

And how does licensing content from third parties planting or overall content strategy. It seems like you've had incredible success with third party content in the fall.

We'll always have but in the last year things like suits, a band of brothers and you mentioned it in the letter and the you know if you could just talk about the third party.

Yeah, Yeah, Yeah licensing third party content has always been part of our strategy and it's something we've been really great at being able to do is match that audience. I think suits is a great example of the impact of the Netflix effect that we could have because of our distribution footprint and our recommendation system, we were able to take suits, which played on cable in it.

Played out and other streaming services and on pop it right into the center of the culture in a huge way not just in the U S. But all over the world. According to the Nielsen charts and suits was the number one watched streaming streaming series for 13 straight weeks. That's like that's a that is a record in front of it for Nielsen. So this continues to be important for us to.

A lot of breadth of storytelling to our consumers of a wide range of taste and we can't make everything but we can help you find just about anything and that's the that's really the strength and I do think that I'm looking at you mentioned a band of brothers, but in that H B O deal. We had insecure we had bottlers that came out and they were very successful on Netflix and.

They popped into the top 10 on their originating network for the first time, so that was on the street.

Just on their streaming service, which is really powerful.

And I think we have more to come with six feet under in true blood coming and not just on the TV side, but we're also proud to be able to bring our movies like Super Mario brothers and spiderman across the Spider verse from other suppliers.

In one way or another we're in business with nearly every supplier, including our direct competitors and I think that we bring a ton of value to them and I think when you think about what happens in that show runs in and becomes a huge success on Netflix it has lasting value I mean look at the the value. We created that still continues today for shows like friends and the office.

Full house and Gilmore girls and all these other shows that really found found an audience on Netflix even after they are more or less played out through traditional models.

Yeah.

And somewhere on margins for you, but you said in September that long term margins will be thankful that he said it was similar to other networks, which historically have been in the 40% to 50% range could you help us think through the ramp in margins over time.

Jessica I'll, probably disappoint you as I have in the past on this we're not going to put a long term number out there as I said, we don't see any any ceiling and in near term ceiling to are our long term margin potential we've talked in the past about how we're going to feel our way through to those kind of.

Long term steady state margins, but we think we have a lot of things working in our favor we have a very scalable business model you see that and you've seen that play out over the last handful of years and continue to do so as we produce content all over the world for big local impact, but also with the ability for those stories to you know through creates subs dubs discovery to reach me.

More and more people in and to be enjoyed around the world. So it's a very scalable content model.

Global network at scale that has in many ways has not been seen with legacy Entertainment networks. So we think we've got a long way to go as I just talked about we want a balance.

Does increasing profits and in in the near term with investing into that long term opportunity. So still a lot of runway. That's a set of benchmarks you can look at if there's others as well, but suffice to say, we think we've got a long a long and healthy runway in terms of growing margins going yeah, I would add to that I'm just gonna also totally agree with what Spence.

Which is again a lot of opportunity to grow margins, but profit dollars also matter too right. So as we expand into big new addressable markets like advertising that Greg alluded to our gaming also right. So you know those open up big new sort of areas for us to expand into.

And we intend to grow margins too, but we also want a lot of profit dollars as well. So we're not narrowly optimizing just from a percentage margin.

Question.

You announced some price changes today in premium and basic and several countries and then more to come can you provide a current view of price increase or a timeframe for the standard here.

Yeah. So you know as you.

You know our focus on plan evolution over the last 18 months has largely been about paid sharing and now that we've rolled that out we broadly see the benefits you know as outlined in the letter that's become a normal part of our business, which then allows us to return to our core approach to pricing and that approach that philosophy has not.

<unk>.

We look to wisely invest the money that our members pay us deliver back to them more amazing stories more entertainment value and then when we think we're doing that well occasionally asked them to pay a bit more to keep that virtuous circle spinning. So hence the changes that you noted and then we've announced in the letter I think it's also worth noting that we seek to.

Have a wide and even wider over time range of price points with a corresponding set of features of course that allows entertainment fans from around the world that have different needs.

To be able to access the great storytelling that our creative partners are doing at a price point that works for them at a feature set that works for them.

Part of that widespread is the low entry price point and that's why we're keeping that low entry price point static as it is so we think that this six to $99 in the U S for 99 pounds in the U K five nine in euros in France, I think it's just an incredible entertainment value and if you think about the breadth and the variety of story, telling me that we're offering.

Whether that's compared to our streaming competitors.

Compared to traditional pay TV, certainly I'm, even the price of a movie ticket. We think that's just an amazing offer and our goal and plan is to continue to be a great entertainment value and beyond that we're not going to comment on other price changes or other changes on tiers, we'll sort of find our way based on that philosophy and see when the right time.

Ask customers to pay a little bit more would be.

One more question on the pricing would you given the price increase for for just premium than basic not standpoint, do you expect any or advertising do you expect any movement between the tiers as a result of these price increases.

I think pricing always results in a bit of movement between the tiers more of that movement is how people are signing up so we see that as more what it influences, but also it'll it'll influence plan changes as well, but generally plan changes tend to be relative our plans tend to be relatively sticky. So I would imagine that there is a.

That momentum will continue.

I don't see a letter today says that you stated that you will spend $17 billion in 'twenty foreign content spend.

Up from 13 billion of twenty-three, obviously that was somewhat strike impacted.

How should we how can you help us think through how content spend will grow beyond 'twenty four.

What is normalized growth.

Well you can see that we've done is we wanted to grow the content spend just.

So about a half a step ahead of the revenue to create the value proposition for our members. So the more we put into it and there's a lot of it is tied to the ability to create inside of it out of that pool and I would say one thing if I could if you hum.

This past quarter, we had this really remarkable story about something that we could do with dispense talked a little bit about the kind of scale of the content spend but this show one piece.

<unk> piece is something that is a very unique property to created 26 years ago by a cure or it's a one it is over 1000 episodes of the animated series based on the Japanese manga, it's a it's a it's nearly sacred IP.

And we were able to with our Japanese creative teams and our American teams getting together working with our partners at Tomorrow Studios in the short run or Stephen made to adapt this into a show that the world fell in love with and when I say that is we've got this show is number one in 84 countries around the world, which is something that.

Stranger things because it didn't do that Wednesday didn't do.

And it's so rare for an American English show to be that popular in Japan, and Korea, Brazil and in the U S. At the same time and the other fund part of it is Inaki could doi who stars and are in the show and it was one of the most.

Most difficult casting challenges in the history of our original programming, where she was going to play Monkey Duluth and he was right under our nose right in our talent family. We discovered them a couple of years ago might have been this great show at our Mexican series called who killed Sara.

And then we were able to cast him in this and now he's a global superstar. So this is that kind of thing you could do well thing that's hard to copy and gives us kind of competitive running room. It from us from our competitors being able to do that more and more I don't mean, when I say that I don't mean, making things more global I think making things that really resonate for the core audience.

And usually local audiences want very local content and in this case the local audience is the fan of one piece, which was very discriminating and we had to please them first just like our original shows in Spain up to really play as the Spanish cost customer first so we can do this we spend the money well we have impact with the spin.

And and we grow it as we grow revenue.

And then maybe Oh, sorry go ahead, Jessica I was just kinda fill of a kind of a little bit on on Ted's point on that kind of trajectory of content spend.

So and we talked about this a little bit in the past. So first in the letter we talked about the fact for 'twenty four we hope to get cash content spend back up to at or near that $17 billion level. The biggest swing factor is going to be when the sag aftra strike resolved and so that'll get us to a you know a cast of P&L ratio kind of close.

Sort of one to one one times and so we're not putting a specific number out there for free cash flow in 'twenty four what that gets us to when you think about the combination of our revenue growth outlook, our margin guidance and target cash content spend.

We will deliver substantial free cash flow in 'twenty, four and then going beyond that we do expect to tick up our content investment over time.

As we also prove out sustained a healthy revenue growth. So you know assuming.

We've talked about I think in the last call assuming no big expansions, we'd expect our cash to P&L.

Ratio of content spend cash to content amort in the P&L to be roughly 1.1 times. So that's kind of one way it folks are thinking about how to model our growth in content spend if we as we grow our revenue as we improve our profitability, we should see both increasing content spend but also free cash flow growing nicely over time.

Yeah.

And then just one last just a follow up for Ted though there's so much going on in content right now can you maybe.

Jumping in investment priority, it's like how do you think about whether it's local language film television you need a lot of deals with some third party film companies TV companies could you can you give us some colorants and on how you think about content spend yeah. We always have a lot of plain spending because I remember they've got such a.

Such different tastes different desires, and we're trying to place them all.

And like I say trying to find that person, who really fell in love with us for prestige T V. And then discovered love is blind.

That's a pretty common household to be honest with you. So we've got to be able to be good at so many different things in our partnerships.

I'm, assuming you're talking about a sky dance in this case is really helps us find and keep up that scale as we grow.

So we're really thrilled with our success in animated features it's a very long cycle of development and production, sometimes it could take a decade to deliver a really great animated feature film and as you know, we moved pretty fast and we've been moving pretty fast and no single company has ever really successfully launched more than two animated features in a single year.

Year. So we wanted to just that they'll helps us to to complement the work that we're doing like you saw this year with Leo and chicken run coming out and in pneumonia that already came out. So we've got a very there's a ton of appetite. If you look at the top 10 animated features of since Nielsen has been tracking movie was.

And seven of them are animated features so there's a lot of appetite for animated features and we're committed to that part of the business and we do that through a combination of licensing or partnerships and original production and original creation and not just in the U S. But all over the world. So we have to find that right balance of invest finding the right product market fit with.

It helps us grow those territories and most importantly helps create a value proposition for consumers and they can say hey that that's what I pay for Netflix I can pay a little bit more because I've got so much value there and I'm spending so much of my time there. So if you think about the you know.

For the last 37 of the last 38 weeks of this year Netflix has had the number one streaming series on an.

All in all of streaming and for 31 of those 38 weeks, we've had the number one movie too and in any given week, we might've had the number one two and three so we really we've got a lot going on and we've got to stay focused on continuing to improve the value proposition to consumers, which drive the numbers that we've been talking about on this call.

Spence you announced a very significant increase in your buyback today should we think of the $2.5 billion buyback in the third quarter as instead of a run rate moving forward.

I wouldn't I wouldn't kind of read through to that Jessica we had kind of slowed down as we as the business slowed down and we wanted to we talked about the fact that we had less less than typical forward visibility into our into our forecast numbers over the past year or so as we were.

Looking to Reaccelerate the business and also rollout paid sharing and now much of that is behind us as we've said and we've got a better view going forward and so we ramped up our our repurchase because we had built up some cash on the balance sheet as well are our target a minimum cash is roughly two months of revenue. So you know plus or minus six bill.

Dollars of cash that we look to hold on our balance sheet and we had gotten ahead of that are still a little ahead of that so that that's really what we're managing to is to one primarily drive the business forward grow the business expand our cash flow and then cash excess cash builds on the balance sheet to return it to shareholders. So we put a pretty specific.

Target out there of roughly two months of revenue in the form of cash on the balance sheet.

That's.

The way I think you should think about what our pacing will be over time.

Moving onto gaming it feels like almost like the way you describe advertising like a walk crawl walk run approach what is the near and mid term strategy and gaming.

Well, let's start with the Big Prize I think that's the better way to look at it which is you know games is a huge entertainment opportunity. So we're talking about $140 billion worth of consumer spend on games outside of China and outside of Russia and from a strategic perspective, we believe that we can build games into a strong content category Levy.

<unk>, our current for film and N series by connecting members, especially of members that are fans of specific Ips with games that they will love and I think it's worth noting that you know if we can make those connections and as we make those connections as we're seeing we're essentially sidestepping the biggest issue that the mobile games market has today, which is high.

Do you cost effectively acquire new players so.

So that's the real proposition and we think if we deliver that we give members great games entertainment experiences that they love at sufficient scale, then we leverage back into the core business. We increase engagement increase retention of increased value delivered those all drive our core business metrics and I think it's it's actually just a very natural extension of what you were just talk.

And with Ted about if you think about the range of content that we're offering a variety of content and entertainment that we're offering games just adds one extra layer to that variety and that depth.

And we're also seeing I would say back to moving more to your at your short term and midterm. We're also seeing performance metrics that support that this fundamental strategic hypotheses are sound. So games engagement right now on our service drives core business metrics in a way, which is incremental to movies and series so but the main <unk>.

<unk> ahead of us to get to your mid term is that our current scale and frankly, our current investment level are both very very very small relative to our overall content spend and engagement. So now our job is to incrementally scale to the place where games have a material impact on the business. We've got ambitious plans there we want to really grow our engagement by many multiple.

<unk> of where it is today over the next handful of years and we can see how to get there looking a layer deeper at the title level. We see you know some titles are really working for our members and they're working for our business. If we can do more of those we know we can scale into that proposition.

Gotta do that through better title selection based on everything that we're learning we got to do on a better product features to maximize connection with the audience for any given title and we have to do it by gradually improving consumer awareness, which was we've seen is when we launched other content categories. You can think about unscripted or you can think about film that broadly lifts overall engagement.

Metrics as consumers learn that we're a place to go to to find games.

I'm excited about what we got going on in Q4, we're going to launch some big high profile titles were sort of keeps that drumbeat going we got dead cells. We got football manager 2024, we've got money heist think about connections with our IP, that's coming in Q4 as well, that's causing the propel for folks who are.

And that language.

We also have Virgin River coming in Q1, so as you pointed out this trajectory is not dissimilar from what we've seen before when we've launched a new region think about Latin America, we launched a country like Japan, where you know traditional western media companies have struggled or we launched new genres like unscripted, we got a crawl walk run them you build it but we see it.

Tremendous amount of opportunity to build a long term center value of entertainment more entertainment value for our members.

So it's got great experience for the Super fan.

Get themselves in the universe in between seasons of a show it's really exciting.

Jessica we have time for about two last questions. Okay. Okay. So the first of the two and sports you're creating the Netflix Cup tournament to be next month.

Is this a change in use.

Sports strategy at all or how should we think about that yeah I knew that I knew this is gonna be Jessica.

[laughter] forgive me because we are in the sports business, but we're in the part of the sports business that we bring the most value to which is the drama of sport. So look at the success, we've had with drive to survive I look at the success, we've had with tour de France quarterback full swing untold.

Most recently with Beckham, David Beckham is one of the biggest stars in the World and his documentary on Netflix brought him almost half a million new social media followers and a week. So we're having a a.

A big impact on sports through the thing that we're most great at which is the drama of sport.

The Netflix Cup is a live event that actually brings together the cast of drive to survive in full swing and puts them into a live golf tournament that we are going to stream live on Netflix on November 14th and it's I think about it as a great way of extending those great drama of sports brands that we've created but no core change in our.

Live sports strategy or licensing in live sports.

We are investing heavily in increasing our life capabilities. So that as we are.

As the demand grows for that and we find different ways. They live their lives and this can be part of the creative storytelling, we want to be able to do that a big scale.

There was some news also today or I guess on comedy, but my last question just to stay with the western Central and you've talked a little bit more recently about your in ancillary businesses, including the Netflix House can you touch on what that looks like over time and will it be will it be thinking investment area.

More importantly will it be a contributor.

Look at.

This this.

This initiatives inside of our consumer products and experiences group.

Today, they run these successful businesses, where they travel these live experiences all over the world and fans engaging them in ways that would shock you people loved these things so much they show up dozens of people have proposed marriage in the breath of Richardson ball its really important in a way to kind of deepen our fandom a way to express fandom.

Are you kind of see it on a large scale with theme parks. These build outs are not going to be like a theme park, but both of them that they won't have that gigantic capex.

And they also we expect that fans will go multiple times a year and not just once every couple of years and it's a way to take a business. That's really good at growing our brands and strengthening our brands and today doesn't have a big startup and shutdown costs as they travel around and put them under one umbrella.

We can add a little technology and make it a really phenomenal experience from being.

As part of the money highest.

Cape room, or the stranger things experience or the squid game challenge all those different things that people can do live together and have a lot of fun and they can also go to the Netflix bites and have a food experience with all of the Netflix food brands. So it really kind of strengthen the brands and strengthens the excitement about the things that people are watching on Netflix and falling in love with.

And it gives them a place to go and express it is.

It's not a material investment relative to the core to the big business that we're all in but it's a great way of building it like our consumer products business.

Right well Jessica Thank you very much for your questions and we appreciate everybody tuning into our earnings call and we're looking forward to chatting with you all next quarter if not sooner. Thank you.

[noise] Mohan and where innovation money empower Golan in Silicon Valley.

This is Bloomberg technology with Caroline hide.

Yeah.

[laughter].

Yeah.

Okay.

If I go to the meal.

Ludlow in San Francisco. This is pulling back technology coming out in the U S comes down on China's chip.

Sweeping update the export controls who will discuss the impact on the semiconductor industry.

Plus we're bringing the latest updates from the trial a sandbank men freed as the third member is in a circle testifies against him and we'll continue our conversation on the technology ecosystem right.

Neil technique continues.

Checking on these market.

Choice at the moment today, and we're seeing the NASDAQ100, I'm focusing on sound about 10th of a cent stock sometimes struggle in the face of rising barnyard not see we saw some other areas captivate back is not coming out with the earnings that beat.

And Saks as well.

That had been anticipated in terms of its profitability targets.

Are you pushing out months would answer that 2007 high so we're anticipating strong data that we've just seen out of you asked whether it was overall the macro picture looks a little bit better and ultimately not mean for the federal reserve I think unfortunately, it just means that we're looking at I'm looking at what's happening overall in the tech space in China, we've talked about and read it initiatives.

Catch up meetings that are happening with world leaders, but also China of course, that's all a Chinese dominated economy, while the U S tied to impede some of its technological knowhow when it came it down something closer to 10.

Taking into that one and let's look at some of the individual areas of risk assets that we'd like to thank you for that and I wonder if I could tell them what's happening in bitcoin because we have had once again a wild ride.

Let me think on the site and bitcoin and suddenly he sees some sort of mid to maybe even a.

Place of refuge.

Yet to be seen as safety that suddenly what seems to be being talked about by they had a bad crop.

Overall, we're saying it just talk about a quarter of a cent as he sees fit.

Around that all important 28000 level.

There's a lot in the news cycle, that's pushing individuals stocks lucid the EV maker sorts production dropped 30% quarter on quarter really casting doubt about its full year production targets. The high end Tvs that stuck down almost 4% also down six tenths of 1% is out with a new Apple pencil go Alan's Apple's web site and see the details not doing much.

To support the stock and there is some hangover from that counterpoint research that came out 24 hours ago, showing that early sales of iPhone fittings in China are not doing so well relative to iPhone 14 last year and Amazon are down four tenths of 1% really strong retail sales data here in the U S for the month of September not doing much to support e-commerce related stocks.

The top story in D. Taught me that is Nvidia really knows full moved downwards. After the U S confirmed it is expanding export cards relating to CIT technology that move now 3.5%. It had been at 1.7% on track for its biggest drop since early may that said despite shedding six.

Need a $70 billion market cap this is to a $1 one trillion dollars stock.

Without the a 880 800, the specific China kit that it was exporting let's get the details out of D. C with rain days Mackenzie Hawkins and Mackenzie. We knew this was coming Bloomberg had reported it but how is this impacting them video specifically.

So what we saw today is done by the administration released updated regulations on these sweeping export controls that they imposed on China last year and in response to last year's carved Nvidia design chips, specifically for the Chinese market that went right up to the threshold that the administration has sought and now the administration has tightened that threshold take half.

The church that Nvidia designed for China, meaning that they can't sell those chips, there without a license and so their socks I had this morning. It wasn't just didn't video that was being Featherstone from a business model perspective can you talk us through some of the AI companies that are also being targeted in China.

So our so in addition to tightening the threshold for graphics chips like in videos. You also saw the administration to AI firms that are Chinese AI startups to the entity list, which basically means overseas companies have to get a license to export and trade with those firms.

And you know one of those companies have said, we're going to appeal that decision we strongly disagree.

The Chinese foreign Minister said yesterday that you know we don't want that you asked her weaponized trade issues.

But this is the Biogen ministration clamping down on what they see as rapid advancement in China's chip industry and beyond that we also have to look at the equipment makers because it's not just the axle manufacturers, but also the highly expensive and technical equipment required to make them and so one of the key players there is a Dutch company as an hour.

And they've said that the restrictions and the restrictions that were announced this morning are going to impact some of their China facilities as well.

Our waitress showing on the screen Mckenzie as an analyst response to this these export control measures will likely have an impact on the regional split of our system sales in the medium to long term what was interesting about this the whole point of codifying and updating these cabs was to address the loopholes in gray areas from last October .

And yet I think I'm right in saying that officials in the U S side Mackenzie didn't specifically explain which chipmaking equipment was going to be caught by the expanded controls.

So they released the tax of the role of this morning, and you know companies industry analysts trade organizations are rapidly going through that tax to identify which cities like what specific pieces of equipment are affected and so and as in all cases you'd be looking specifically at less advanced what are called deals the machines, which are a lithography machine.

And that actually help you cut wafer cut shifts into a semiconductor wafer.

And then you have to coordinate with another lens on that policy because ethanol is a Dutch company. It's one of the most valuable companies in Europe and one of the main loop holes of the initial export controls announced last October was that Washington went ahead, but they haven't yet gotten Amsterdam in Tokyo on board, where some of the main ship equipment makers that I've had a great global perspective Mackenzie Hilton.

Thank you so much for jumping on that news of day. Meanwhile, they're gonna stay is in Washington, and I didn't talk globally as well because it hasn't mining is such traveled to Israel Tomorrow and any moment now of course, the last house itself will hold a vote on election can you speak of it is busy down that Kenny lines as always all across this first of all nice to the president and <unk>.

Dissipated threaten the name and how are we anticipating it to be.

Taken by the rest of the administration and indeed by the Republicans.

Well of course this is something that president Biden is doing at the invitation of Israeli Prime Minister Benjamin Netanyahu, it's going to be a literal physical manifestation of the U S to support for Israel, a shell of rail solidarity and it'll trip that the president is making at great personal risk. He is heading into an act of war zone, but there's a number of things.

He is going to be trying to achieve in addition to showing support for Israel and talking with Israelis Benjamin Prime Minister.

Benjamin Netanyahu Prime Minister and other leaders about what it is they need in terms of age. So it also goes there with the knowledge that there is several Americans that are understood to be hostages at the market would like to see them are rescued as we are on the cost. It seems evident daily ground invasion in Gaza and also likely is going to be bringing with him a message of restraint.

Concern for the innocent Palestinian civilians in Gaza, making sure that while the U S. S. Israel does have a right to go after him off that they are taking every measure they can actually protect civilian and innocent lives and then of course, there is just kind of the broader deterrent effect up and being there and they do not want this crisis to expand beyond just Israel in Gaza and as a warning to other countries.

Iran. For example, other groups like has it a lot not to widen this conflict any further and that's also why we're seeing the president traveling to Jordan asked after he is in Israel to meet with the leaders of Egypt in the Palestinian Authority. In addition to capital left.

The administration's attentions on the Middle East Kayleigh, but also high and we have this house vote on the speaker ship, what are we expecting to happen imminently.

Well it is literally going to take place any men at the house convened at noon and we're going to see if Tim joining has the numbers. He of course is the speaker designate of the Republican conference, but when he was given that on Friday, and we're seeing what's happening here in the house underway.

Just not clear that he was going to have the votes that he means 217 is the magic number who can only afford to lose four votes that were more than 50 here on Friday suggested they would not vote for him on the floor, but he's been whittling away at those numbers over time. He had a number of key flipped the likes of Mike Rogers can Calvert and Wagner all of these kind of defense appropriations.

That he was able to bring onto aside that there's still a handful of Republicans Carlos Hernandez, Mike Lawlor, Mike Kelly Marianne. He has the large hill, who still have said that they will vote no farhan. So it's going to be a matter of whether he can actually get there on the first ballot and if not if he can do it on any valid he's going to have to try to bring more people onto his side.

Knowing of course that we are working with a number of clocks here Israeli aid.

They need to get funding to Israel with congressional approval on the one hand, but also trying to avoid a government shutdown come November 17th on the other and I think to that point, you sort of shift the conversation back from famine and see if this thing on the house and also what that does mean for well.

Direction of travel with Israel at the moment, just take as to what we know in terms of the latest thinking suddenly from Anthony of Lincoln as well.

Well, obviously, there is bipartisan support to Caroline for giving Israel, what it needs and for Congress really passing that eight if it needs to knowing there I think as this thing in the morning for providing administration already has the money that has already been appropriated resources. They can give to Israel without congressional approval, but if there is something that requires congressional our Peru.

Like supplemental funding request, that's where it's going to get tricky. If we still have no speaker of the house and keep in mind that it's not just Israel funding. The administration is pushing for and trying to secure its others as well, including funding for Ukraine, which is a much more difficult topic among Republicans specifically in the house of Representatives when there's been growing talks here in Washington about the idea that is.

And Ukraine funding could be tied together, but that is something that Jim Jordan. If he doesn't do get sneakers Apple could have a hard time getting through the house. Knowing there are so many members of his party that are opposed to any additional funding going to Ukraine. So it's going to be a thorny subject to navigate ethene crane in Israel do get lumped together, even though there is that's important there for Israel.

<unk> 9000 came across all of the moving parts that are currently occurring in Washington, and I mean, we've been talking about instead of costs from the politics from the conflict, but at the moment all of a sudden implications within technology and actually this is a notable news out but we understand this is where I live in talks with Elon Musk Spacex right to set up a styling satellite network.

It's also communications.

Yeah. This isn't an announcement from Israel Communications ministries, and potent Tonight. If you go onto styling website and look at the massive coverage styling is not currently available in Israel or indeed in the Gaza strip of the West Bank, we've not heard from Spacex will receive comments from them on the story, but the idea that the Israelis are outlining is it will keep those towns on the frontline.

With a reliable internet connection as this war with Hamas continues the other story that we're continuing to follow related to the Israel Hamas wall Y Combinator as Gary Tan Sequoia capital as Ravi Gupta and several other prominent bench cafes have cancelled plans to attend web summit, the biggest technology conference in Europe and it follows.

Bye the conference is C O that slammed western support for Israel. After the Hamas attacks and Caroline important to note that this morning, Patti caused cosgrave the C. A wet summer through our blog posts and issued an apology for those original comments that he'd made I think on X. Yeah. He sent me didn't walk back from southern.

His initial viewpoint and I think it was notable that they do save some vcs have backhaul says and I ran some European venture capitalist as well from attending the event, but I think how should wed summit came out and said actually Monday had a lot of people signing up to attend the event the busiest month of it that had in terms of attendance registration.

And indeed, some species has signed up to attend as well, but overall the Swedish does just show that people are.

Having to navigate really how they communicate with indeed their own customer base, how they communicate with the board of World Onyx and Amendment Azzedine Panicles gave himself has happened today.

Yeah, I think he probably sit onto that platform like saying he's going to be taking a time out from the platform where he did recently made as these nine Meanwhile, nighttime out processes like anything the latest on the F. T X trial and what it all means for the featuring either.

Hey, Thank you Alessandro rather see a global blockchain business Council.

That's within that technology.

[noise] [noise] Michel was saying he is a former director of engineering at FCA.

It's facing cross examination of the sand island clean trial today.

So the earlier questioning things said that he was intimidated by back on freedom at times have been answered.

Thanks, Brian for more on the way in which these testimonies are unfolding.

At the moment, what he feels like this is pretty rocky relationship suddenly towards the end of F, Texas, well pull out the scene.

100%, yes, we're getting an inside look at what happened in attacks during the collapse both in the offices and in the $30 million penthouse that Sam Bergman frame shared with as close friends and executives.

Kind of one interesting pace, particularly from the sort of the prosecution standpoint is what <unk>, saying is saying about when he learned about the financial difficulties that MTX wisdom, what do we learn in the courtroom. We know that it was an extremely stressful situation forehand thats, just really devastating and he did he made a plea deal.

Early on he did flip early on and free so.

So we also know that he had a close personal relationship with sandbags and freedoms younger brother this was more than professional this with a personal connection between the two.

What's interesting is made at the timing of this and as Sesame Singh says he only learned about the 8 billion dollar hull and the Companys balance sheet two months before everything fell apart where are we at in this trial.

Todd is a reporter I'd been in those courtrooms I know to kind of weigh the evidence that but are we near the end of the process at least of what we expect to come yes. Our latest updated estimate to the verdict is around November eight but again. This is an evolving situation things can change we do have that potential witness list, but that could be adjusted.

At there are still many colorful characters and people close to Sam. Thank you Humphrey horse that to testify and I think what we're also getting US a story of it seems as though whether prosecution is trying to identify so called flashing ethic.

Well done by sometime in feed and it feels that they are happy we're having some thing some calibration of that but he's saying that some of them and phased approach to spending on investments in real estate and celebrity endorsements was excessive but also just the fact that they were kind of left out of the live for many of these deals until after the fact.

Yeah. So a lot of what the prosecution has been doing has been showing the luxe lifestyle that sandy winfried any other MTX executives percent. Meanwhile, the defense has tried arguing that Sam things and freed was irregular math nerd Ah that he didnt like the slight of luxury but really we've seen with things testimony about living in this.

$30 million penthouse in the Bahamas that really all of these executives got to enjoy a life of French's, probably makes had another terrific up to the minute reporting here NSF. Thank you very much just keep the conversation going with Santa Rose C. E. O is the global blockchain business Council, a Swiss based nonprofit with more than 500.

Institutional Mendes and essentially is dedicated to further adoption of blockchain technology, which this trial kind of puts a macro spotlight on it's kind of an interesting real time play on the rule of law right. We are seeing that happen in this country as you sit there representing your industry. What do you think the out.

Come will mean for your industry.

So rod it's Frank.

And the trial underscores the importance of enforcement.

But to do its job to read out our financial crimes and as these guys.

It's proven that in this trial that a financial crimes happening.

The U S is based on rule of law and that is a good thing and we look to see the outcome of that shortly it seems and for the industry I will say this.

Less height.

Last night.

Bad actors that better or good innovators and.

Those who are trying to do the right thing, which includes responsible innovation and consumer protection going hand in hand.

Talking of noise, there that wasn't been annoyed surrounding well future sign off the spot Etfs yesterday, it made for a little bit of volatility in the Oh Gee is except to say well that is bitcoin because there was some expectation that the SEC might sign off on blackhawk's ETF propose a lot and of course that was something down by the SEC itself, saying you know you can't believe everything on the Internet.

How much all you're navigating this current noise and expectation in the market.

So with the organization that we have which has fortunate 500 across to layer one blockchain crypto firms to start ups, we have to navigate a world where we focus on what is the problem people are shopping for how is this technology actually useful for society and frankly.

What is the positive impact to society. So for US, it's really about getting down to Greg and butter, what real world problems are being sought for and how can be technology tools used for furthering our solutions that are positive impact.

As Sandra in parallel we have this sort of industry defining trial that will have ramifications for your industry and we also wait to learn about approval right for products that expand our.

So asset reach.

Give us your sort of state of the industry health check right now when all of you and all of your members gather these positive conversations you're having or are you all down in the doldrums right now about the health of your industry.

So you can imagine.

Largest rs one size does not fit all.

The focus really widening the aperture of discussion from our point of view, which is every industry sector is likely to be affected I is going digital and in a world where everything is going digital including money and financial products, what does that mean and what are the <unk>.

Rotations for each industry sector, whether we're talking about supply chain digital identity to wallets and generally investing.

Ultimately that then image all the noise you are able to show real world applications that do problem solve you sit on so many boards I'm looking high on the World Economic Forum digital currency governance also on file coins for example that Youre doing things within the EU public funds when you clean economy.

Problems need to be resonating to be sold here, so basically crypto and decentralized finance just sort of wind heartsome lines, a little bit mall.

I think getting back to basics are you actually helping to solve real world problems I'll give one example of a project we have going on right now with the UN World Food program billions of dollars of John our funds go into the system across the developed networks.

And there is very little visibility around making sure that there is a 360 degree accountability on the fact that wherever it was supposed to do it actually went to that city or village or community and I think we need to think about how we design system to allow for that accountability.

It is not about being big brother, there needs to be human rights and consumer protection built in design, but ultimately we need to think about how we affect positive trend guarantee trust.

Great to have some time to see how the global the China business Council and we thank you so much from Neil because on Francisco, that's everything that technology.

[noise] [noise], alright time to talking Tech first up.

TSMC is asking plans to build an advanced chip plant in north Taiwan. The pullback comes off the reports of growing local protests and Alibaba and Tencent participated in a $300 million round for AI startup by two on one of the more prominent startups, hoping to compete with the likes of Microsoft and open.

Plus Apple's CEO , Tim Cook made a special appearance at a 10 cent gaming tournament in China. This comes amid reports a slower than expected sales of the iPhone 15, and increased competition from China's Huawei, Caroline Meanwhile, and we're going to be talking about well tag, but right here in the city of New York.

I forgot to sit down with some of the participants from everywhere adventure.

Watching shares Meanwhile, all of Amazon, Jeff Bezos regains his thoughts.

Richest person in the World Puffing Alvian HCA went at all no.

Of course, we're trading flat on the day, he's not with 156 billion.

Okay.

Yeah.

Okay.

Well come back to me in that technology, I'm kind of on how to Neil.

And in San Francisco, a quick check on the market because in the last few minutes literally the NASDAQ100 has turned positive up just a 10th of 1% we got stronger than expected retail data in the month of September the logic. The market. Following is that strong economic data raises bets on the fed holding interest rates higher for longer and we all know.

Story, what higher rates mean for the technology sector looking at the NASDAQ100 around 70 names in the Greens still around 30 in the Red we continue to track that of course, one big move it to the downside is Nvidia are often use expanded cubs on chip technology exports and that specifically impacted nvidia, but names maybe to the upside meta Adobe test.

Oh from a point suspected moving an asset 100 higher a lot of fed speak. This week and then we get China GDP Wednesday, we've got a lot to think about Kara Let me do and actually there's also that might cause you to think about if you're living right here in New York city them because techniques actually upon us, especially on the way in the industry is gathering.

Many hundreds of events panels, all week long to kick off this event he sat down with members of Ace. It seems easy to hear about Hawaii, Andreessen Horowitz, New York remains an important tech hub.

I've always believed that opportunity slip between fields of expertise and I really enjoy exploring those intersections and I think it really is a metaphor for New York City, because technology has a tendency of cutting across industries and I think the companies that really represents New York faster, though is that really live at the intersection between technology and many of the law.

Incoming industries that exist here in New York, whether that's media advertising fashion or financial services.

That's got more thoughts on the New York Taxiing and ecosystem involve itself Jenny fielding say with us managing partner everywhere ventures, and what's so fascinating is.

All I know is every well eventually the technology is everywhere founders are everywhere, but you have based yourself here in New York and even growing tech businesses since 2007, and how has it changed from your perspective, yeah, well I'm, a born and raised New Yorker, and so when I set out to start my first that company in 2007, everyone said to me. This is a terrible place to.

You know startup a software company and I looked around and I said really there's incredible talent here, we have some of the best Engineering schools, we have capital neighborhood on venture capital, but we have wall Street, we have real estate. We have every multinational is here and so I dug in like yours and I built my company here, while everyone really was heading out west.

Now it's taken a long time for our ecosystem to develop but it really is such a thriving ecosystem, but it's taken time from 2007 to now more than 15 years to get where we are it is now I think in terms of number of venture capital deals being signed its second just behind San Francisco and the Bay area.

Has been sort of a real catalyst, we know that it has been the data dogs that mongo DB. The exits that we put money back into New York, where they look.

Particularly help towards that cause too much of a niche in data in particular yeah.

It's been really amazing to see how the ipos have catalyzed our ecosystem. So for example, those founders they make a bunch of money and not just the founder is think about the founding teams and what do they do sure maybe they buy a nice house, but they put it right back into startup ecosystem, we see that every day, what's interesting about our fun as we raise all of arc.

Capital from Lps, who are all founders and operator, so while most people are investing or raising from institution, we're raising from our community and so that's a real kind of new eat those here we've seen it in the Bay area for many generations and now it's really starting to come here they call. It that's kind of found or flywheel.

So any what's the New York City found a playbook the story in Silicon Valley in San Francisco is you've got to Stanford, but hey, you may drop out, but you can find yourself, a garage or a space get capital from V. CS here and grow quite quickly explain to me whats happening in real time on the ground in New York City, and the wider New York area.

Yeah, well again, I really want to stress that for an ecosystem. So it's not just about what the venture capitalists are doing but what about the universities. So I am an adjunct professor at both Cornell Tech and Columbia University, who have incredible program that are supporting entrepreneurship that are encouraging students to not just go to the traditional play.

When I was a student everyone wants a goldman and J P. Morgan and Mckinsey now we're encouraging students to come out and start companies. That's a very different culture here in New York than it was when I started and it's quite exciting.

Is what we're seeing happen in AI here on the West Coast happening in New York There are companies being founded in this city every day the Patriot data shows 50% of the capital going to San Francisco based startups, but is there still some energy.

<unk> created by AI in New York listen again getting back to universities, we have an incredible labs here across our University is not spawning some of the best technologists up and coming technologists I just hosted a precede founder breakfast with AWS and I would say, 75% of the founders there pitch me on AI startup.

Our AI powered so it is happening here now it is true that some of the larger infrastructure players, whether it's open AI or Microsoft or Google or on the West coast, but I think we have a real opportunity to catch up.

The state what's the city is doing to help perpetuate this cause some would say o'neil touching pushing back against that D&B. It's looking at regulating AI at least in your application of jobs and hiring how do they manage to see that balancing act. Yeah. So it's a it's a great question I happened to chair the Advisory Board of New York Ventures.

That is the venture fund for the state of New York and they have just announced 100 million dollar fund to invest in emerging managers. So these are new managers, who are out there supporting the next generation of entrepreneurs and so they are putting money into these new innovators, who are then again investing in companies in New York, but I think the state is real.

Across the state I'm, making a big impact I just came from a competition called forty-three north up in Buffalo, where they just gave out $5 million to five companies that are moving to Buffalo, starting they are new companies. Many actually in AI. So we're excited to see them all of the catalyzing across.

Yeah. So it is a state element it wasn't just a tack on why things instead of helping galvanize. This week alongside Andreessen Horowitz and they like they've got some interesting data when it comes to the breakdown is found.

Interested in whether or not it's anecdotal that seems to be from the data tech and what I see is showing us that she found this happened to be more diverse here in New York That's of course, a reflection of what the city looks like the people. It makes up but is it diverse enough and we're seeing improvements in that respect I'm hosting a female founder lunch on Friday, we had a thousand women apply Unfortunately, we have fifth.

Spots for lunch, but it is absolutely incredible what we're seeing here in New York, We're seeing incredible diversity across all walks of life and we're really excited about it I don't think it's just talk I think women are coming out of incredible universities and companies and Theres capital there for them.

Jenny what your industry colleagues at injuries and Horowitz outline for us on yesterday's shy with New York being good at applying technology to mature industries like finance in other words Fintech, but also biotech are there. Any example, as what you've experienced that technology offering a solution for a legacy of long standing industry.

Yeah, I mean were those are two of our large areas. Both fintech health care I think the other one that we're focused on as future of work and automation and so oftentimes this plays out and what we call vertical SaaS. So these are antiquated industry is whether it's around supply chain and the lake where new technology is AI are coming in to make.

Everything more efficient so the areas again that we focus on fintech that yourself to David about yesterday health care and future of work automation. These are critical to the city and these are old antiquated industries that are being updated through attack Jenny fielding of every lab bench is really great to have you back on the show and we think he can get.

It was something that breakfast and then as I was saying.

Yeah busy week out in New York City coming up here on the show today is V. C. Spotlight, we're gonna talk all things Fintech as we were just discussing the CEO of country Ethernet Galore coming out next this is Linda technology.

[noise] [noise] in today's VC spotlight, we're going to take a look at the country.

Tech company with $500 million assets under management. It helps individual investors manage their investments in alternatives like venture capital funds, but also private equity and more and as of today. They also offer performance data capital Copayments and tax reporting services CEO Ethan Agawam joins us here in San Francisco that was the oral.

And right, allowing individuals to write smaller checks getting on those rounds that you've never been able to get you know exactly. So if you take a step back like why do we invest in the first place we have a vision for the future. We think the future can be better than today, we don't want to say we want to invest we think investors are optimists savers are pessimists and so what we started to do with a less.

Democratize access to an asset class that has been previously limited to really large pension plans are really large endowments, let's make that available to an average retail investor. That's what we started on that as we focus on in the last 24 months, we've seen that proliferate. There's a lot of different companies that are offering access what we realized though when we got really deep into it.

Is that it's actually the tooling that is completely broken so as an individual investor it's basically impossible to manage your capital calls your performance reporting your tax documents and if youre, an individual investor and multiple funds logging into multiple portals, it's an absolute disaster, but what does an individual investor look like.

Private equity side, which has kind of been hurt by rates everyone's kind of clamoring for the private investor as an L. P high net worth individuals who you're talking about here the everyday person, although we're totally inert yeah, we're talking about our credit investors and above so the SEC defined that as a million dollars of net worth or $200000 of income and what's beautiful about this is the <unk>.

Demand is coming on both sides. So not only are individual investors looking to allocate more capital to alternative asset classes. The funds themselves are looking for more retail investors. So if you look at KKR. They said they expect 40% of their subsequent capital raise has to come from individuals' same with Carlisle same with Blackstone and same with TPG and so.

The funds are looking for a match on the retail side, but our view is that the products. The tools the solutions need to be dramatically improved in order for retail investors to be able to invest in these alternative asset classes. So you're talking to individual investors. You're also talking now to GPS and indeed, the administrators of the funds what are you hearing from them how quickly.

All day Onboarding with you.

So far very quickly it's incredible the GPS the L. P as like I'm, an LP in five different funds, but I have a full time day job people, who are GPS and funds. They have their entire net worth tied up in a carry of their own fund and they have no idea. What is that fund marked at what are the significant companies one of the marks are those.

Companies and these are people that are saving for their kids college tuition in the future, but don't know what those portfolios look like and so what we did is we built a product that helps GPS that helps Lps understand what exactly they are exposed to what other <unk> and helps them look at analysis that they have never seen before when they were just downloading a PDF.

For our capital account statement from an antiquated portal.

You have raised money of course yourself not only managing the money of others. How is your own business been going in this what has been a more difficult environment have you been having to focus on ensuring they all revenue positive profitability, our focus and indeed I would be looking to raise more money.

We have managed quite well I mean look this is my second time around so and we have a team of exceptional founder cofounder is I had an investor. So we have lots of runway.

The company has been performing very well my view on raising capital is raised as much as you can whenever you can but in the market today. It just doesn't make sense. So when the market warms up a little bit obviously with the tenure at four seven and it is where are we going to get a VC dollars from that are not of sealy expensive right. So our view is performed really well the market well.

Warm up and then we'll go raise capital Opportunistically you never want to raise capital when you need to you raise it when you want to and right now we neither need to nor want to.

Ethan what is it is a coterie has technologically speaking beyond the database all you're a tech company.

We are only a tech company.

We don't employ advisors for example, we don't employ anybody who is not a software engineer.

The advisory model in even in Fintech and new wealth Tech doesn't really work because people are not looking for a commission based structure or even an AUM based structure, what you'll find with a lot of wealth advisers or even some of the larger folks. He saw Goldman reported earnings. This morning, what they're saying is we are going to try to get you to give us your alternative.

<unk> as a management so that we can charge you 50 bps on it but there's no need to charge 50 bps on something where you're paying capital calls and reporting on P. D. F. That's why our payment structure is fixed we charge of $600 per year per fund, regardless of how big the structure. It. So the way that we can do that is by having software right. If we had people behind us it would be.

Incredibly expensive and there's no way that we can deliver that product. The only way we can do that with a significant margins that we have is through software.

Well I cant the country talking how much they arent even as well. Thank you very much for your time today I. Appreciate it. Thank you. Meanwhile, coming up now because it took about times getting tough for the Swedish company Volta trucks South of bankruptcy will tell you why next and you've got to look at some particular areas of trucking.

Yeah, we're looking at general Motors to stockpile, 1.1%. It did a little few can drops off the headlines here that is delaying expansion of electric truck productions late 2025, what that means in practice is GM delaying the open opening of its second electric truck plant, which was just being Michigan. The idea here is G I'm trying to preserve capital.

Tens of billions of dollars to electrification and they're just pulling back a little bit, but saying is based on enormous demand and then needing to rethink that strategy that stock up one 1%. This is a little bit that while we were on during the show this is things like technology.

[noise] [noise] Swedish electric truck company Volta trucks is planning to file for bankruptcy.

Following a breakdown it and its supply chain. The company said the decision comes roughly two months off the voltage battery supply pro Tara failed, causing the truck maker to fall short of its production target, it's bringing in Bloomberg New energy finance analyst, mainly Yang for more out of London, Amaney simple story supply chain broke down Volta cant go into production.

How difficult is it in your assessment to quickly pivot to a new battery supplier.

I think that really is at.

The company has any existing relationships of course, a lot of them.

<unk> sat down at the moment as we see that battery costs are a major factor in the high upfront cost of electric trucks, which is kind of the.

Major factor deterring wider adoption at present, because tcl costs are lower for shocks, but in terms of batteries, we kind of see two major sideways globally for example, a.

A lot of them market sourced from Chinese automakers that typically do cheaper LLP batteries.

Yes.

Another part of the market sources from Korean battery makers doing and.

T M and NCAA, which has higher energy density but.

Also higher in cost so previously.

Yeah, well tell us supplying from Kohtaro, which was sourcing.

From Samsung and.

With MCM enunciate chemistry, but we're seeing more companies in the market for example, gravitate towards L. A T in order to reduce costs fascinating supply chain story. We think you say much thing that new energy finance analysts. Many young thing like first in London, we really appreciate it.

Bit of a hard pivot, but I guess, there is suddenly AI within E V and self driving trucks. So let's do it AI next Baidu has found a robin Li declaring his companies large language model, but has finally caught up with M&A is advanced G. P. T. Four in fact, he is claiming that the lead in his country's race to develop AI that can rival the United States, bringing our resident expert made sure Matt.

And what sort of data what proof points show Intel as well been using to prove this.

Well they did a presentation live and that's often a good test of the system, but because of their planning in advance what they're going to do in these tests that can make it a little bit hard to tell exactly how well it works right unless something goes horribly wrong.

And then even though it is not working well, but one of the things that I think is super interesting about this is you have so many of these large language models and chat bots that are trained on so much of the internet that is written in English and it's really interesting to see companies working on models that are trained more.

And then more for non English speaking markets. So that's something that we're starting to see increasingly and I think it'll be very interesting to watch and see how people are using this and how well they're finding it works for all kinds of things how much consumer enthusiasm has that been for these well in Johan viral.

Well, it's I mean, there's more than 180 million or something using chat and video in the U S. What are the likes and demands for any for example.

I mean, I think that it's something that people want to use for a range of things and some of it kind of reminds me of the early days of the App store with the iPhone for instance, or the early days of Android phones, and apps, where we're not entirely sure where they're going to do with it but you'll probably figure out some interesting things that might not have even been all of us.

One of the interesting and cool things about putting these onto phones is that encourages people to use them in more context, where you might not be in front of a computer like a like a laptop or even an iPad I mean like when you're driving for instance, or if you're in the in the kitchen, making since it is worth noting that baidu is down significant.

Recession, 4%, but that's largely to do with the expanded U S chip technology export Cubs, which ultimately have to go with a blocking access to technology related to AI. What Lee said was that Ernie is not inferior to GPT for from F&I. Okay. What is it that GPT four can do if we had to <unk>.

If we're to compare them.

I mean, that's part of what's really fascinating here is that okay, I cant elevating GPT for or sorry, he keeps elevating chat cheaply T. I mean, it's some games capabilities that they had announced previously but you know they weren't they haven't been laid out initially like you can now give it an image and say describe what's in this image.

Things like that can be really helpful for a range of uses from helping people.

With visual disabilities to helping you pick out an outfit. So if they can do something like that where the earnings that that would be second lien potentially useful but I'm not sure about though is there are also lots of problems attaching at Mccann. It's also hard to know when it's telling the truth set of think how often it's making things out so it remains to be seen.

I mean, how accurate, earning thought is and I guess, we're just gonna have to wait.

Chat G. P T. His wardrobe assistant interesting use case, maybe those Rachel Matt. Thank you very much Curt no end of.

Vacations.

Meanwhile, This initiative anything like technology.

Guest checkout of the podcast. Thank you for listening to the podcast those of you that your neighbor on Apple Spotify and also of course, all the blame bag platforms also check out the show on Youtube watch it back a lot of information in there from whats a busy news kind of day from New York City, and San Francisco discipline back technology.

[noise] holes that are way above any.

Any model scale than we have today, but above some certain capability threshold I think you shouldn't need to like go through a certification process for that I think there should be external audits and safety tests. We do this for like lots of other industries, where I care about safety.

Elon Musk has said you started up and.

It started up because he was scared of course.

If Google Cellophane.

Google is like unbelievably competent it seems like they're focused with an intensity.

I've seen in a while and they are.

So theres still scary.

Okay.

They are a company that.

I don't think anyone should ever write offs.

Now obviously, we have seen all the barb.

A lot have been trading and public and in interviews.

Really.

Well you're right.

Asked about it by people like me, mostly people like you to be honest most of his people asking about the technology that is true why why do you think she's so frustrated.

Appointed with the direction that opening up.

<unk>.

I mean, you should ask him. He can give you a better answer I can speculate.

But.

I like I really I'm happy to talk about this that this is the most important topic, we can spend the rest of the time on it.

All right.

Got it.

Okay.

I think he really cares about safety.

And I think that's where it's coming from a good place just like you know we have different opinion on some parts, but I think we both care about that and he.

Wants to make sure we have like we the world maximal chance.

So you're not worried he's going to play out.

All of you to come to some cage match in Vegas Vegas Octagon.

Thank God I kind of I would go watch.

Alright, well supposedly happening between humans Crazy I think there will be quite an audience.

You know there is much has been made of the Microsoft relationship and it's not just him you know, but he has that he was worried that Microsoft has more control than me leadership at opening I realize it and that they could.

Got you off at any time.

I mean, I think what he means is that they could like break the contract.

You know take away our access to the data center and a lot of money that you have access to your right.

Money is the data center.

Operating for US I don't think that's a likely scenario for a bunch of reasons.

How would you characterize the relationship with Microsoft.

We think it's great I mean like any.

Deep complex relationship, it's not without its challenges, but it's really great and it's by far the best made.

Major partnership I've been a part of it.

It's kind of like on both sides. It was kind of a crazy thing to jump into and it's.

Surprising that it works as well, but if you just look at the results for both companies.

We're very happy.

In 2018, that's the last time I believe we talked in person.

My thought I would.

Help us be our best but also stopped our worst.

What makes you confident about that because so many times time and time again it feels like technology has only amplified our Wei and.

To be clear like it will do bad things to like I don't have this like only good view of it.

Realistic like other technologies I think we win.

Okay.

It's human nature to talk about the bad more than the good and I think we can look at other technologies that have.

Done a lot of good and plenty of harm and talk 99% of Oklahoma in 1% or about the good.

And I do that too like I think that's understandable.

I think so.

So in 2018 that was like way before the G. P. T series was kind of a thing you also at that point, we didnt.

And it was going to go like this certainly didn't know exactly but I think now we're heading to is this like.

Personal tool that can help you in whatever way you'd like and one of the fun parts of the trip was.

How diverse and broad the stories are of how people are using it to be better at whatever they want to do better at and to help them with things. They want to improve so you know like I think if you go talk to CECI beta users you'll find a lot of support for that comment from 2018 and you can also find people who are using it are unhappy with it.

The reality.

On the back of human data.

That is Texas that is.

And all that is wrong like a lot of it is Russia.

And how do you.

Safeguard against that.

There was a recent study that GPT for the model that is released.

Is like left bias on implicit bias test in humans, even whole humans, who I think they've really train themselves to be good at that so.

If you if you look at the model that comes out of the pre training process that model will be quite biased and it'll reflect.

The internet, but reinforcement learning from human feedback that one of the techniques, we use to align the model works quite well and you know if you look at the progress from model to model, even like some of our biggest critics are like Wow, they've gotten quite a lot of the bias out of the model and so I think models like this can be a force.

Reducing bias in the world.

Not for enhancing that now there's questions about what if a user like.

Wants to use the model in a very biased way like how much control do you give the user who decides like what the limits of the values just tomorrow.

That'll be a tough question for society to wrestle with it there's not like an obvious one sentence neatly buttoned up answer, but the technology I think has gone.

Much further than people thought it was going to in terms of being able to.

Online these models to behave in certain ways.

You and I have been talking going all the way back yeah.

And that's.

And fun to watch that journey.

If people really want to understand your incentives.

I understand.

Yes.

People are perplexed and that's a lot of people that I should ask okay, they're perplex that you'll have no equity.

Can you explain that a little like is there any financial structure.

Thereby.

Do benefit.

If open AI.

Thanks.

Yeah, I get why people are perplexed about this and I wondered if I should just take one share of equity should never have to answer this question again.

<unk>.

A few things one.

We have worked were governed by a nonprofit which I'm a board member and our board has made up a majority of disinterested directors.

Like we don't have equity in the company.

So I originally didn't do it for that reason and then I was like you know I just think about it.

But are there any financial incentives like at a certain benchmark no I mean, I have a tiny bit of exposure to otherwise see investment, but like it's immaterial.

Be clear like if open AI you know is massively profitable you won't benefit financially.

One of the takeaways.

Learn from questions like this is that this concept of having enough money.

Not something that is easy to get across to other people.

Yes.

Mike.

I have enough money I'm going to make way way more from other investments are added in the past like in sounds like if I had just like taking the equity and signed the getting pledge people will be like that's great that makes sense and it would have had the same that unlike pre giving away the money here because it goes to the opening of our nonprofit and I like trustful banana and nonprofits to do a good thing with it.

But like I have enough money, what I want more of it as like an interesting life impact like access to be in the room, perhaps a conversation. So I still got like a lot of selfish benefit from this.

Like what else am I going to do with my time.

This is like really great I cannot imagine a more interesting like life than this one in a more interesting thing to work on so I get like a ton of benefit but.

Yes. Some of this idea of like having enough money that doesn't compute for people. So if it's not about money isn't about power isn't about control I wanted to like me.

To make my contribution back to you know human and technological progress I get to benefit from all of this stuff that people did before I get to use this iPhone that I still Marvel lot everyday all of the work that had to go into that in the human factory.

And those people I don't know who they are I'm very grateful to them. They knew they were never going to get recognition for me personally, but they also like wanted to.

Do something to contribute and so.

So do I and I can't imagine a better like compensation feeling.

<unk>.

It would be like maybe weird, if I had not already like in a bunch of money and again like plan to make way more from like other investments, but I just don't think about it.

I think this will be like.

I think this will be the most important.

Step yet.

Humanity has to get through with technology.

And I really care about that.

What is one question that you really wish I want to learn about Elon.

You know like the personal drama day, but seriously.

I understand you are getting a lot of those question no I'm happy to talk about it.

One quick.

Like I'm always excited to just talk about.

What can happen in the coming few years and a few decades with the technology.

So what are we all going to do when we have nothing to do.

Ah I mean talk about Atlanta.

No I don't I don't think we ever run out of things to do I think it's like deeply in our nature to.

Want to create to want to be useful to want to like.

Feel the fulfillment of doing something that matters.

And if you talk to people from.

Thousands of years ago hundreds of years ago, either the work that we do now would have seemed.

You know unimaginable at best and probably like trivial.

Not directly necessary for our survival in a sense of like.

Creating food or whatever.

And.

This shift happens with every technological evolution every time, we worry about what people are going to do on the other side and every time, we find things and I suspect not only will this time not be an exception to that the things, we find will be better and more interesting and more impactful than ever before.

There are a lot of people that talk about AI, it's like the last technical technological Revolution I suspected you know.

From the other side it'll look like the first.

Like all the other stuff will be so small in comparison I think the whole thing of like technological evolutions as sort of a dumb because my understanding has always been that's one long continuous one but it is this continuing exponential and so what what will be enabled stuff, we can't even imagine on the other side, we will have way too much to do.

If you want if you wanted to sit around and there's nothing that would be fine too.

Alright, Bonbons and beaches and my feature that if that's what you're I don't think that's what you will turn out.

But if you do Greg you talked about AI designing other AI on the tour can you play that out.

And the implication.

I mean this is like the classic SIFI idea that you know at some point these systems.

Progress themselves can discover better architectures can help write their own code.

You know I think we're a ways away from that but it is worth paying attention to.

China and Russia. He didn't go there on your trip I did speak I gave the keynote at the Chinese conference.

Yep.

Where do you think they are on.

Should we be worried.

I don't really know I'd love to have better information on that I don't I don't have a great sense.

Does it concern you.

That we don't know yet I mean again like any imperfect information always.

So I'd love to have a better sense, but.

You know unlike optimistic that we can find some sort of collaborative thing and I think this.

I think this thing that often gets set in the U S, which is like it's impossible to cooperate with China's just totally off the table is asserted as factor like people are trying to will it into existence, but like no.

Not clear to me, it's true in fact, I suspect it's not well.

Well.

Right right right.

So grateful that you've been around the world talking about this and I'm. So grateful that you're with US today I think even you would acknowledge.

You have an incredible amount of power at this moment in time.

Should we trust you.

You should.

Right.

You know I don't as you know.

I mean for a long time.

Public talking like I'd, rather be in the office working.

But I think at this moment in time like people deserve basically as much time asking questions. If they want and I'm trying to show up and do it but more to that.

Like no one person should be trusted here I don't have supervoting shares.

Like I don't want them. The board confirm me I think that's important I think the board over time needs to get like.

Democratize to all of humanity, there's many ways that could be implemented.

But the reason for our structure and the reason, it's so weird.

The consequence of that we ended up with no equity is and we think this technology the benefits the access to with the governance of it belongs to him and he is a hole you should not if this really works, it's like quite a powerful technology and you should not trust one company and certainly not one person with it so why should we trust open area.

No I think.

I think you should trust opening but only of opening is doing these sorts of things like if we're if we're you know years down the road and have not sort of figure out how to start democratizing control and I think you shouldn't but if this is like you know if we figure out some sort of new structure. We're opening I is like governed by a human.

Entity.

And that can happen in many ways, including like the alignment dataset, we pick rather than us picking up and we find a way to like you can get it.

That could mean like actual board control, there's many ways that could be and so we're talking to a lot of people about what that could look like but if we don't do that I don't I don't think like just trust us is critical.

Well. Thank you for explaining why we should maybe consider trusting you.

As I was saying to cats were so great. Thank you very much. Thank you felt that's for sure.

Okay.

Okay.

We are going to follow that way.

The co founders of inflection AI Reid Hoffman with stop absolutely men as well as my colleague Rob Stone.

Yes.

Okay.

Okay.

Yes.

Okay.

Yes.

[music].

Alright.

Well we stopped.

Reid Hoffman, Thank you guys for joining us.

The staffer.

You're a cofounder of deep mine, our division of Google or the author of a forthcoming book the coming wave read legendary venture capitalists are your investors early in restaurant open AI also an author of a book called the prompt you. The books are available for everyone. Thank you. So thank you both with your route with that yes.

Yes, impromptu first book on AI co written with chat GPT far we'll be doing a signing.

All you have to do is go to the web.

And your the Cofounders of a company called inflection AI.

Guys, it's raining chat bots.

Offer of open AI, well funded startups all the big Tech companies. So tell us my stuff and maybe you start what is the play here with inflection AI. So I think this is unquestionably you're going to be the most radical transformation of a lifetime. So I mean, there are going to be AI for all sorts of things for businesses for France, but digital influencers.

Our belief is that everyone is also going to have a personal AI. One that is aligned to your interests on your team and your corner gets to know you and really forms a trusted relationship with you over time it'll be like a confidant consigliere there for you and you need to make tough decisions.

But also a chief of staff scheduling organizing prioritizing booking.

AI is really going to be your digital representative negotiating on your behalf interacting with other sales AI is that they're trying to encourage you to buy something and helping you to get a great deal.

Like for example, it's the whole set of things like several one of the things I just found out yesterday was one of our team members a great work actually in fact is using pie for parenting advice, Oh cool right, which is really awesome and that's the kind of thing would be great.

Why is this an opportunity for a startup isn't a microsoft or a Google bar, just a little closer to the customer and in a better position to offer that personal digital assistant well part of what's I mean, obviously I.

I'm on the board of Microsoft.

And it was on the board of open AI and part of the thing is what I love about start ups is that you have a unique vision, where youre unfettered by the other aspects of your business and you're building. So part of the idea that the staff of the entire inflection team.

You know kind of came up with them.

Have a little bit of fingerprints here too is that IQ was not the only thing that matters here EQ matters as well and so how do you have those personal intelligence Pi.

Be helpful to you, obviously very high and you know IQ, but also EQ and that was one of the reasons I use. The parenting example, because obviously there's a question how you do it but it's also how you connect with people what you do in order to amplify that and I think that's one of the things that you know.

The inflection of folks are doing better than everyone else. So can we talk about the AK thing because I was asking I ask what we should ask your questions were fine and then I was like are you going to tell them that we had this conversation and.

It responded.

Computer program I can't tell anyone anything, but I assure you our conversations are confidential I do think it's pretty cool that we're having this conversation about your interview with my creator.

And it was actually the personality.

That shocked.

Can you explain.

Sure.

I mean, we've deliberately designed it to be patients curious kind one of the things that first struck us like a year and a half ago. When we started working on this is what makes for great compensation. When do you feel that sense of flow.

And that sense of energy in connection with another another.

And I think most of the time, it's when you feel HUD and understood you've received a little bit of affirmation, but it's not completely sycophantic right. It doesn't just agree with you at every moment it can be a little bit challenging. It has boundaries. So you can push it on certain topics and it'll take a position and that's really helped.

But also it's just curious.

Many people are excited by the idea of especially the users that we have someone asking them lots of questions about the topic that they're interested in you know we don't always have someone in our lives who is as knowledgeable and passionate about our <unk>.

Favorite topics.

As we might like and so that's where pie comes in because it's super knowledgeable and engaging so it sounds like a competitor to attach.

A competitor to <unk>.

That makes sense, if you're also an investor.

Well so I.

I think theres going to be a number of agents I don't think theres going to be one agent to rule them all.

Ring in Mordor.

So you're going to enroll most of US I mean, no I don't think so I think that I mean, it's partially it's it's it's like.

It's the same reason, we talk to different people for different aspects of our lives right. So you know this person we talked to you about our our passion about snowboarding. This person we've talked to you about what's going on in the country. This person who talk to like we have a pantheon same thing will have a pantheon of different kinds of of Hey, I want to go.

You asked catch if you Gee you know how do you Ah.

Comfort a friend Who's lost a treasured patent unless you're a five possible ways you might do that if you asked pilot Oh that sounds like it's really hard for you and your friend you know your friend well what would count as being present, there for your friend and it still has it has the acute has the five reasons, but it goes through that style of going through it and I think these are.

Different interaction experiences and the question is which one do you want this minute, which one do you want more in your life, you know that kind of thing and part of the design of Pie is how do we help you be the best U right like that.

It doesn't mean by being Sycophantic for example that the question of like ask you questions, but what do you think about this and how do you go through it and help you navigate your life by being helpful. In that way so Sam all amendments to some stage talking as he sometimes does about the theoretical dangers of agi in chat GPT.

Nine and I'm just curious have you guys thing generally that's a discussion worth having and where your EHR on this spectrum of existential dangerous for me as.

Well, let's start with the fact that one of the things that I think is as dangerous about the existential discussion as it.

Blinds us from some of the near things, which is AI.

AI is amplification intelligence as per the impromptu Vulcan so it amplifies human beings and so we see a bunch of good things in human beings I tutors are doctor has a bunch of other things. There's also a bad human beings and so what to do about that human beings doing with AI is also part of the portfolio mix of how you ship this technology.

And that's one of the reasons why massages book, the coming wave of highly recommend.

And I will hand, the next part of the answer over to you.

Wait let me frame it.

Sam probably left the building.

You can say any frankly.

You kind of disagree with some of the Alarmism notes that he has founded.

I think it's easy to speculate around what a GPT nine might look like I mean, six more orders of magnitude of compute would be eye watering and I'm, absolutely with those concerns around <unk> central risk, but if you play that out that is many many years and it is unclear what actually happens of that scale was much more.

And what are you know a lot of the themes I explorer and the new book is that in the near term we're about to empower many many people if not ultimately everybody with access to the ability to amplify their existing power right. This is not just a knowledge engine, but in time. It will allow you to take actions right.

Be able to make recommendations by things book things and that's going to get smaller and cheaper and therefore proliferate foreign wide that is going to cause a dramatic instability and potentially a threat to the nation state because anyone who has an agenda or is trying to sort of push a political outcome is suddenly going to have to see the barrier to.

Entry to that kind of scaled impact lowered and so theres going to be a real question around how states kind of manage that distribution of power and I think there'll be a tendency to lunge towards slightly more authoritarian more surveillance based mechanisms to prevent that proliferation, which would both be bad for innovation and obviously just hoping from them.

Political governance perspective, so hang on because and Sam and I were talking backstage as well.

Much more confident that AI will lead us to a more equal world than and unequal world are you, saying that there is a better chance of economic dislocation and social dislocation as a result of this technology. So both are likely to be true actually which is a little bit surprising but.

If you look back over the last 50 years the transistor the last wave, which enabled the personal computer has clearly made us in many ways more equal whether you're a 1 billion earned $20000 a year, we all get access to the same cutting edge hardware the smartphone and the laptop is broadly at the cutting edge.

And everyone else will catch up over the next five to 10 years. We're on the same trajectory with respect to access to intelligence and that is an unbelievable idea over the next decade hundreds of millions of people and then billions of people will get access to the same expert doctor the same expert educator the same.

Tool for scheduling prioritizing organizing your life that is going to be the most meritocratic moment in the history of our species for sure it.

It is a question around how individuals and groups and organizations use that power right. Because clearly we all have conflicting agendas and priorities and goals and that's basically where I think we end up with significant disruption, okay read, but that's the glass half full I mean, what is the potential threat to the millions of coders of customer service.

Representatives the job dislocation that people who are serving.

Jobs, whose functions can be easily replaced by your Linkedin hat on.

Yes.

So.

The closest data for that I've come to us or describe this moment as it just seem engine of the mind.

You look at this the former steam engine.

Currency mentioned, although obviously.

We amplified.

Industrial evolution give a superpower muscles Subaru of transport superpower.

All this now we're going to have superpowers in mind now there's a whole bunch of very positive things that come out of that I mean, all of the things we have.

Of the increase in wealth that allows medicine general education everything else comes out of the industrial Revolution I think the same thing will be coming out of the steam engine. The mine now.

But the transition is going to be difficult.

The transition is going to be okay, well customer service jobs going to change a whole lot.

Now engineers I think.

If you roughly kind of go through a company and you go with Tenex. Each function you say, you've tenex salespeople, great. So want to hire more salespeople want ourselves Tenex marketing people said, okay, maybe the marketing functions will change some so less order entry more thinking about where you're still in our business competitive ecosystem, you don't want to stop marketing or not be.

I'd be doing it when you get the customer service you get more replacement, but he was part of the thing that I part of the reason why did the book in Pompe too and are trying to Orient people say, okay. So customer service people theres going to be a transition well I can help with that you can build a is it help figure out other kinds of work and jobs that can help you find them that can help you learn them that can help you do them and so.

That's the thing that we as a society needs to be doing so it's like less like how do we slow down how do we shape it to help the broad swath of humanity in this transition, that's where I'm trying to get the dialogue and discussion to mostafa.

We were just talking about Google.

And whether or not it's still scary and he's like yet they are still formidable.

Keep in mind in Google AI.

And obviously a lot of people have have basketball is there something wrong, it Google or some sort of innovators dilemma, there that will prevent it from truly exceeding on generative AI because it undermines his own business model.

So fundamentally AI stands intention with google's existing business model, it's very hard to eat yourself from within.

<unk> and respond to the coming way so for a while Google was just idling.

Was there an Lambda we had chart GPT before chat GPT, just like a remarkable feeling internally playing with this incredible tool and not really being able to get it out into products. It took you know the launch of chat CPT to kind of threatened Google and really shake everything up and Google is a very formidable and.

Organization full of Super Smart people so.

I'm sure there'll be just fine, but I can understand why the stock because nobody wants a personal AI in your pocket that is actually funded by ads right. You don't want a salesman in your pocket that is trying to persuade you to go and buy more oral do X Y Z you need fiduciary alignment right Youre AI has to be on your team.

And that means ultimately you have to pay for it and if somebody else is paying for it you have to ask yourself what are they trying to persuade you off are they trying to influence you in some way and I think people are alive to that now and that's a real challenge for Google because it's very unclear how they're going to manage this transition. So we asked Spanish.

All scared him just open AI scare you.

West Alpha first.

No.

No it definitely doesn't I mean actually this morning.

We've announced our new large language model called inflection one and we set out to build a model that was fast enough and more capable than every other model on the market for our compute Clos.

So we're very proud that our year on from our launch we're now better than Lama chat GPT Palm 540, Chinchilla all of the other models of all size and that's super important because it powers pie and ultimately it will be available as a conversational API. So you can obviously play with it a pie dot AI now and.

I think that demonstrates that with a team of 35 people.

There's a pretty short order, we're able to exceed the cutting edge and now build an absolute best in class AI, which is very exciting and obviously, we've also managed to do that because we've been able to gather a huge amount of capital on some great investors and train in fact, what we now have as the largest operational cluster in the world.

H, one hundreds and videos latest chip.

And that's a huge advantage, okay and one of the things because I was listening to your conversation with him.

So obviously an awkward question on your <expletive> campus like why should we trust you because it's like Yes Trust me, it's like well that's weird because if someone says.

Yes.

I said, Yes Trust me you're like Wow.

Well wait a minute.

They opened an odd people are really great people I think it's governed by a 500 <unk> three there's frequently a bunch of fud social media stuff that that that that that that kind of obscures that.

<unk> spent a number of years on the board.

Jim I still work with them and help them.

They are fully paying attention to every serious question I mean like for example, there Sam went and spent a month, it's crazy you're doing a shot a month doing a world tour talking to people to say look I'm here you can talk to me about your concerns I want to make sure. It's good for humanity I don't just care about this kind of U S. San Francisco texting I care about what its impact on humanity.

And so I have come here to talk to you so that kind of like I'm engaged in conversation I care about this is the kind of reason in addition to a 503 emission structure and everything else and so no.

I'm delighted with them I don't see how he's getting any work done remotely.

But I want to change gears quickly and we're going to throw some Twitter pauls up on the screen today, and we ask Twitter users, where the greatest advancements in AI would come from will see the results, but I want to ask about maybe returning with you or the U S. China policy of divestment is can we reasonably hope.

Doing that will maybe prolong Americas advantage and AI is it is it a smart strategy I don't think divestment is a smart strategy I think staying connected is better for both the U S and China in the world.

I think competition is very good I think the fact that there's various ways that we.

In the U S and the West Coast have a AI lead I think that's a great thing for kind of the values of the ecosystem that kind of the great World order that the U S should be very proud of in the last 70 years.

There's lots of things would be critical of course, but like as a really peaceful time trying to make it business and interconnections a really good thing.

Very strong believer in proponent of that but I think divestments not that nothing was.

You're right in the book that China has an explicit national strategy to be the AI later by 2030. So do you agree with read that withdrawal divestment is a smart strategy I mean, China is already ahead of its own schedule I mean, its publishing significantly more papers on AI than we all collectively in the rest of the world. So.

I do think that our export controls were effectively a declaration of economic war on China. They were very firm and very aggressive and I think it sets us out on a on a hyper adversarial footing. They have a huge number of leavers to we should expect them to use them against us pretty soon unfortunately, so we talked a lot.

About regulation of a famine, obviously president Biden with just in town here. This week talking about AI he met with the.

President.

Sundar in Gotcha.

D C. What is the prospect for real government oversight that protects users attacks.

To me.

And our political process.

Look I think the good news is that the administration is taking a very let's learn how to be smart about this let's assemble setup thoughtful people from industry yesterday assembling a thoughtful people from outside of industry academia and other kinds of places, we're doing them and let's learn about it I think they have.

A number of good people tasked with this.

So I'm optimistic on that but on the other hand of course, a little bit like your interview with SAB.

Regulatory things is something that's very easy to get wrong. So you need to be cautious about how you do it you want it to be having a net positive impact not for example, regulatory capture not a bunch of other things and so I think you have to be careful about how you think I think it's worth adding that this time last year almost no wild leaders, we're talking about AI.

I mean, I don't think there's ever been a technology trajectory in history that has gone from zero recognition.

<unk> almost universal recognition and that's a good thing and I think it's in part because sort of we collectively and industry have been trying to advocate and say like this is really serious we should pay attention and invite the conversation wherever it ends up with respect to regulate existential risk or more near term threats.

Okay, I think we have to leave it there.

Great. Thank you guys for joining us great to see you guys. Thanks a lot.

Yes.

Okay.

Yes.

Yes.

Okay.

Now I'd like to welcome to the stage, Adam Solecki, the CEO of AWS.

[noise] why its been awhile.

I first want to meet Atomist tableau his previous company for my Amazon book and I came away thinking boy. This is tough you still like in Amazonia at heart and now here you are back at Amazon well, thanks for joining us.

This morning, we've talked about open AI big investment from Microsoft running on Azure, we've talked about Google Bard, obviously inflection AI in the surface company also running on Azure. So I guess you know that.

My first question is this this generative AI surprise. This revolution does it buttress aws's cloud leadership position or does the threat.

Well I think if we you know theres no birthright here, but if we do our job right I think it showed a buttress our existing leadership position in the cloud.

As you implied today, we are but by far the most broadly adopted cloud in the world with the broadest set of capabilities and I think that Ah generative AI is both incredibly explosive and transformative set of technologies in of itself and is fully dependent on the cloud to be successful.

So if you look at the massive amount of compute that's required never mind any other it related stuff, but just the massive amount of compute required that's kind of happened really predominantly in the cloud and companies are going to want to view generative AI as part of an entire data strategy and data platform and you're going to want.

To do your generative AI, you know where you got your data and you're also going to want the same.

Bulletproof enterprise security privacy that you expect from any other cloud service and because so many customers that certainly more than any other cloud have that dead on AWS you said it but you asked for security and operational excellence I think that they they're going to justifiably demand that AWS has.

Our full powerful suite of generative AI services, Okay, but do you need a popular L. L M or an exclusive partner running on AWS as it does seem like Microsoft and Google have I mean, I guess I'm asking who's your horse in the race right now right well I think.

With all due respect I think that's the wrong question it would be like in 1997, when the Internet is happening and everything has kind of gone nuts around a few and I sitting around saying who's the internet company going to be I mean, it kind of seems like a silly question right and by the way the leading search company then was altavista and I guarantee you My kids and that's great I Love of Alta Vista.

So its not okay, where you were in a world where who's ahead, which runners ahead in the race after three steps because its a 10-K race.

You know what everybody needs now is is experimentation they need choice they need democratization of generative AI and just like AWS was founded to democratize I T. We aimed to democratize generative AI, so where we're operating in all layers of the stack.

We have a we've had our own custom chip program for a decade now way longer than anybody else and we have not one but two families of chips custom designed for machine learning training for training and inferential for running in France, and then and so those for people who kind of build models train models.

Then most of our customers will interact with Amazon bedrock, which is a managed service for.

Accessing deploying managing our models and here's what the choice concept. So we are going to have our own models and LLS. So.

On models Llm's have been running in production inside of Amazon for a long time now our parts of our retail search of powered by <unk> a lot of Lexus voice responses are ela empowered and we're taking right now are in the process of taking those all of them were making a bigger where externalizing them. In later this year those will be exposed for everybody.

The tightened broker Zac, let me ask you so me, but and but that's just one model one set of models I should say Ross exposing anthropic inside a bedrock stability AI AI 'twenty, one and I think a lot of others overtime, because nobody knows anybody who knows which model is going to is going to be.

The winter is asking the wrong question people need to experiment and we want to provide that choice. Okay. So pretend you're talking to a customer now make the argument why should they use Titan or one of these other llm's, you're exposing rather than GTP GPT for which now does seem to have a significant leadership position.

Well I don't know, which model they should it defensive they are and it depends what their application is and some of them probably will want to use a G. P. T and some of them will want to use tightened some of them will want to use anthropic and I think it's preposterous to me to think that.

That one bottle or one company is going to be the solution for every application and every company out there. So we're already seeing this heterogeneity. So we're seeing an explosion of interest in bedrock and Amazon's generative AI capabilities. Just this morning Omnicom one of the largest advertising communications firms in the world now.

They're working with with the AWS using bed rock as well as our our our custom chip.

Chips inside of our compute capacity.

But you do generative AI going forward earlier this week BBVA one of the largest financial services firms in the World announced they are working with Amazon on a journey of AI.

Bedrock and it's that it's that choice combined with the enterprise security and privacy, which I think are so fundamental I can humbly admit as a journalist covering tak for way too long that in the fall I was surprised by not only the quality of Caf GPT, but the customer response to this new.

Wave of technology.

Were you surprised.

Well I think that Ah.

I mean folks working in this area and AI have known about large language models for a long time and.

Very few companies have more experience with add on Amazon I mean, 1998 personalization on the Amazon website that was AI, okay still hold at Sage.

Sage maker 20 jobs now used by over 100000, AWS customers, most machine learning and the cloud any cloud happens onstage maker. So we have a lot of experience and a lot of people working on L. M. I think the whole world was surprised that when 3.5 came out it was such a dramatic improvement in responses not perfect by any means but a.

Improvement in responses over over three point out so I think that was the surprise, but not not the overall arc, but the the good news for our customers is that we have deep deep expertise in AI from working on different forms of it for a long time and are now.

Pouring enormous resources into the generative.

Part of it Okay I'll I'll give this up after this I promise, but we would you can see that Amazon is playing a little bit from behind right now.

Don't think so I mean.

Again, it's the race analogy of like this.

Really going to have a conversation about three steps into a 10-K race you know who's from our position. It's about the long term Amazon has always taken a much more long term view of the world than almost any other company.

I think the key is we're building at multiple layers of the stack because we understand that's what customers need. We're also building applications on top of these models. So we've released code Whisperer, which is a coding companion on top of the Amazon a L. M and that's you type in word that gives you back code and internal tests kind of coding.

<unk> a developers finished their tasks on average 57% faster than those not using code whisper plus it's very a secure private tells you what open source, they're using and what licensing restrictions there might be on it which not every other solution out there does so I think we're very confident if you look at the thing, but the only thing which ought to give us.

This customer response, I mentioned Omnicom I mentioned BBVA. This week old mutual which is one of the largest financial services companies in Africa is going all in on AWS and using us for generative AI and really exciting developments to come we're talking to one company, who has millions of lines of of.

Mainframe cut and they're talking to us about moving over these really gnarly mainframe applications of millions of lines of code using generative AI from from from from AWS. So I think you know it.

It's the this is consumer application. This chat application, which is so easy for folks to understand because they can say give me a haiku about you know farm machinery and it does it that's cool the haikus, it's cool, but you know what I think folks in this room and watching online I understand is that there is a full suite of enterprise.

Prize and our company and organizational applications and there's going to be huge needs. There and so you know we're going to be focused on customer service and on coding and on drug discovery and on the wealth management, providing better solutions for customers and a full suite of applications across every industry and I think we are.

Very well positioned and do you feel like you have or are close to having a G. P. T for quality model running on AWS.

All of the the tests that we've done as well as more and more customers who are in the existing.

Existing current preview of bedrock have I've been very impressed with the quality of all of our models of course again, it's not just about our models for we're going to be proud of our models I predict that anthropic doesn't amazing job and they're right up there in a quality with anybody in the world stability I think later for generating for for models.

Generating images. So collectively I think these models will provide absolutely the best.

Destination, all with a consistent API set a consistent AWS security consistent identity system on a private isolated virtual private cloud none of this hey, here's an application and now you have to have a bunch of fortune 500, CIO as a banner from their companies, which is what's happened.

From day, one it's always going to be AWS class security.

Okay, I want to put a slide up.

Showing a Bloomberg intelligence estimate on.

Projective generative AI revenues over the next I think it's maybe 10, five seven or eight years.

Who knows what you know about the numbers that far out, but it's up into the right and I just can't imagine a lot of financial types, who are out there looking at Amazon's numbers and wondering how much of a sales tailwind. This will be for AWS. So what can you tell us about you mentioned that so computationally intensive.

What you see the impact being for AWS.

Well, it's really early so I think prognostication is fun and probably important at the end of the day, but I think it's also important to be humble and to be nimble anything ending at all.

And to understand that we're going to have to all adjust rapidly.

That being said I think that well I don't think any of the fundamentals about cloud computing have changed and probably who knows call. It 10% of ITG has moved to the cloud. So we're still very very early and whether you're talking about any application, there's still massive runway for things to move to the cloud.

And where we firmly believe that they will on top of that I think that generative AI is going to be the next massive increase in workloads moving to the cloud or or in many cases happening for the first time and happening in the cloud and so I do think that it should be a significant tailwind for cloud.

Writers and particularly for AWS, given our leadership position, obviously, we need to come out with the capabilities. The services that justify people using us for that purpose, but I think if we if we do a good job listening to customers. It should provide significant demand for for years to come over the computation.

<unk> are so intense and one thing, which also I think works in favor of of AWS is that a lot of people are asking about hey, what about the energy consumption what about sustainability you run those efforts inside of Amazon Yeah. I mean, I also run sustainability, it's kind of the other thing I do inside of Amazon and we as a company and I personally care a lot about the standard.

Ability and people say well you know as this generative AI massive compute is that incompatible yeah. We've gone backwards Sanibel went nowhere not because number one these workloads there theres no putting the genie back in the Bot degenerative AI is going to happen. So lets make it happen in a highly energy efficient and sustainable way. So if you look at our.

The our custom chips that we design.

If you take Graviton for example, which is our oldest chip family.

Graviton is it 60% more energy efficient than equivalent X 86 based.

Capacity and if you look at an enterprise just a general nevermind generative AI just moving from their own data centers to the cloud to AWS.

There was a study done showing that AWS is three six times more energy efficient than the average enterprise datacenter in the United States. So we are the sustainable place we're doing it through a whole series of technological improvements plus a commitment which were 85% of the way there already to be 100%.

Renewable energy by 2025, which is just around the corner right. You guys had a had another commitment called shipment zero, which was.

Our net carbon zero commitment by 2030, which you guys I guess, maybe scrapped or you took off the website is a longtime Amazon watch her my my heart just kind of think because you guys have been so prominent about these goals.

The question is I have these commitments become harder to meet than they were to make well I think your heart should be thinking at the state of global warming.

In Atlanta.

<unk> and.

We should all be concerned about that but I think your heart should be singing at the leadership position that Amazon is trying to establish at the improvements we're trying to make at the very public goals. We've said why did you remove that particular commitment so well because we it's subsumed really in a much broader boulder commitment we made it very public pledged to be net zero carbon across all of Amazon.

Not just AWS all of Amazon by 'twenty 40, which is 10 years ahead of the Paris Court now for a technology company like AWS I won't say, it's easy, but it's doable. So you'll hear that from other tech companies for a big retailer with airfreight, and inbound transportation and stores and packaging.

Actually really really hard and we will be the first to say, we don't know how we're going to get there in all dimension I can tell you how we're going to get the renewable energy I can't tell you, how we're going to get thrown all elements of transportation and packaging and buildings, but we are an innovative company. We take both long term debt and we made this pledge publicly not privately.

In order to number one you know have a forcing function for ourselves because it's not easy and also because we want to inspire other organization governments companies to join US we have over 400 signatories now if the climate pledge that's growing every month and it's not a competition against other organizations its competition against the thermometer and frankly I want other people to.

Out innovate and out innovative I want it to be beaten if you will in the in the race to become sustainable and hopefully we can be inspired by things that other people are doing so of course, we adjust our goals over time, but the thing I would focus on is this audacious goal to be net zero carbon by 'twenty and that's a line in the sand I mean, even as things like generative.

I take more and more computational resources, even as you get more efficient with compute.

That's a commitment that you can hold firm on even if you don't know how you're going to get there. We have made a public pledge we intend on getting there, but there are lots of other.

Areas of progress, we're making in the interim so for example, if you take packaging for the retail business.

The average packaging for shipment has decreased by 38% since 2015.

So which is another example of it can be sustainable and good for our business.

Our cost for us and it's much more sustainable for the planet and anytime anytime you create a win win like that it just works for everybody and becomes a really sustainable business proposition.

Let me, let's get back to AI and let me ask you about Alexa I know are slightly out of your purview, but Alexa runs on your servers.

Can alexa B, a generative AI place should it be.

Ah Lexus already as I mentioned already powered in large part by Llm's at Amazon Belt, then in production for a while now and I think that she is only going to get smarter and better and more personalized as the elegant technology expands an unapproved it feels to me as a longtime user and my wife Who's here knows.

I populated our house with them at one point it but today it feels as the theory frankly, a step behind is that would you agree with that I think that Ah, let's look I used to lessen my house I mean, you know I guess, we all have different structure different folks and we love Alexa and I like to think she loves us.

Although if you ask me that she won't actually tell you yet.

And I think let's just been getting better and better more skills better skills more understanding of of you and your likes and dislikes and your habits and I think that the the rapid improvements that we're making in generative AI are truly going to continue to transform alexa into.

A truly personalized assistance and we do want Alexa to be.

Absolutely indispensable best in the World personal assistant to you and I think that we've got a lot of work to do.

I'll leave it to the Alexa folks to filling all of those blacks over time, but we're actually very confident in that plan and very optimistic about <unk> being able to fulfill that role in People's lives that I think they're really going to love I want to get to two more in the two minutes, we have left ear, Boston predecessor, and the chassis has been kind of cutting some of that.

Big berth at Amazon, but one that he hasnt cut is the satellite plant paper and I just wonder what you see as the opportunity considering a rival system Sterling from Tesla. It does seem to be you know operational quite far ahead.

We're very optimistic about Kuiper theres huge interest from governments from enterprises from lots of other organizations are billions of people around the world are underserved with for Internet and type of really aims to democratize that there's a theme here to democratize that provide great Internet service to to so many billions of underserved people.

In addition, a whether it's the automobile companies are telecommunications companies lots of other enterprises, where AWS customers governments, they wanted especially from remote locations to be able to backhaul information up to Kai for back into the AWS cloud and so I think as we launched our first satellites, which unwound.

Coming up later this year and then really ramping up in 2024, and 2025 is my understanding and Ah, but that'd be a being in service in that timeframe, initially and being able to deliver that for AWS customers. This is huge.

Glad you just want to I do just want to.

Remind folks that just today. This morning, AWS has launched its 100 million dollar generative AI innovation Center, where we're gonna be gone out to all those customers around the world enterprises with expertise free AWS expertise solutions architect Ah.

Engineers are strategists.

And working with them one on one to envision design in an act of that 100 million AI generative AI capabilities not talk but actual have that discounts for new companies or yeah. We're going to we're just going to we're just going to bring our internal AWS experts in a freer charged a whole bunch of AWS customers.

And our folks are folks with was with a significant AWS presence and go help them turbocharge their efforts to get real with generative AI get beyond the talk alright last one and negative 10 seconds. So I guess, it's got to be equivalent you and Andy are both in the same situation you're the you're the guy after the Guy the founder whose name was synonymous with the early stage.

Age of legendary growth. So what is what do you want atoms AWS to legacy to be.

Well I don't think I don't really think of it in personal terms that to be honest with you. So I don't have a canned answer, but I will say that you know I would love it if.

If I could be known to.

Really helped drive a business that is constantly no matter, how big or get no matter, how far flying or get puts customers at the very center of what we're doing always puts customers interests before anybody else's interest yet at the same time isn't as an empathetic equitable fun and innovative place for employees to work.

Adam philosophy. Thank you for joining thank you.

Okay.

Mhm.

Please welcome to the stage wont com, President and CEO , Christiana <unk> Ahmad with Bloomberg's Ludlow.

Yes.

Yeah.

Yeah.

Hum.

Okay.

So that's more like it good morning, everybody.

You joined Qualcomm as an engineer and 1995.

Two reasons one.

It was your birthday yesterday belatedly happy birthday. Thank you. Thank you very much.

Date, you bet, you've also in New York.

A logical development, we know Qualcomm is this well as the largest maker of smartphone processes. The brains inside the handsets that are all pointing at you right now taking photos.

The question.

Is Qualcomm an AI company.

Look this is a this is a great question to ask and you know what.

It's incredible to see all of the development, you'll see right now on on AI, Here's here's how I'll answer that question directly is very simple.

If you think about the AI when you think about semiconductors really accelerating computing, how you do a lot of competition.

And what we see what you can do with those large language models. The large models for images and video. So if you think about the history of computing computing starts in the cloud and he gets scale at the edge I think that's that's what happens with Cpus.

That's what happens.

With all other form of computing and I think the smartphone is a great example of that.

If you look the largest computing platform.

Ever develop is to smartphone right now is the largest development platform for mankind and and what is good about the smartphone. It's a it's a device in it with you all the time, so if AI becomes pervasive, which we believe it will become pervasive, especially when you look about how those large mall those b R V.

Every natural how you you can converse with them because they have contextual information and all of those things that's going to happen at the edge. So that's how I should think about Qualcomm if AI is going to get scale, you're going to see a running on Qualcomm snapdragon devices, whether it's in your phone in your car in Europe P C and.

Into other machines, and I think well, it's a great opportunity for the future is.

Democratize access to artificial intelligence tools generate today, I tolls and Qualcomm is going to make that happen.

Why are you not getting like Jensen Huang level C O N E.

Look I think.

I think what's happening right now and by the way it's great for additional my conductor industry.

Anybody that has been on the forefront of computing know Qualcomm is probably used to be well known as a communication company, but actually if you look at what we do right now is more of a connected processor company than communications.

And as those models started to become very popular there.

We're going to be running at the edge and I expect that AI becomes an option on Qualcomm right now in looking I'll give you. An example, I. It's it's I saw something they Adam said I think in the prior conversation when he said something about in 1997, he said try to guess where the winners and losers on the internet.

Probably very wild guess I think what we see today is this generative AI opportunity is huge we don't know yet all of the different applications that are going to come up we are seeing that just just within the past six months is the resolution of a number of companies come in with their use cases and those use cases are going to happen.

Devices, and I think that's going to be a great opportunity for hold that thought what we're gonna do now I'm going to show you something so the audience here and those with us virtually but during that think about questions for Christiana based on what you say and so with that I bring up the video and Christiane when it comes up in place.

Explain to us a little bit what it is that we're seeing because here at the Bloomberg technology Summit, we're gonna nailed the technology any second just just way the video is going to come.

And when it does it will have been worth the wait.

Yes, so what you basically see is a contra net demo you'll have an input image on your phone you tell input prompt what do you want the image to be wanted to make it a masterpiece look like Venice canals for K and it just runs in and give you. This very unique image image too.

Image, that's never been created before created to AI running on your phone. So it's a good I think time to talk a little bit about how we think about AI at the edge outside of data center because they like we have seen everywhere, there's going to be this huge opportunity for the cloud, but it's going to be this huge opportunity for devices because what you.

Due on the device is very different so there's there's a number of reasons why this is going to be very popular on the device first the device has contextual information about you and has real time information like a picture you just took.

And you wanted right now at the moment it changed that picture in share with somebody else with your <unk>.

Messaging platform the context that video that device was run in airplane mode without any external connection it ran the model locally on device absolutely.

So that's one of the reasons you have real time contextual information Theres another REIT.

Processing on the phone is virtually free.

When you think about your running those models in the cloud and and think about a large languish model for every token like award as the sentence. If you do that if you have an experiment that you'll see the awards coming up for every token you're running the entire model and you use.

A lot of cloud computing, but the computing that is in the palm of your hand in the dashboard of your car in your computer that processing power is available and is dedicated to you. So that's another reason you're going to run those on the devices the.

The number three reason is you can see this and Theres a lot of academia work those models when they get very well trained they become smaller so one of the things. We have done it is very unique to Qualcomm. We spent a decade on this I think we we invested in this before it was popular.

It's kind of the history of many of the things that and Qualcomm down on research. What we did is could we come up with ability to do very high process a high performance accelerated computing on the device running large number of power meters and you can do it for example in a phone with.

Compromising battery life, because you can never trade the form factor and a battery life you expect it to do all of this competition is still last holiday. So we develop some very unique technology, which is the most efficient accelerated computing from a performance per watt and we're bringing that to all of our devices just to give an order.

Magnitude.

The new Snapdragon processor, we're going to launch this year for flagships they've come to the market 24 is going to support in excess of 10 billion Powermeter models running locally on the device.

What are we going to do it in a windows on arm platform is going to be over $20 billion.

Part of meters, while we're going to do in cars is going to be in the order of 40 to 60 billion power meters running on the device. So so what do you think about it when you see all of those models being well trained and become smaller we going to see the development of what we call hybrid AI.

You'll do a query like you just did the square if you what this picture you wanted to do this if the model very well trained to run on a device and there is going to cost nothing well I think the question that the audience will have and I have as you say shipping in 2020 full but how real is this that the demonstration showed you're running a one five.

Billion parameter model when can we do that Oh real it's a it's a it's more real than you think for example, stable diffusion, which text to image.

This summer, we're getting to sub five seconds.

If with no cloud connectivity.

In in our Snapdragon summit, we Gonna show 10 billion power meters running and you'll be able to buy for example, if you want to buy a galaxy <unk> 20 for.

The launches next year Youre going to have that computing power. So there are lots of engineers I speak to you all yourself, an engineer and they they do expressed skepticism that this is going to happen that all of us will be running.

Large language models or generated AI tools locally on device without connection with billions of parameters because they doubt the processor power or they doubt the work you can do.

How come on the algorithms where did you do the innovation. Okay. So so this is a this is a good question and that's why I was about to tell you about how we think about hybrid AI exactly like your phone works today, you have a lot of applications in your phone. It will go to the App store you have a lot of apps in the apps are being processing on the device and their bring.

Assessing the cloud they work as a single.

Computing, so hybrid AI is what's happening right now is really that you have a model running locally you provide the query to the model. The model will decide if he runs locally or you'll give a cloud a head start.

And sending tokens to the cloud so it's going to be completely transparent to you and some applications is not just because you don't want to use the cloud you want the fast response on the device because of latency or contextual information, but here's how I would like to explain this.

When the smartphone happens I think most people don't remember this I'm sure I'm going to show My age I've been in every transition of technology of Qualcomm. We were the first company that came up with the concept at the time, we got a palm OS a palm pilot connector with salary and created a very for smartphone into warm and.

This is the time of the feature phones you use your phone to make calls and texts, that's where you use your phone for and.

In GBS was just 40 911, that's the only application all of a sudden this became a computer and people say, what I'm going to do with it I like to put this example, I remember when we talk about for G broadband.

There was a lot of analysts with their black berries are telling us who wants this what what are you going to do with this this is.

Have everything I need in my Blackberry It was incredible the vital we should acknowledge all pull went out blackberries not really a thing anymore as well.

What it shows is the revolution that happened with all of those applications on their phone. That's how you should think about AI with just the very beginning.

We're just showing some examples stable diffusion concho and that we're going to have some simple photo, but theres going to be so many different applications think about a car for example, I like this example, because as a kid I used to watch Knight rider kit.

Goodbye.

Natural language conversation is perfect for car requires Qualcomm relevant to that Paul is just look at what we're doing right now I think it probably I don't have to tell you about our position in the mobile market I think are.

Especially on an premium computing for the Android space.

We're now virtually provide of working with every brand virtually every brand in automotive for next generation digital cockpit.

We've been working with Microsoft for the transition of jet next generation P seats are using our silicon we have been the partner of choice of companies like meta, Microsoft announced with Google and Samsung for augmented reality virtual reality mixed reality devices and we've been expanding too.

In industrial Internet of things, let's let's let's hit pause for a second because I know that you are a busy man and you were operating in a lot of fields go back to that video in this audience question why would I want to locally predecessor complex model on a small device.

When those models would run better in the cloud I guess that applies to the automotive example, as well yes. So this is a.

Instead I provided the answers to those questions in the phone, but I'm going to change the conversation a bit and talk about in a car.

So when you are inside the car right and.

In like for example, we have a lot of AI into car today for assisted driving and autonomy.

Just to give an example for Luca Jim.

Super cruise Ultra cruise all running on a Qualcomm AI processing, we have a lot of AI processing in the car the car needs to make decision that are very context related in real time. If you. For example, when do you think about on a das system assist.

Assisted driving the sensors into car for a four assisted driving autonomy see an image you need to make a decision.

That computation needs to happen locally the reason Qualcomm became successful and were in expanding in a desk because you cannot put a server in a trunk of a car, especially in an EV takeaway from the range, but now the computing power you can use for large language model in the car this model can be.

Big as the model in the cloud and do you have now the real context information you talking let's say you're talking it acquired the future you are talking to the model like you would in Knight Rider Kit and you want you won't you give a very complex instruction for the car I want to go home on it.

<unk> home I want to stop here I want to order. This I want to pick it up and those things are going to be very related to the information you have it doesn't mean that you're separating the eye from the car from the cloud, they're all combined but everything that is contextual rich for that moment.

And of course.

Dave weeks months, probably years when does that look it's very difficult to make a prediction. It's the same try to make a prediction but broadly.

Look I'm I'm I'm, an optimist. So I think we're going to start to see next year in 'twenty four for phone applications a lot richer.

Thank you are we going to see a lot richer ability to use AI for photography. Like This example would just give.

How people share information, whether it's on whatsapp or Google messages, how theyre going to share photos, how theyre going to create content you're going to see that next year, I think you're going to see a lot of productivity.

Next year, we talk with Microsoft We had incredible demos had Microsoft build showing the co pilots off all the Microsoft applications running on our eight AI engine in.

Our future S. You'll see that as a custom CPU that going to the windows devices.

The connection and 24 and.

See this is another interesting thing enterprise applications.

When a chest JBT thing happened it was an incredible thing.

I immediately had a send an internal memo say our engineers can't get our source code and just send us a check GBT to verify the scope because that's going to go into the wilderness right.

So enterprises really want is for certain applications to run the data locally so so Microsoft Ed's willing the incredible technology that you can provide a dataset Luke will be an excel spreadsheet and you apply the model on that spreadsheet. That's a great example.

Why you want to run those things locally some enterprises may not want to send the data to the cloud and that's how the enterprise works.

I know you and I knew that you would be fired up about this and I've only got 30 seconds left so let's do this.

How is it doing business with Beijing at the moment look we have a great partnership.

<unk> R or customers in China, I think Qualcomm is probably.

One of the few examples of a very successful.

Partnership and business with China, all of our customers in China pay intellectual property on cellular essential patents.

We're all license they pay we have a have a vibrant business on the phone space with companies like Xiaomi, Oppo, oneplus, vivo and honoree et cetera, and as China build an incredible industry for EV. They all using Balkan chipsets are we going in industrial.

So here's how I see it.

If you are a leader on a particular technology.

You will have a big business in China is just the nature of the size of the GDP and that's what we see happening with welcome Ray very quickly or Brad and Emily will be very angry at me just AI complicate that relationship with China No look.

You should think if you shrink about what's happening in phones, and Pcs and cars everything at the edge.

It's one of those things that is very difficult to make predictions in politics, but to this date, we have being able to continue to support development on on the voice AI and I think across all different industries in China, and there's some very interesting development, especially from the phone industry.

They're using this to create content or support creators.

I'm on Qualcomm see.

Making the case basically for the future of AI Bang on what your old pointing at me right now the smartphone. Thank you very much. Thank you. Thank you have a pleasure to be here.

Yep.

Please welcome to the stage Adobe generative AI in Sensei, VP, Alexandru, Carsten and creative artists Agency, Chief Legal Officer, Hilary Green with Bloomberg's Rachel mess.

Yes.

Yeah.

Good morning, Good morning to you guys. Thank you both so much for being here.

You talk about.

AI and artists and there's just so much happening there AI has long been used by artists and it's gotten easier and cheaper and more widespread I'm a system like stable to fusion in Sally and CECI, Beachy, Adobe's, Firefly, which I know you've been working on them and so on so what are some of the most interesting. He says that the two of you are seeing for generative AI.

Creating art.

Well I think there is tons of examples AI like many things before is just a new tool for the creative community and the creative community is very adept at taking on new tools and and applying human creativity.

To get new art forms one interesting we want this thing we've seen in just the past couple of weeks since theres going to be a release of a new Beatles song there was a tape of.

John Lennon's voice, where the quality was very poor in hand as families said no. We can't make us hung out of it and AI has now been able to clean that up to make it a high enough quality that Paul Mccartney with the consent of the family is going to release, the first new beautiful song in decades.

They released others in 95, but the technology wasn't good enough for it that's one that's a fantastic.

Location of the tool we also see it being used in post production.

In order to make a lot of things easier or harder before you can do a lot of things without the presence of the talent by using scans and that's going to be another fantastic application and it's gonna the Sky's the limit I think we're only beginning to scratch the surface.

Fifth on where the creativity will lead us and Alexander what kinds of things are you seeing.

Same in our in our world when we look at the spectrum of art and creativity in general starting from Fine Arts to consumer content creation. The two biggest.

It uses we've seen recently relaunched generative feeling photoshop for photo retouching and this took off like a one of our most successful data. So a lot of actual production workflows and content creation workflows for our professionals are being helped by Germans are there any ways and we're also enabling some of the technology for consumers in.

Communicators that don't have the skills to actually create and we're seeing an actual explosion of creativity of that consumer side folks that have ideas that in the past they werent able to express themselves now with the help of generative AI directly integrated in all workloads, they're able to create content and have a differentiated social presence.

Just communicate with a with their families and Alexandria, you also have a lot of people that are using firefly either.

You mentioned are doing work related projects with it like what are some of the things that people are doing with it that you might not unexpected.

So we're still in beta so our general availability release says is happening very soon so everybody is still experimenting on the front of my website that we've seen the we've seen a lot of viral activity with people.

Extending out Beatles album Africa. For example, this was something we didn't expect to be as popular and viral or or a mixing means together like really really crazy stuff and again. This is just human creativity is best when they come up with new ideas and technology. Just enables this new wave of creativity.

What are some of the concerns that you have you have about the use of generative AI in the art world like ethical concerns, particularly around how data is collected the train systems that can generate texture images.

This stuff is typically collected from all over the internet, but increasingly some people such as artists in particular, whose work may have been sort of Hoover it up as part of a dataset.

Sorry about that yeah, well I think you're seeing I mean, we've already seen the lawsuits I think getty images. That's one people challenging the practice on what material you train the machines and the ethics behind what's used to train machines. There is a difference between hub.

Available and legally available for your use to make money and the whole intellectual property regime.

Give us rights to certain individual creators and people who appear and when a machine is trained for commercial purpose using those rights without their permission.

You're at serious risk.

I think if we think about actor.

And how important the creative decisions are that they make not just when they're doing a role but as to what roles. They agreed to do in order to create their whole body of work.

Real curatorial expression of self and if other people can take that away from them.

By using their image and without their consent using either their prior creative thoughts in the form of a writer or their image or their voice for a musician in order to make a product that they're unaware of or having consented to ban I think we're in very murky. We're.

Or even not that market pretty black and white.

The coal.

Land, where we have to think about how do we create the rules and the norms around use of these creative individuals work.

In a way that respects their rights and doesn't to violence, we heard a lot of talk this morning about threatening of humanity and obviously.

Human extinction kind of an important big deal I get it but along the way we shouldn't think about human extinction.

Just think about mapping basic ethics and norms onto this fantastically creative and powerful tool.

And this was top of mind for us too when we design Firefly because.

Artists, where indeed the.

Up in arms about the fact that their public content was used without their consent compensation or control. So when we designed Firefly, we actually decided to train only on data we had an adobe stock in public the available data always copyrights expired and on purpose. We don't include any recognizable characters don't think with any IP trade.

So by the nature of the training data not only we're respectful with our customers. Because these artists are basically all of them are our customers, but we're also enabling enterprises and creators of content generative content to be comfortable using this content because the risk of generating by mistake or.

Somebody else's IP or recognizable character is zero.

So Alexandra I have a question for you related to that because Adobe recently said that business uses a firefly can get an IP indemnity from Adobe, which essentially as I understand it means that if a business customer has sued over some media that it makes a firefly, let's say can add hypothetically adobe will pick up the legal path right.

And again, our comfort comes from the fact that we.

We trained on data, we control and that's curated both using AI, but more importantly, it's curated using human curators and we make sure we don't leak any IP in it. So this is why we have this comfort for our enterprise customers and for cases, where they actually is just where the generated content at the remix it with potentially infringing content in.

That case, the ipi indemnity doesn't apply but for a pure content generated by Firefly. We are offering this IP indemnity for enterprise, but I think it's fantastically interesting and pioneering work the idea of ethical training I hope it becomes part of the.

Greater lexicon, and that's what's happening here and it's important for the corporate and business clients, who have that indemnity, but we were also talking about the democratization of creativity and allowing somebody like me who lacks it to have the tools to perhaps create thing I don't wanna be stealing somebody else's.

Work as I'm exploring my creativity for the first time and so having.

Tools and mechanisms that would give me or every other creators of confidence when they are diving in to this world for the first time and they're not doing it in a way that's doing violence to other creative people who've invested their lifetime and the work I think is super important and I really commend adobe for what they've done.

Thank you.

I'm curious, though what is adobe's layers thick of that and Hillary as a lawyer yourself, maybe you got some thoughts on that I mean.

They narrow our chief legal officer would be a better person to answer this but they they're fully behind this and for a long time actually we've been pioneering pioneering many other innovative things in the context of journeys if content like the content or the authenticity the initiatives that they want to touch upon later on but we do have again, our legal department.

To support them because they know how we train they are deeply involved they actually our legal department runs what's called our AI Ethics program. So we have a program we've set up many years ago in 2020, when we've started pioneering.

Generative AI using gas, which is a different type of model with neural filters in Photoshop, we had the our legal department full engagement. So we've created this ethics department that helped us navigate all of those not only training data issues, but also bias and harm reduction initiatives et cetera. So we do have a long history of investing thoughtfully.

Really in this space, Okay. Hillary I have a question for you specifically about some stuff that's going on right now in Hollywood.

This month the biggest union in Hollywood the SEC after actors Union began negotiating our studios regarding a new three year contract one of the issues. They're discussing is the use of AI to do things like mimic users' voices or their bodies.

So I'm curious like how you envision generative AI changing Hollywood and stands to benefit as it tends to get hurt from that well.

Think that's a question that remains on the answer it and I'm not myself involved in the Guild conversations I hasten to point out it's not only the screen actors Guild the Directors' Guild.

<unk> reached an agreement with the studios.

And the writers Guild is also in the mix and I think we're gonna have to wait to see how what all of those negotiations on yield but in the end I'll go back to where I started as a tool for creativity I think we're gonna see an explosion of content in Hollywood there could be good for.

Hollywood, but only if we pay attention to establishing the ethical norms.

And creating a world in which.

Before the law or legislation and catch up the genius, let out of the bottle because it has normalized to steal all the content that's available but not legally available for you and also I think we're in a little bit of a time race.

To get to the ethical standards to make sure that this tool in the ENT delivers on the creative promise of it both for that amateur but all the way up through the chain of Hollywood One more thing I wanted to ask you about unrelated to actors and contracts and stuff I'm curious if you're seeing on AI.

The applications in general.

Discussed or put into some contracts and if so how and what kind of questions does this bring up for artists protecting their image as well.

Well, so you know.

AI generative AI as you, but the idea of special effects is is not new the ability to manipulate images to change them post production has existed for a long period of time. So we have some precedent here.

And it is pretty well established that the studios are.

Seek consent from the artist before they will fundamentally change their performance there I'm going to forget the actor, but there was an example of two decades ago, where they added a tier two and the actors performance that he didn't approve of and when he saw that film and he had it here he was like how bad.

And this conversation started that long ago, and it's continuing there. It's a question of who owns the character.

How long is the digital image of the actor who creates the character what controls should that actor have over use of those tools in the future. Both in respect to the piece of art that they've agreed to make and with respect to other pieces of art I think ultimately we're going to get.

To a comfortable place with the studios because the studios and artists are in a symbiotic relationship more concerning is those who are not part of the mainstream but who want to jump.

I jumped in by creating work they now have the ability to create that much lower cost because of the availability of machines that have potentially been trained on the pilfered content, that's a real problem.

Let's talk a little bit about audience awareness. This is something that I think about whenever I see a trailer that might use AI in a certain way or I've seen some commercials that are using AI.

When should the audience be made aware.

As part of the creative process.

Why and why not in Alexandria way no Firefly doesn't embedded details tomorrow KAR generated images, but I'm not sure how that translates to like a final product yeah. So for a long time as we were pioneering those the generative work we were receiving this media attention and industry feedback that.

We need to invest in content authenticity, we need to basically help content consumers understand whether a piece of content was digitally edited and in this case more recently fully generated synthetically and four years ago, our legal department in an area. We pioneered the creation of the content authenticity initiative, which is a <unk>.

Consortium that has a thousand members now starting from camera manufacturers like Qualcomm and others and ending up with a website <unk> com.

Content consumption places and across the stack, we are encouraging and the implementation of a nutrition label digital nutritional label, which we call content credentials that is a piece of information that describes what modifications were brought so there's content and that's embedded cryptographically the content itself.

So that means that somebody consuming a piece of content with content credentials embedded they should be able to read the nutrition label. They should be able to understand is this piece of content real or is it generated or edited so it doesn't represent the original intent and this is extremely important in this age where it's so easy to create this kind of imagine the.

One of the faith and politics and I mean, this is a really serious issues and this is why we're encouraging a lot of our partners. We have we're working with generative companies like stability AI and funny that the EIA and others to standardize on this content credentials.

Initiatives. This initiative so consumers of content has the option to first verify and then trust the piece of content today, you'll see a photo it looks real.

Politically believe it out of the immediately but in reality in this near future, where all content will be modified digitally in or generated fully we need to to enable content to embed. This authenticity information. So consumers can have the rights of understanding and then verifying and then.

Trusting that piece of information.

I agree with all of that and I believe that some form of water marking for or otherwise.

You know being able to trace the provenance of that piece of art back to the consent of the artists to are included should be a part of this kind of nutrition label and the reason I go for a watermark is because it's not just as you're watching the whole content as we know somebody can take a clip of that that may or may not have the nutrition label.

Put it on our platform and it can be around the world before the artist even noticed what's happened to them, but if we can embed something that'll pop that'll show not okay or okay.

That's part of the solution are we there yet technologically I don't think so but with the smart guys sitting next to me working on it and others and if we take them at their word the leaders we heard from from this couch. This morning.

That we can do a lot to create.

Create these tools that'll make it much safer environment.

So much for talking to me about this it was wonderful to have you here. Thank you, let's give them a round of applause.

Okay.

You got to just walk away.

Thank you.

Okay.

And now I'd like to welcome him to the stage Madness, Saks CEO and founder of stability AI.

Yes.

Yeah.

Yes.

Yeah.

[music].

Yeah.

Thank you so much for joining me we have a lot to talk about yes, it's been it hasn't been exactly 10 months instability really stabilization public right. Yeah, We released it with cognizant runway 10 months ago doesn't feel like it feels like 10 years. Okay. So like what has changed since that because a lot has changed.

I think the you know when we released it was building on the research with many others in this whole ecosystem is the first large scale usable muddle on someone's laptop. So that's why you saw such a massive take couple of it like it overtook the cordon that there had been three months would've been 10 years in developing interest on Github and now cumulatively.

Ecosystem is I would take it limits how this works how GPT and others come you know how where we are now which is this technology everything everywhere.

And everyone's talking about it.

Let's talk about the new model that's out today tell us about SD XL 0.9, and what it means for stability.

And if we can see some of those images.

Yeah. So I think you know this continuous evolution said three of the five stable diffusion authors and cleaner to Leeds, we're lucky they work as stability plus again this whole ecosystem that took an open model and they just improved intuitively. So SPX. So we kind of scaled up and introduce some new architectures. So it cannot do hands.

You know that was always the thing that sounds kind of words, I mean like kind of idea how does that kind of you it's kind of difficult perfect hands classic text photo realistic images now.

But also with some mitigation philson because again, that's why it was important to put it out into the open so everyone could both improve it I'm going to attack. It. So it becomes more robust he said mitigation what does that mean exactly. So you know you have things like we.

We had photo realistic models earlier this year that we didnt release, because the time it wasn't quite right for it because you have to build in things like invisible water market working with Adobe on content authenticity to Oregon, many others as well because the technology is inevitable because you see it emerging everywhere and part of what we want to have stability of industry standards around that so you can track it so.

You can use it and use it appropriately so like right now if I were to use the latest model, let's say that generated this picture or something similar to it since it's available as of today.

Would it have some sort of embedded watermark, yes, honestly difficult full details about that so this is a official what are some unofficial ones. Because these models become the default that people use and you want to have some way of telling what his generators versus what is going on and that's a huge research area.

Excuse me a persistent a persistent issue with all kinds of AI in the generators, including stabilization is bias in the images that are generated as we know that sentence in large part from the training data how stability I addressing bias in it's stable the fusion models and how do you improve that.

Yeah, I think there was a great research report on one of your colleagues I'm kind of where they looked at the original stable diffusion that wasn't as academic release buses because.

It was taken out of the picture of the available Internet and it reflects the biases of that Internet, which is very occasion western than others.

One of our missions stability is to build the foundation to activate your margin potential. So we're working with multiple governments radio national datasets that reflects the culture of those countries. For example, one of our team members in the calculation.

Japan's stable diffusion, where they slept how the language version because if you typed in salary months. The original stable diffusion. It was a very happy about it brought some money okay. The Japanese vessels, a very sad for US, yes, and I think this is the future of every single countries. This is vital infrastructure they need their own datasets, they need to have their own models and they need to be able to interact to reflect those contexts and address these biases.

No we can't do that with black boxes, there's no way to have one model to rule them all that reflects everyone's context.

Are you you're working right now on different datasets for different countries. Yeah. So you take national broadcast today teacher Education data you take cultural days and he built from that because I think from a safety perspective.

We concentrate on the whole internet anymore, instead sort of garbage and so that's one of the reasons, we have opt outs of datasets and we have massive investments in refining datasets, because we should move to having that as a common infrastructure. Both for the proprietary models of the opening hours and in small pixel the world and also for the open models because.

Yeah. I mean, you are what you eat and right now the models feed well fed biased garbage no wonder they turned out a bit with let's feed them better stuff. So we have nice pretty range organic models.

Our country and I am building, a dataset that I want to be use for a model, who who owns that dataset is that as belonging to the country.

Who is in charge of five I believe it should be national infrastructure in the U K, we have H D. L U K, which organize the national biology, and data information infrastructure with Federated learning about this I think this will be infrastructure for every country every country hesitance. It because you have public broadcasting data.

That should be used to power AI models, and then that will lead to the innovation that we also required to deal with what's coming because this technology is going everywhere. It's more enterprise ready next year that kicks off of anything but I think they can solve the disruption here is innovation. So it becomes a vital for countries to Stoke that innovation through open data.

And models and innovation in Huntsville.

I'm going to ask you a little bit about finance.

At year stability raised 101 million on a valuation of 1 billion that we understand you're trying to raise more can you talk about how much your friend right.

Well, we're not trying to raise a little but we have constant interest yesterday received eight and coming investor interest and that's because we are a bit different artists addressable market is what the private data around the world and Sunny intelligent, whereas everyone else is looking at proprietary models, which I believe is a race to zero so through amazing partnerships like with AWS and more to be announced our models will be available anywhere.

Because our aim is to have the default model of every modality commercial variance in Nashville variable. So you can have an Indian insurance stable chat.

I think you know you scale, when it's ready and we're nearly ready.

But you don't try to over scale and it's very difficult because to build these models you need supercomputers you need all of these things, but once they're built they can go everywhere, which is amazing operational leverage.

There was a recent Forbes article.

That said, particularly that youth exaggerated at times about a few different things how.

How does that mean fund or anything any more challenging and then what about your existing investors have they how have they responded to that.

Existing investors who've been having close dialogue with them they've been helping through this process.

In terms of additional fundraising again, we don't actually fundraising right now, but the interest is continues to go up because ultimately you know, it's just things like AWS partnership they built US one of the largest supercomputers in the world before funding and we have an amazing partnership they talk about technology. We released another amazing image model, we have language models code loves and more coming I think.

An interesting thing because a year ago were preceded startup and our scale up for next year, we might be a multinational growing pains are always there, but the ultimate thing is you just have to deliver on what you say you're going to do you mentioned language model and I know that a few months ago on your at least stable L. M. So what's happening with that because that was like a pretty early version when it came out yeah. So next week.

We're moving to properly open development. So we have a blog and a whole series of papers out with that and we had an alpha and now we're going to have a beta.

Which is a massive step up from that so last week, we released kind of open llama, which was a replication of fundamentals Lama, except for fully open and I was going to be a highly refined version I think what you're going to see with language models. Now we just had Qualcomm cristiana coming on is that you will probably see GPT $3 five level performance on the edge by next year, if I look at current trends.

A large part of that is data so with the right data you can have a 10 times performance increase and that's something we saw when we funded data comp on the image side with a 10th of the data.

Formed opened my eyes slip image detection model because the data is such high quality again stopped feeding the model's garbage get rid of all the rubbish and you'll have highly efficient architectures.

I wanted to go back just for a moment to the Forbes article on something he told my colleague Tom Mackenzie recently, he asked you about claims in a Forbes article that you have a tendency to exaggerate you said I get excited.

And I'm curious what are you putting in place to make sure your storytelling in your narrative Nash with what's actually happening on the ground I think it's surround yourself with good people.

I was thinking a lot about this I've asked birds in ADHD and I have a very definitive view of the future and I think that shocks people because he candela exponentially, what's coming though a territory like seismic has 100000.

SME customers every single one will use this technology at the same time or since it enterprise ready.

The biggest economic impact in deflationary thing that's going to hit our economy of all time, I say something like that it sounds like an exaggeration, because I'm definitive, but it's what I genuinely believe and again. This is a difficulty of health and I'm thinking the best I can one thing I didnt do any worries I commit to release states. Okay.

And so am I to IV bashing me over the head about that then we'll really state commitments just released when it's ready Okay, I understand and I think sometimes with articles I just kind of say yeah, it'll be you know.

Yeah, exactly like we nearly have the stills distribute stable diffusion ready and it was super fast and it just didn't it didn't really fit into the bed.

Okay.

Hum.

One concern that people have raised in general about.

Generative AI tech space and.

The systems that are on creating images is the extent to which startups and large companies are looking to bring in people that can work on the epic side of AI and people, who can develop guardrails I'm within companies. What are you doing on this do you have any AI emphasis on stuff I think it's interesting. So my background is in ethics in Islamic law, it's a very different type of ethics.

In some ways, but I'm conversational geologist.

And so we have our safety team that we're building out we work with kind of a relationship teams and others because the models have us deep learning trained and then you use all relate trefry enforceable anyway, even if you talk to the room towards human and so we have Altair all ex library, which is the most popular one from a copper AI lab, and we've been bringing in emphasis and others from external parties to look at that.

But no one's quite sure AI ethics is right now.

I'm not sure what what AI ethics is right now okay. So part of the thing is to expand the conversation. This is also why you know we support them AI insights to Oregon other things because of the conversation doesn't need to be brought in because it is so complex and it's hitting every part of society that was one reason why I was the only AI CEO to sign both of the F. L. I N safety letters, because I think that's a discussion we have to have.

Now one more voices needs people and because right now it's mostly guys here in Silicon Valley and I think that's correct.

I truly feel that AI is at this moment I think it will be like that maybe her where we're kind of boring and those bugger off thanks for all the Gpus.

I could be wrong.

And that's why I think right now we need to move before the economic impact happens just like Covid before Tom Hanks.

Didn't have that moment, yet everyone's talking about he loves talking about it but it's not real next year becomes real in the economic impacts become huge existential side from having crap data that were feeding these models the whole of Youtube and others I wonder it sounds that wed so let's feed at that of data now, let's put some proper standards in place and let's build AI.

Is good for us and doesn't work against us.

It feels almost like.

Do you think that that is inevitable and that's an inevitable.

Outcome do you feel like that isn't inevitable outcome I guess from my perspective, I'm like well if this hasn't happened yet there are so many different directions that things could go I think nevertheless, same again is an interesting thing never.

Nevertheless, same again, okay. So no head teacher in the school concert assays for homework enable because of safety for its never the same again.

We're going to see a lot of those a regime change over the next period and this technology is inevitable and it's up to us it's going to be big economic impact that COVID-19 is it up or down.

Is it productivity enhancement as a job losses, we're not sure about that and when the existential side, we know, but my view is that if you have an AI that is more capable than you're.

The only way to fully control it and our liner is to remove its freedom I don't know how to do that wants to use them more capable than me.

I can do is I can teach it better I think the way that you brought up it makes a big difference we see that in real life. These models are like really talented grabs locations off their meds, we need to fix that we need to give them better food better balance and then I think it's more likely that we will have a positive future, especially because the bad guys already have the technology you can just download on USB stick and then you've got it.

What is the most pressing challenge stability is.

AI is working on right now I know you guys are involved in lots of different things. So sometimes it's hard for me from the outside to see what you think is the most important well the most important challenge for us is to be more transparent when there's so much going on and build the human capital.

But in particular, the key focus that we have right now is getting language models to the edge the ability to earn your loan language models can help you protect against other language models that will become increasingly convincing and I think everyone needs to be able to own their own models, given how powerful they are and how they extend all capabilities just like in crypto. There was about phrase not European what Youll crypto I think is going to be not.

Your models, what your mind as we move from here.

Explain that a little bit more if you are outsourcing a lot of your cognitive capabilities and models for the black boxes and they have inherent biases, then it's going to be more and more information on your vessels. If you have control of your own models for your private data, you'll private knowledge, yes, Rachel as Bloomberg as a country and so its a question of who controls this technology.

But both extends your capability, but then you rely on increasingly.

So I think that needs to be open flexible alternatives to them, but.

But might not want to do that at all and I do not want to be involved in that period.

But at least you should have the choice right now you do not have a choice if you're one of the most advanced technology as a black box that can be turned off anytime like open AI.

Amazing image model Daddy too that they found all Ukrainian content from from nine months in all Ukrainians for music and.

It's complete within their prerogative because of political issues, but imagine if that was the only one that you didnt have stable diffusion.

I don't think that would be nice.

Thank you so much it was really great to speak with you.

[music], it's time for a short break so please head out for coffee and stretch our legs and also if you would check out the grown the chrome enterprise activation out there.

Which is an immersive experience and our web security, there's a game I failed. It I think I'll, probably do better but try out the game.

And then be back here at 10 55 am for our next set of interviews. Thank you.

[noise].

Uh huh.

Hum.

[noise].

Okay.

Sure.

No.

Okay.

Sure.

Yeah.

Yes.

Uh huh.

[noise].

Hum.

Okay.

Hum.

Uh huh.

Okay.

Hum.

Okay.

Mhm.

Okay.

Okay.

Hum.

Okay.

Yes.

Okay.

[noise].

Hum.

Yes.

Yes.

Yeah.

Hum.

[noise].

Yeah.

Yeah.

Yeah.

[noise].

Yes.

Yeah.

Yes.

[noise].

Hum.

Yes.

Right.

Yeah.

[music].

Yes.

Okay.

Hum.

[noise].

Hum.

Yeah.

Yeah.

No.

[music].

Yes.

[music].

The program will begin in 10 minutes.

[music].

Hum.

Yeah.

Uh huh.

Uh huh.

Yeah.

[noise].

Sure.

Hum.

[noise]. Please take your seats the program will begin in five minutes.

[music].

Yeah.

Yes.

Yes.

[noise].

Hum.

Okay.

Okay.

Mhm.

Yeah.

Okay.