Q2 2025 NVIDIA Corp Earnings Call - Q&A

Abby: Good afternoon, my name is Abby, and I will be your conference operator today. At this time, I would like to welcome everyone to Nvidia's second quarter earnings call.

Operator: At this time, I would like to welcome everyone to NVIDIA's second quarter earnings call. All lines have been placed on mute to prevent any background noise.

Operator: After the speaker's remarks, there will be a question-and-answer session. If you would like to ask a question during that time, simply press the star key followed by the number one on your telephone keypad. If you would like to withdraw your question, press star one a second time.

Speaker Change: All lines have been placed on mute to prevent any background noise.

Speaker Change: After the speaker's remarks, there will be a question and answer session. If you would like to ask a question during that time, simply press the star key, followed by the number one on your telephone keypad. If you would like to withdraw your question, press star one a second time.

Stewart Stecker: Thank you, and Mr. Stewart Stecker, you may begin your conference. Thank you. Good afternoon, everyone, and welcome to NVIDIA's conference call for the second quarter of fiscal 2025.

Speaker Change: Thanks for watching!

Speaker Change: and Mr. Stewart Stecker, you may begin your conference.

Stewart Stecker: Thank you. Good afternoon everyone, and welcome to Nvidia's Conference Call for the second quarter of Cisco 2025.

Stewart Stecker: With me today from NVIDIA, our Jensen Wong, President and Chief Executive Officer, and Colette Cress, Executive Vice President and Chief Financial Officer. I would like to remind you that our call is being webcast live on NVIDIA's Investor Relations website. The webcast will be available for replay until the conference call to discuss our financial results for the third quarter of fiscal 2025. The content of today's call is NVIDIA's property. It cannot be reproduced or transcribed without prior written consent. During this call, we may make forward-looking statements based on current expectations. These are subject to a number of risks, significant risks, and uncertainties, and our actual results may differ materially.

Speaker Change: With me today from Nvidia, for Jensen Huang, President and Chief Executive Officer, can collect Kress, Executive Vice President and Chief Financial Officer.

Speaker Change: I would like to remind you that our call is being webcast live on Nvidia's investor relations website. Our webcast will be available for replay until the conference call to discuss our financial results for the third quarter of fiscal 2025.

Speaker Change: The content of today's call is in video's property. It cannot be reproduced or transcribed without prior written consent.

Speaker Change: In this call, we may make forward-looking statements based on current expectation.

Speaker Change: These are subject to a number of risks, significant risks, and uncertainties, and our actual results may differ materially.

Stewart Stecker: For discussion of factors that could affect our future financial results and business, please refer to this disclosure in today's earnings release. Our most recent forms, 10-K and 10-Q, and the reports that we may file on Form 8-K with the Securities and Exchange Commission. All our statements are made as of today, August 28th, 2024, based on information currently available to us. Except it's required by law, we assume no obligation to update any such statements. During this call, we will discuss non-GAAP financial measures. You can find a reconciliation of these non-GAAP financial measures to GAAP financial measures in our CFO commentary, which is posted on our website.

Speaker Change: For discussion of factors that could affect our future financial results and business.

Speaker Change: Please refer to this closure in today's earnings release, our most recent forms 10K and 10Q, and the reports that we may file on Form 8K with the Securities Exchange Commission.

Speaker Change: All their statements are made after today, August 28, 2024, based on information currently available to us.

Speaker Change: Acceptance required by law. We assume no obligation to update any such statements.

Speaker Change: During this call, we will discuss non-GAAP financial measures. You can find a reconciliation of these non-GAAP financial measures to GAAP financial measures in our COFO commentary, which is posted on our website.

Stewart Stecker: Let me highlight an upcoming event for the financial community. We will be attending the Goldman Sachs, Communicopia, and Technology Conference on September 11th in San Francisco, where Jensen will participate in a keynote fireside chat.

Jensen: Let me highlight an upcoming event for the financial community. We will be attending the Goldman Sachs Community Copia and Technology Conference on September 11th, in San Francisco. For Jensen, we will participate in a keynote fireside chat.

Stewart Stecker: Our earnings call to discuss the results of our third quarter of fiscal 2025 is scheduled for Wednesday, November 20th, 2024.

Jensen: Our earnings call to discuss the results of our third quarter of fiscal 2025 is scheduled for Wednesday, November 20, 2024. With that, let me turn the call over to Colette.

Colette Kress: With that, let me turn the call over to Collette. Thanks to our Q2, was another record corner. Revenue of 30 billion was up 15% sequentially and up 122% year on year, and well above our outlook of 28 billion.

Speaker Change: Thanks to our Q2 with another record corner. Revenue of 30 billion was up 15% sequentially and up 122% year on year. And while above our outlook of 28 billion.

Operator: At this time, I would like to welcome everyone to Nvidia's second quarter earnings call. All lines have been placed on mute to prevent any background noise. After the speaker's remarks, there will be a question and answer session. If you would like to ask a question during that time, simply press the star key, followed by the number one on your telephone keypad. If you would like to withdraw your question, press star one a second time. Thank you.

Colette Kress: Starting with data center, data center revenue of 26.3 billion was a record, up 16% sequentially and up 154% year on year, given by strong demand for Nvidia Hopper, GPU computing, and our networking platforms. Compute revenue grew more than 2.5x; networking revenue grew more than 2x from the last year. Cloud service providers represented roughly 45% for data center revenue, and more than 50% stem from the consumer internet and enterprise companies. Customers continue to accelerate their hopper architecture purchases while gearing up to adopt Blackwell. Key workloads driving our data center growth include data to AI, model training and inferencing, video image and text data pre-imposed processing with CUDA and AI workloads, synthetic data generation, AI-powered recommender systems, SQL and vector database processing as well.

Speaker Change: Starting with data center, data center revenue of 26.3 billion, or they record up 16% sequentially and up 154% year-on-year, even by strong demand for Nvidia hopper, GPU computing, and their networking platforms.

Speaker Change: Compute Revenue, more than 2.5x.

Stewart Stecker: And Mr. Stewart Stecker, you may begin your conference. Thank you. Good afternoon, everyone. And welcome to Nvidia's conference call for the second quarter of fiscal 2025. With me today from Nvidia, our Jensen Wong, President and Chief Executive Officer, and Colette Cress, Executive Vice President and Chief Financial Officer. I would like to remind you that our call is being webcast live on Nvidia's investor relations website. The webcast will be available for replay until the conference call to discuss our financial results for the third quarter of fiscal 2025.

Speaker Change: Networking Revenue, Group More Than 2X from the last year.

Speaker Change: Bob and Service providers represented roughly 45% for data center revenue, and more than 50% stemmed from the consumer, internet, and enterprise companies.

Speaker Change: Customers can use to accelerate their hopper architecture purchases while gearing up to adopt Blackwell.

Speaker Change: Key Workloads, Driving our data center growth include, Deter to AI, Model Training and Infrincing.

Speaker Change: Video, Image, and Text data pre-improced processing with CUDA and AI workflows, synthetic data generation.

Stewart Stecker: The content of today's call is Nvidia's property. It cannot be reproduced or transcribed without prior written consent. During this call, we may make forward-looking statements based on current expectation. These are subject to a number of risks, significant risks, and uncertainties, and our actual results may differ materially. For discussion of factors that could affect our future financial results and business, please refer to the disclosure in today's earnings release. Our most recent forms, 10K and 10Q, and the reports that we may file on form 8K with the Securities and Exchange Commission.

Speaker Change: AI powered recommender systems, SQL and vector database processing as well.

Colette Kress: Next generation models will require 10 to 20 times more compute to train with significantly more data. The trend is expected to continue. Over the trillion four quarters, we estimate that inference drove more than 40% of our data center revenue. CSPs, consumer internet companies, and enterprises benefit from the incredible throughput and efficiency of NVIDIA's inference platform. Demand for NVIDIA is coming from frontier model makers, consumer internet services, and tens of thousands of companies and startups, building generative AI applications for consumers, advertising, education, enterprise, and healthcare and robotics. Developers desire NVIDIA's rich ecosystem and availability in every cloud.

Speaker Change: Next-generation models, full-require, tends to 20 times more compute to train with significantly more data.

Speaker Change: The trend is expected to continue.

Speaker Change: Over the training for quarters, we estimate that inference drove more than 40% of our data center revenue. CSPs, consumer internet companies, and enterprises benefit from the incredible throughput and efficiency of Nvidia's inference platform.

Stewart Stecker: All our statements are made as of today, August 28, 2024, based on information currently available to us. Except it's required by law, we assume no obligation to update any such statements. During this call, we will discuss non-GAAP financial measures. You can find a reconciliation of these non-GAAP financial measures to GAAP financial measures in our CFO commentary, which is posted on our website.

Speaker Change: Demand for Nvidia is coming from frontier model makers, consumer internet services, and 10,000 of companies and startups, building general to AI applications for consumers, advertising, education, enterprise and healthcare, and robotics.

Speaker Change: Developers, Desire, Nvidia's rich ecosystem and availability in every cloud.

Colette Kress: CSPs appreciate the broader adoption of NVIDIA and are growing through NVIDIA capacity given the high demand. NVIDIA H200 platform began ramping in Q2, chipping to large CSPs, consumer internet, and enterprise companies. NVIDIA H200 builds upon the strength of our helper architecture and offering over 40% more memory bandwidth compared to the H100. Our data center revenue in China grew sequentially in Q2 and is a significant contributor to our data center revenue. As a percentage of total data center revenue, it remains below levels seen prior to the imposition of export controls. We continue to expect the China market to be very competitive going forward.

Speaker Change: CSPs appreciate the broader adoption of Nvidia and are growing their Nvidia capacities given the high demand.

Stewart Stecker: Let me highlight an upcoming event for the financial community. We will be attending the Goldman Sachs, Communicopia, and Technology Conference on September 11 in San Francisco, where Jensen will participate in a keynote fireside chat.

Speaker Change: Nvidia, H-200, PopForm, BeGAM, RAMBING, NQ-2, Chipping to large CSPs, Consumer Internet and Enterprise Company.

Stewart Stecker: Our earnings call to discuss the results of our third quarter of fiscal 2025 is scheduled for Wednesday, November 20, 2024.

Speaker Change: The Nvidia H200 builds upon the strength of our helper architecture and offering over 40% more memory bandwidth compared to the H100.

Stewart Stecker: With that, let me turn the call over to you. Thanks, Stuart.

Speaker Change: Our data center revenue in China grew sequentially in Q2 and is significant contributor to our data center revenue.

Jensen Huang: Q2 was another record corner. Revenue of 30 billion was up 15% sequentially and up 122% year-on-year, and well above our outlook of 28 billion. Starting with data center, data center revenue of 26.3 billion was a record up 16% sequentially and up 154% year-on-year, during by strong demand for Nvidia Hopper, GPU computing, and our networking platforms. Compute revenue grew more than 2.5x, networking revenue grew more than 2x from the last year. AutoService providers represented roughly 45% for data center revenue, and more than 50% stemmed from the consumer, internet, and enterprise companies.

Speaker Change: As a percentage of total data center revenue, it remains below the level of seam prior to the export controls.

Speaker Change: Continues to expect the China market to be very competitive, going forward.

Colette Kress: The latest round of ML Perth, inference benchmarks, highlighted NVIDIA's inference leadership with both NVIDIA Hopper and Blackwell platform combining to win gold medals on all tasks.

Speaker Change: The latest round of ML Perf, Infrin Spendmark, highlighted Nvidia's Infrin leadership with both Nvidia, Hopper, and Blackwell.com, combining to win gold medals on all tasks.

Colette Kress: At Computex, NVIDIA, with the top computer manufacturers, unveiled an array of Blackwell architecture-powered systems and NVIDIA networking for building AI factories and data centers. With the NVIDIA MGX modular reference architecture, our OEMs and OEM partners are building more than 100 Blackwell-based systems designed quickly and cost-effectively. The NVIDIA Blackwell platform brings together multiple GPU, CPU, GPU and the link and link switch and the networking chips systems and NVIDIA CUDA software to power the next generation of AI across the cases industries and countries. The NVIDIA GB-200 NBL-72 system, with the fifth generation NBLINK, enables all 72 TPs to act as a single GPU and deliver up to 30 times faster inference for LLM's workloads and unlocking the ability to run trillion-parameter models in real time.

Speaker Change: At Competex, Nvidia with the top computer manufacturers, unveiled an array of Blackwell architecture-powered systems and Nvidia networking for building AI factories and data centers.

Speaker Change: With the Nvidia, MGX Modular Reference Architecture, our O-RAN and O-DM partners are building more than 100 Blackwell-based systems designed quickly and cost-effectively.

Jensen Huang: Customers continue to accelerate their hopper architecture purchases while gearing up to adopt Blackwell. Key workloads driving our data center growth include, Center to AI, model training and inferencing, video image and text data pre-imposed processing with CUDA and AI workloads, synthetic data generation, AI-powered recommender systems, SQL and vector database processing as well. Next generation models will require 10 to 20 times more compute to train with significantly more data. The trend is expected to continue.

Speaker Change: The Nvidia Blackwell Platform brings together multiple GPU CPUs, GPUs, and the link and the link switch and the networking chips systems and Nvidia CUBE software.

Speaker Change: to power the next generation of AI across the cases, industries and countries.

Speaker Change: The Nvidia GB-200 NBL-72 system with the fifth generation NBLink enables all 72GP used active single GPU and deliver up to 30 times faster inference.

Jensen Huang: Over the training for quarters, we estimate that inference drove more than 40% of our data center revenue. CSPs, consumer internet companies, and enterprises benefit from the incredible throughput and efficiency of NVIDIA's inference platform. Demand for NVIDIA is coming from frontier model makers, consumer internet services, and tens of thousands of companies and startups building generative AI applications for consumers, advertising, education, and a present healthcare and robotics. Developers desire NVIDIA's rich ecosystem and availability in every cloud.

Speaker Change: For L-L-M's workload and unlocking the ability to run trillion-paramher models in real time.

Colette Kress: How for demand is drawn and Blackwell is widely sampling. We executed a change to the Blackwell GPU mass to improve production yields. Blackwell production ramp is scheduled to begin in the fourth quarter and continue into fiscal year 26. In Q4, we expect to ship several billion dollars in Blackwell revenue. Proper shipments are expected to increase in the second half of this go to 2025. Proper supply and availability have improved. Demand for Blackwell platforms is well above supply, and we expect this to continue into next year. Networking revenue increased 16% sequentially. Our Ethernet for AI revenue, which includes our Spectrum XNN Ethernet platform, doubles sequentially, with hundreds of customers adopting our Ethernet offerings.

Speaker Change: CopperGaband is strong, and BlackRoll is widely sampling.

Speaker Change: We executed a change to the blackwell GPUs mask to improve production yields.

Speaker Change: Blackwell production ramp is scheduled to begin in the fourth quarter and continue in fiscal year 20.

Speaker Change: And GeForce, Re-Expectives, Bip, several billion dollars in Blackwell revenue. Opportunity are expected to increase in the second half of this goes 2025.

Jensen Huang: CSPs appreciate the broader adoption of NVIDIA and are growing their NVIDIA capacity given the high demand. NVIDIA H200 platform began ramping in Q2, chipping to large CSPs, consumer internet and enterprise companies. NVIDIA H200 builds upon the strength of our helper architecture and offering over 40% more memory bandwidth compared to the H100. Our data center revenue in China grew sequentially in Q2 and is significant contributor to our data center revenue. As a percentage of total data center revenue, it remains below levels seen prior to the imposition of export controls.

Speaker Change: Oppersupply and availability have improved. Demand for Blackwell platforms is well above supply. And we expect this to continue into next year.

Speaker Change: Networking Revenue Increased 16% Sequentialing.

Speaker Change: Our Ethernet for AI revenue, which includes our Spectrum X, and Ethernet platform, double sequentially with hundreds of customers adopting our Ethernet offerings.

Colette Kress: Spectrum X has broad market support from OEM and OEM partners and is being adopted by CSPs, GPU cloud providers, and enterprise, including XAI, to connect the largest GPU compute cluster in the world. Spectrum X supercharges Ethernet for AI processing and delivers 1.6x the performance after traditional Ethernet. We plan to launch new Spectrum X products every year to support demand for scaling compute clusters from tens of thousands of GPUs today to millions of GPUs in the near future. Spectrum X is well on track to begin a multi-billion dollar product line within a year.

Speaker Change: Spectrum X has broad market support from OEM and ODM partners, and is being adopted by CSPs, GPU Cloud providers and enterprise, including XAI to connect the largest GPU compute cluster in the world.

Speaker Change: Spectrum X Supercharges Ethernet for AI processing and delivers 1.6X the performance after digital Ethernet.

Jensen Huang: We continue to expect the China market to be very competitive going forward. The latest round of ML Perth inference benchmarks highlighted NVIDIA's inference leadership with both NVIDIA hopper and Blackwell platforms combining to win gold medals on all tasks. At Computex, NVIDIA with the top computer manufacturers unveiled an array of Blackwell architecture powered systems and NVIDIA networking for building AI factories and data centers. With the NVIDIA MGX modular reference architecture, our OEMs and OEM partners are building more than 100 Blackwell based systems designed quickly and cost effectively.

Speaker Change: We'd plan to launch new Spectrum X products every year to support demand for scaling compute clusters from tens of thousands of BPS today to millions of BPS in the near future.

Speaker Change: Spectrum X is well on track to begin a multi-billion dollar product line within a year.

Colette Kress: Our sovereign AI opportunities continue to expand as countries recognize AI expertise and infrastructure at national imperatives for the society and industries. The Pan National Institute of Advanced Industrial Science and Technology is building its AI bridging cloud infrastructure 3.0 super computer within video. We believe sovereign AI revenue will reach low double-digit billions this year.

Speaker Change: Ourself, our NAI opportunities continue to expand as countries recognize AI expertise and infrastructure at national imperatives for their society and industries.

Speaker Change: The Pan National Institute of Advanced Industrial Science and Technology is building its AI bridging cloud infrastructure 3.0 supercomputer within video.

Jensen Huang: The NVIDIA Blackwell platform brings together multiple GPU, CPU, DPU and the link and link switch and the networking chips systems and NVIDIA CUDA software to power the next generation of AI across the cases industries and countries. NVIDIA, GB200, NBL, 72 system, with the fifth generation NBLINK enables all 72TPUs to act as a single GPU and deliver up to 30 times faster inference for LLM's workloads and unlocking the ability to run trillion-parameter models in real time.

Speaker Change: We believe, sovereign AI revenue will reach the level of double digit billions this year.

Colette Kress: The enterprise AI wave has started. Enterprises also grow sequential revenue growth in the quarter. We are working with most of the Fortune 100 companies on AI initiatives across industries and geographies. A range of applications are fueling our growth, including AI powered chat box, generative AI co-pilot, and agents to build new, monetizable business applications and enhance employee productivity. And Docs is using NVIDIA generative AI for their smart agent, transforming the customer experience and reducing customer service cost by 30. Cursor. ServiceNow is using NVIDIA for its Now Assist offering, the fastest growing new product in the company's system.

Speaker Change: The Enterprise AI Waves is started, Enterprises also drove sequential revenue growth in the quarter. We are working with most of the Fortune 100 companies on AI initiatives across industries and geographies.

Speaker Change: A range of applications are futile, are growth, including AI powered chatbox, Gen. Today, I co-pilot, and agents to build new, launch-hisable business applications and enhanced employee productivity.

Speaker Change: And Docs is using Nvidia Generatives AI for third smart agent, Transform the customer experience and reducing customer service cost by 30%.

Jensen Huang: Copper demand is strong and Blackwell is widely sampling. We executed a change to the Blackwell GPU mass to improve production yield. Blackwell production ramp is scheduled to begin in the fourth quarter and continue into fiscal year 26. In Q4, we expect to ship several billion dollars in Blackwell revenue. Overshipments are expected to increase in the second half of fiscal 2025. Oversupply and availability have improved. Demand for Blackwell platforms is well above supply and we expect this to continue into next year.

Speaker Change: ServiceNow is using Nvidia for its now assist offering, the fastest growing new product in the company's history.

Colette Kress: SAP is using NVIDIA to build dual co-pilot. OECity is using NVIDIA to build a gender-to-day-I agent and lower gender-to-the-AI development cost. Snowflake serves over 3 billion queries a day for over 10,000 enterprise customers. Is working with NVIDIA to build co-pilot. And lastly, WisDrawn is using NVIDIA AI Omniverse to reduce end-to-end cycle times for their factories by 50%.

Speaker Change: I say P is using Nvidia to build dual co-powered.

Speaker Change: and C-C-C-C, is using Nvidia to build a genet today iAgent and lower genet to the AI development class.

Speaker Change: Snowflake serves over 3 billion queries a day, for over 10,000 enterprise customers, is working with Nvidia to build coal products.

Speaker Change: And lastly, WistRong is using InviniAI, Omniverse, to reduce end-end cycle times for their factories by 50%.

Jensen Huang: Networking revenue increased 16% sequentially. Our Ethernet for AI revenue, which includes our Spectrum XNN Ethernet platform, doubled sequentially with hundreds of customers adopting our Ethernet offerings. Spectrum X has broad market support from OEM and OEM partners and is being adopted by CSPs, GPU cloud providers and enterprise, including XAI to connect the largest GPU compute cluster in the world. Spectrum X supercharges Ethernet for AI processing and delivers 1.6X the performance after additional Ethernet.

Colette Kress: Automotive was a key growth driver for the quarter, as every automaker developing autonomous vehicle technology is using NVIDIA in their data centers. Automotive will drive multi-billion dollars in revenue across on-prem and cloud consumption and will grow as next-generation AV models require significantly more compute. Healthcare is also on its way to being a multi-billion dollar business as AI revolutionizes medical imaging, surgical robots, patient care, electronic health, record processing, and drug discovery. During the quarter, we announced a new in NVIDIA AI Foundry service to serve for charge gender-to-AI for the world's enterprises with Meta's Lama 3.1 collection of models.

Speaker Change: Automotive was a key growth driver for the quarter as every automaker developing autonomous vehicle technology is using a video in their data centers.

Speaker Change: Automotiful Drive, multi-billion dollars in revenue across on-prem and cloud consumption. And we'll grow as next-generation AV models require significantly more compute.

Speaker Change: Healthcare is also on its way to being a multi-million dollar business as AI revolutionizes medical imaging, surgical robot, patient care, electronic health, record processing, and drug discovery.

Jensen Huang: We plan to launch new Spectrum X products every year to support demand for scaling compute clusters from tens of thousands of GPUs today to millions of GPUs in the near future. Spectrum X is well on track to begin a multi-billion dollar product line within a year. Our sovereign AI opportunities continue to expand as countries recognize AI expertise and infrastructure at national imperatives for their society and industries. The Pan National Institute of Advanced Industrial Science and Technology is building its AI bridging cloud infrastructure 3.0 super computer within video.

Speaker Change: During the quarter, we announced a new Nvidia AI-Flap Foundry Service to separate, different charge, generating today AI for the world enterprises with Metas, Lama 3.1, Collection of Loms.

Colette Kress: This marks a watershed moment for enterprise AI. Companies, for the first time, can leverage the capabilities of an open source frontier level model to develop customized AI applications to encode their institutional knowledge into an AI flywheel to automate and accelerate their business. Accentures the first to adopt the new service to build custom Lama 3.1 models for both its own use and to assist clients seeking to deploy gender-to-AI applications. NVIDIA NIMS accelerate and simplify model deployment. Companies across healthcare, energy, financial services, retail, transportation, and telecommunications are adopting NIMS, including Aramco, Loves, and Uber. AT&T realized 70% cost savings and eight times latency reduction after moving into NIMS for gender-to-AI, call transcription, and classification.

Speaker Change: This marks a watershed moment for Enterprise AI.

Speaker Change: Companies for the first time can leverage the capabilities of an open source frontier-level model to develop customized AI applications to encode their institution knowledge into an AI flywheel to automate and accelerate their business.

Accenture: Accenture is the first to adopt the new service to build custom Lama-3 platform models for both its own use and to assist clients, seeking to deploy the new today I application.

Jensen Huang: We believe sovereign AI revenue will reach low double digit billions this year. The enterprise AI waves has started. Enterprises also grow sequential revenue growth in the quarter. We are working with most of the Fortune 100 companies on AI initiatives across industries and geographies. A range of applications are fueling our growth including AI powered chatbots, generative AI copilot and agents to build new monetizable business applications and enhance employee productivity. Andox is using Nvidia generative AI for their smart agent, transforming the customer experience and reducing customer service cost by 30.

Speaker Change: Nvidia, NIMs, Accelerate and Simplify Model Supplement, Companies across healthcare, Energy, Financial Services, Retail, Transportation, and Telecommunications are adopting NIMs, including Rampo, Lowe and Uber.

Speaker Change: AT&T, Realize 70% cost savings, and 8 times latency reduction after moving into NIMS for dinner today, I call transcription and classification.

Colette Kress: Over 150 partners are embedding NIMS across every layer of the AI ecosystem. We announce NIMS agent Blueprints, a catalog of customizable reference applications that include a full suite of software for building and deploying enterprise-generative AI applications. With NIMS agent Blueprints, enterprises can refine their AI applications over time, creating a data-driven AI flywheel. The first NIMS agent blueprints include workloads for customer service, computer-aided drug discovery, and enterprise retrieval-augmented generation. IBM. Our system integrators, technology solution providers, and system builders are bringing NVIDIA NIM agent blueprints to enterprises. NVIDIA NIM and NIM agent blueprints are available through the NVIDIA AI Enterprise software platform, which has great momentum.

Speaker Change: Over 150 partners are embedding men across every layer of the AI ecosystem.

Speaker Change: We announced NIMA-GIN's Lufrinch, a catalog of customizable reference applications that include a full suite of software for building and deploying enterprise-general-to-AI applications.

Jensen Huang: RSA, ServiceNow is using NVIDIA for its now assist offering, the fastest growing new product in the company's system. SAP is using NVIDIA to build dual co-pilot. OECD is using NVIDIA to build their Gen. A.I, agent and lower Gen. A.I, development cost. Snowflake serves over three billion queries a day for over 10,000 enterprise customers is working with NVIDIA to build co-pilot. And lastly, WisDron is using NVIDIA AI Omniverse to reduce end-to-end cycle times for their factories by 50%.

Speaker Change: With NIM, Aged BluePrench, Enterprises 10 Refining, their AI applications over time, creating a data-driven AI flywheel.

Speaker Change: The first name agent BluePrens, which includes workloads for customer service, computer-aided direct discovery, and enterprise retrieval augmented generation.

Speaker Change: Our system integrators, Technology Solutions, Fighters, and System Builders are bringing Nvidia and M, Agent BluePrinces to enterprises.

Jensen Huang: Automotive was a key growth driver for the quarter, as every automaker developing autonomous vehicle technology is using NVIDIA in their data centers. Automotive will drive multi-billion dollars in revenue across on-prem and cloud consumption and will grow as next-general generation AV models require significantly more compute. Health care is also on its way to being a multi-billion dollar business as AI revolutionizes medical imaging, surgical robots, patient care, electronic health record processing and drug discovery.

Speaker Change: Nvidia, NIM, and Ninja BlueField-3, are available through the Nvidia AI enterprise sample platform, which has great momentum.

Colette Kress: We expect our software, staff, and support revenue to quote a $2 billion annual run rate exiting this year, with NVIDIA AI Enterprise notably contributing to growth.

Speaker Change: We expect our software, SaaS, and support revenue to approach a $2 billion annual run rate exiting this year. Within video AI enterprise, notably, contribute to growth.

Colette Kress: Moving to gaming and AI PCs, gaming revenue of $2.88 billion increased 9% sequentially and 16% year-on-year. We sell sequencer growth in consoles, notebook and desktop revenue and demand a strong and growing and channel inventory remains healthy. Every PC with RTX is an AI PC. RTX PCs can deliver up to 1,300 AI TOPS and are now over 200 RTX AI laptops designed from leading PC manufacturers. With 600 AI powered applications and gains and an installed base of 100 million devices, RTX is set to revolutionize consumer experiences with generative AI. NVIDIA ACE, a suite of generative AI technologies, is available for RTX AI PCs.

Speaker Change: Moving to Gaming and AI PC's

Speaker Change: Gaming Revenue of 2.88 billion increased 9% sequentially and 16% year on year, which helps to ensure growth in consult, notebook and desktop revenue and demand is strong and growing and channel inventory remains healthy.

Jensen Huang: During the quarter we announced a new NVIDIA AI FLAB Foundry service to serve for charge Gen. A.I, for the world's enterprises with Meta's Lama 3.1 collection of models. This marks a watershed moment for enterprise AI. Companies for the first time can leverage the capabilities of an open source frontier level model to develop customized AI applications to encode their institutional knowledge into an AI flywheel to automate and accelerate their business. Accenture is the first to adopt the new service to build custom Lama 3.1 models for both its own use and to assist clients seeking to deploy Gen. A.I, applications.

Speaker Change: Every TC with RTX is an A-RTC.

Speaker Change: RTX, DCs can deliver up to 1,300, AI-TOPs, and are now over 200, RTX AI-LAP-TOPs designed from leading PC manufacturers.

Speaker Change: We 600 AI powered applications and games, and an installed base of 100 million devices, RTX is set to revolutionize consumer experiences within RTBI.

Speaker Change: Nvidia Ace, a suite of general-to-AI technologies, is available for RTX AI PCs. Macabreek is the first game to use Nvidia Ace, including our smart, small large.

Colette Kress: Mega Break is the first game to use NVIDIA ACE, including our small large, small language model, Minotron for the optimized on-device inference. The NVIDIA gaming ecosystem continues to grow. Recently added RTX in the OSS titles, including Indiana Jones and the Great Circle, David Awakening, and Dragon Age, the Vailguard.

Jensen Huang: NVIDIA NIMS accelerate and simplify model deployment. Companies across health care, energy, financial services, retail, transportation and telecommunications are adopting NIMS, including Aramco, Loves and Uber. AT&T realized 70% cost savings and eight times latency reduction after moving into NIMS for Gen. A.I, called transcription and classification. Over 150 partners are embedding NIMS across every layer of the AI ecosystem. We announced NIMS agent Blueprints, a catalog of customizable reference applications that include a full suite of software for building and deploying enterprise Gen. A.I, applications.

Speaker Change: Small Language Model, MiniTron, 4B, Optimized, ARM, Device Infrage

Speaker Change: The Nvidia Gaming Infi ecosystem continues to grow, recently added RTX and the OSS titles include Indiana Jones and the Great Circle, during awakening and Dragonade, the bail guard.

Colette Kress: The GeForce NOW library continues to expand with a total catalog size of over 2000 titles, the most content of any cloud gaming service.

Speaker Change: The GeForce now library continues to expand with total catalog sizeable with 2000 titles, the most content of any proud gaming service.

Colette Kress: Moving to Pro Visualization. Revenue of $454 million was up 6% sequentially and 20% year-on-year. Demand is being driven by AI and graphics use cases, including model finetuning and Omniverse related workloads. Automotive and manufacturing were among the key industry verticals driving growth this corner. Companies are racing to digital ties, workflows to drive efficiency across their operations. The world's largest electronics manufacturer, Foxconn, is using NVIDIA Omniverse to power digital twins of the physical plants that will produce NVIDIA black hole systems. And several large global enterprises, including Mercedes Benz, signed multi-year contracts for NVIDIA Omniverse Cloud to build industrial digital twins of factories.

Speaker Change: Weave into ProVizalization.

Jensen Huang: With NIMS agent Blueprints, enterprises can refine their AI applications over time, creating a data driven AI flywheel. The first NIMS agent Blueprints include workloads for customer service, computer-aided drug discovery, and enterprise retrieval, augmented generation, system. Our system integrators, technology solution providers, and system builders are bringing NVIDIA NIM agent blueprints to enterprises. NVIDIA NIM and NIM agent blueprints are available through the NVIDIA AI Enterprise software platform, which has great momentum. We expect our software, staff, and support revenue to approach a $2 billion annual run rate exiting this year, with NVIDIA AI Enterprise notably contributing to growth.

Speaker Change: Revenue of 454 million was up 6% sequentially, in 20% year on year. The demand is being driven by AI and graphic use cases, including model-finding and Omniverse-related workloads.

Speaker Change: Automotive and Manufacturing were among the key industries verticals driving growth this corner.

Speaker Change: Companies are racing to digital advice, workflows to drive efficiency across their operations.

Speaker Change: The World's largest electronic manufacturer Fox Hunt is using Nvidia omniverse to power digital twins of the physical plants that produce Nvidia blackmail systems.

Speaker Change: and several large global enterprises, including Mercedes-Benz, Science, Multi-Year Contracts for Nvidia Omniverse Cloud to build industrial digital twins of factories.

Colette Kress: We announced new NVIDIA USD NIMS and connectors to open Omniverse to new industries and enable developers to incorporate generative AI, co-pilot, and agents into USD and workloads, accelerating their ability to build highly accurate virtual world. World, WPP is implementing US-DNM microservices in its generative AI and able content creation pipeline for customers, such as the Coca-Cola Company.

Speaker Change: We announced a little Nvidia, USDNM and connectors to open Omniverse.

Speaker Change: to new industries and enable developers.

Speaker Change: to incorporate generative AI, Colette pilots, and agents into USD's workflows, accelerating their ability to build highly accurate virtual worlds.

Jensen Huang: Moving to gaming and AI PCs, gaming revenue of $2.88 billion increased 9% sequentially and 16% year-on-year, who saw sequential growth in consoles, notebook, industrial revenue, and demand a strong and growing and channel inventory remain healthy. Every PC with RTX is an AI PC. RTX PCs can deliver up to 1,300 AI tops, and are now over 200 RTX AI laptops designed from leading PC manufacturers. With 600 AI-powered applications and gains and an installed base of 100 million devices, RTX is set to revolutionize consumer experiences with generative AI.

Speaker Change: WPP is implementing U-SBNAM micro-services in its general today I enabled content creation pipeline for customers, such as the Coca-Cola company.

Colette Kress: Moving to Automotive and Robotics, revenue was 346 million, up 5% sequentially, and up 37% year-on-year. Year-on-year growth was driven by the new customer ramps in self-driving platforms and increased demand for AI conflict solutions. At the Consumer, at the Computer Vision and Pattern Recognition Conference, NVIDIA won the Autonomous Grand Challenge in the end-end writing at scale category, outperforming more than 400 entries worldwide. Austin Dynamics, BYD, Electronics, Figure Intrinsic, Seamons, Skills, AI, and Parodine Robotics are using the NVIDIA Isaac Robotics platform for autonomous robots, arms, humanoid, and mobile robots.

Speaker Change: [inaudible]

Speaker Change: Revenue was 346 million, up 5% sequentially, and up 37% year-on-year. Year-on-year growth was driven by the new customer ramps in self-driving platforms, and increased demand for AI-configure solutions.

Speaker Change: As the computer vision and pattern recognition conference, Nvidia won the autonomous grant talent in the end-end writing at scale category, outperforming more than 400 entries worldwide.

Jensen Huang: NVIDIA ACE, a suite of generative AI technologies, is available for RTX AI PCs. Mega Break is the first game to use NVIDIA ACE, including our small-large, small-language model, Minitron 4B, optimized on device inference. The NVIDIA gaming ecosystem continues to grow. Recently added, RTX and the OSS titles, including Indiana Jones and the Great Circle, due to Awakening and Dragon Age, the Veil Guard. The GeForce now library continues to expand, with total catalog size of over 2000 titles, the most content of any cloud gaming service.

Speaker Change: Austin Dynamics, B-Y-D Electronics, Figure, Infinsics, Seenance.

Speaker Change: Skills, A-I, and Terradise Robotics are using the Nvidia, Isaac Robotics platform, or Autonomous Robotics, ARMs, Humanoids, and Mobile Robotics.

Colette Kress: Now moving to the rest of the panel. Gapgrove margins were 75.1%, and non-Gapgrove margins were 75.7%. Down sequentially, due to a higher mix of new products within data center and inventory provisions for low yielding black-well material. Sequentially, Gap and non-Gap operating expenses were up 12%, primarily reflecting higher compensation-related costs. Cash flow from operations was 14.5 billion. In Q2, we utilized cash of 7.4 billion towards shareholder returns in the form of share repurchases and cash dividends, reflecting the increase in dividend per share. Our Board of Directors recently approved a $50 billion share repurchase authorization to add to our remaining $7.5 billion of authorization at the end of Q2.

Speaker Change: Moving to the rest of the panel.

Speaker Change: GAAP Gross margins were 75.1% and non-GAAP Gross margins were 75.7%. Galaxyquenchily due to expire mix of new products within data center and inventory provisions for low yielding Blackwell material.

Speaker Change: sequentially, GAAP and non-GAAP operating expenses were up to 12% primarily reflecting higher compensation related costs.

Jensen Huang: Moving to Pro Visualization. Revenue of 454 million was up 6% sequentially and 20% year-on-year. Demand is being driven by AI and graphics use cases, including model finetuning and omniverse related workloads. Automotive and manufacturing were among the key industry verticals driving growth disorder. Companies are racing to digital ties, workflows to drive efficiency across their operations. The world's largest electronics manufacturer, Foxconn, is using NVIDIA omniverse to power digital twins of the physical plants that produce NVIDIA black hole systems.

Speaker Change: Cash flow from operations for 14.5 billion.

Speaker Change: In Q2, we utilize cash of 7.4 billion to our shareholder returns in the form of share-reportances and cash dividend, reflecting the increase in dividend per share.

Speaker Change: Our Board of Directors recently approved a $50 billion share repurchase authorization to add to our remaining $1.5 billion of authorization at the end of Q2.

Colette Kress: Let me turn the outlook for the third quarter. Total revenue is expected to be 32.5 billion plus or minus 2%. Our third quarter revenue outlook incorporates continued growth of our helper architecture and sampling of our black-well products.

Jensen Huang: And several large global enterprises, including Mercedes Benz, signed multi-year contracts for NVIDIA omniverse cloud to build industrial digital twins of factories. We announced new NVIDIA USD NIMS and connectors to open omniverse to new industries and enable developers to incorporate genitive AI, co-pilot and agents into USD workloads, accelerating their ability to build highly accurate virtual world. World, WPP is implementing USD&M microservices in its generative AI and able content creation pipeline for customers, such as the Coca-Cola Company.

Speaker Change: Let me turn the outlook for the third corner.

Speaker Change: Total revenue is expected to be 32.5 billion, plus your minus 2%. Our third quarter revenue outlook incorporates continued growth of our hopper architectures and sampling of our Blackwell products.

Colette Kress: We expect Black-Well production ramp in Q4. Gap and non-Gapgrove margins are expected to be 74.4% and 75%, respectively, plus or minus 50 basis points. As our data center mix continued the shift to new products, we expect this trend to continue into the fourth quarter of fiscal 2025.

Speaker Change: We expect blackwell production ramp in GeForce.

Speaker Change: GAAP and non-GAAP growth margins are expected to be 74.4% and 75% respectively, plus or minus 50 basis points.

Speaker Change: As our data center next continue, the Shirt's new products.

Speaker Change: We expect this trend to continue into the fourth quarter of fiscal 2018-25.

Colette Kress: For the full year, we expect gross margins to be in the mid-70% range. Gap and non-GAAP operating are expected to be approximately 4.3 billion and 3.0 billion, respectively.

Speaker Change: For the full year, we expect to roast margins to be in the mid-70% range.

Speaker Change: Gap and non-gap operating expenses are expected to be approximately 4.3 billion and 3.0 billion, respectively.

Jensen Huang: Moving to Automotive and Robotics, Revenue was 346 million, up 5% sequentially, and up 37% year-on-year. Year-on-year growth was driven by the new customer ramps in self-driving platforms and increased demand for AI conflict solutions. At the consumer, at the computer vision and pattern recognition conference, NVIDIA won the Autonomous Grand Challenge in the end-end writing at scale category, outperforming more than 400 entries worldwide. Austin Dynamics, BYD, Electronics, Figure Intrindic, Seenance, Skills, AI, and Parodyne Robotics are using the NVIDIA Isaac Robotics platform for Autonomous Robot, ARM, Humanoid, and Mobile Robot. Now moving to the rest of the panel.

Colette Kress: Full year operating expenses are expected to grow in the mid to upper 40% range as we work on developing our next generation of products. Gap and non-Gap other income and expenses are expected to be about 350 million, including gains and losses from non-affiliated investments and publicly held equity security.

Speaker Change: Full-year operating expenses are expected to grow in the mid to upper 40% range as we work on developing our next generation of products.

Speaker Change: GAAP and non-GAAP other income and expenses are expected to be about 350 million, including gains and losses from non-affiliated investments and publicly held equity securities.

Colette Kress: GAAP and NGAP tax rates are expected to be 17%, plus or minus 1%, excluding any discrete items. Further financial details are included in the Seattle commentary and other information available on our IR website.

Speaker Change: GAAP and non-GAAP tax rates are expected to be 17%, plus or minus 1%, excluding any discrete items.

Speaker Change: Further financial detail, are included in the Seeable Commentary and other information available on our IR website.

Operator: We are now going to open the call for questions.

Operator: Operator, would you please help us in both the questions? Thank you. And at this time, I would like to remind everyone: in order to ask a question, press star and then the number one on your telephone keypad.

Speaker Change: We are now going to open the call for questions. Operator, would you please help us and pull for questions?

Speaker Change: Thank you.

Speaker Change: And at this time, I would like to remind everyone in order to ask a question, press star in the number one on your telephone keypad.

Operator: We will pause for just a moment to compile the Q&A roster, and as a reminder, we ask that you please limit yourself to one question.

Colette Kress: Gapgrove margins were 75.1%, and non-Gapgrove margins were 75.7%. Down sequentially, due to a higher mix of new products within data center and inventory provisions for low yielding black-well material. Sequentially, Gap and non-Gap operating expenses were up 12%, primarily reflecting higher compensation related costs. Cash flow from operations was 14.5 billion. In Q2, we utilized cash of 7.4 billion towards shareholder returns in the form of share repurchases and cash dividends, reflecting the increase in dividend per share. Our Board of Directors recently approved a $50 billion share repurchase authorization to add to our remaining 7.5 billion of authorization at the end of Q2.

Speaker Change: We will pause for just a moment to compile the Q&A roster, and as a reminder, we ask that you please limit yourself to one question.

Vivek Arya: And your first question comes from the line of Vivek Arya with Bank of America Securities. Your line is open. Thanks for taking my question. Jensen, you mentioned in the prepared comments that there's a change in the Blackwell GPU mask. I'm curious, are there any other incremental changes in backend packaging or anything else? And I think related, you suggested that you could ship several billion dollars of Blackwell in Q4 despite the change in the design. Is it because all these issues are being solved?

Speaker Change: And your first question comes from the line of Vivek Aria with Bank of America Securities. Your line is open.

Vivek Aria: Thanks for taking my question. Jensen, you mentioned in the prepared comments that there's a change in the Black Web, GPMorgan. I'm curious, are there any other incremental changes in back-end packaging or anything else?

Speaker Change: And I think related, you suggested that you could ship several billion dollars of blackware in Q4, despite the change in the design. Is it because all these issues will be solved? I then just help us size.

Jensen Huang: I then just help us decide what is the overall impact of any changes in Blackwell timing, what that means to your revenue profile and how our customers are reacting to it. Yeah, thanks, Vivek. The change to the mask is complete. There were no functional changes necessary. And so we're sampling functional samples of Blackwell, Grace Blackwell, in a variety of system configurations as we speak. There are something like 100 different types of Blackwell-based systems that are built that were shown at Computex. And we're enabling our ecosystem to start sampling those. The functionality of Blackwell is as it is, and we expect to start production in Q4.

Speaker Change: What is the overall impact of any changes in BlackSale timing? What that means, C-LOC and a revenue profile, and how our customers reacting to it?

Colette Kress: Let me turn the outlook for the third quarter. Total revenue is expected to be 32.5 billion plus or minus 2%, our third quarter revenue outlook incorporates continued growth of our helper architecture and sampling of our black-well products. We expect black-well production ramp in Q4. Gap and non-Gapgrove margins are expected to be 74.4% and 75% respectively plus or minus 50 basis points. As our data center mix continued the shift to new products, we expect this trend to continue into the fourth quarter of fiscal 2025.

Speaker Change: Yeah, thanks and back. The change to the mask is complete. There were no functional changes necessary.

Speaker Change: and so we're sampling functional samples of Blackwell, Grace Blackwell in a variety of system configurations as we speak.

Speaker Change: There are something like 100 different types of black-well-based systems that are built that are shown at ComputeX.

Speaker Change: and we're enabling our ecosystem to start sampling those.

Speaker Change: The functionality of Blackwell is as it is, and we expect to start production in Q4.

Colette Kress: For the full year, we expect gross margins to be in the mid-70% range. Gap and non-Gap operating expenses are expected to be approximately 4.3 billion and 3.0 billion respectively. Full year operating expenses are expected to grow in the mid to upper 40% range as we work on developing our next generation of products. Gap and non-Gap other income and expenses are expected to be about 350 million, including gains and losses from non-affiliated investments and publicly held equity security, and non-GAAP tax rates are expected to be 17%, plus or minus 1%, excluding any discrete items. Further financial details are included in the Seattle Commentary and other information available on our IR website.

Toshiya Hari: And your next question comes from the line of Toshia Hari with Goldman Sachs. Your line is open. Hi, thank you so much for taking the question. Jensen, I had a relatively longer term question. As you may know, there's a pretty heated debate in the market on your customers and customers' customers' return on investment and what that means for the sustainability of CapEx going forward. Internally, at Nvidia, what are you guys watching? What's on your dashboard as you try to gauge customer return and how that impacts CapEx? Is that a quick follow-up, maybe for Collette? I think a couple of billion.

Speaker Change: And your next question, come to the line of Toshiya Hari with Goldman Sachs. Your line is open.

Toshiya Hari: Hi, thank you so much for taking the question. Jensen, I had a relatively longer term question. As you may know, there's a pretty heated debate in the market.

Toshiya Hari: On, you know, your customers and customers, customers, return on investment.

Speaker Change: and what that means for the sustainability of CapEx going forward. Internally at Nvidia, what are you guys watching? What's on your dashboard as you try to gauge customer return?

Speaker Change: and how that impact CapEx. That a quick follow-up may be for Colette. I think you're a sovereign AI number for the fully-year went up maybe a couple of billion. What's driving the improved outlook and how should we think about fiscal 26th? Thank you.

Toshiya Hari: What's driving the improved outlook and how should we think about fiscal 26? Thank you. Thanks, Toshia.

Jensen Huang: First of all, when I said ship production in Q4, I mean shipping out. I don't mean starting to ship, but I mean I don't mean starting production by shipping out.

Operator: We are now going to open the call for questions, operator, would you please help us and pull for questions? Thank you. And at this time, I would like to remind everyone in order to ask a question, press star and then the number one on your telephone keypad.

Speaker Change: Thanks, Toshia. First of all, when I said ship production in Q4, I mean, shipping out. I don't mean starting to ship, but I mean, I don't mean starting production by shipping out.

Operator: We will pause for just a moment to compile the Q&A roster, and as a reminder, we ask that you please limit yourself to one question.

Jensen Huang: On the longer term question, let's take a step back and you've heard me say that we're going through two simultaneous platform transitions at the same time. The first one is transitioning from general purpose computing to accelerated computing, and the reason for that is because CPU scaling has been known to be slowing for some time, and it is slow to a crawl. And yet, the amount of computing demand continues to grow quite significantly. You could maybe even estimate it to be doubling every single year. And so if we don't have a new approach, computing inflation would be driving up the cost for every company, and it would be driving up the energy consumption of data centers around the world.

Speaker Change: on the long term question. Let's take a step back. And you've heard me say that we're going through two simultaneous platform transitions at the same time.

Speaker Change: The first one is transitioning from accelerated computing to general purpose computing to accelerated computing.

Vivek Arya: And your first question comes from the line of Vivek Arya with Bank of America Securities. Your line is open. Thanks for taking my question. Jensen, you mentioned in the prepared comments that there's a change in the Blackwell GPU mask. I'm curious, are there any other incremental changes in backend packaging or anything else? And I think related, you suggested that you could ship several billion dollars of Blackwell in Q4 despite the change in the design.

Speaker Change: And the reason for that is because CPU scaling has been known to be slowing for some time, and it is slow to a crawl.

Speaker Change: And yet the amount of computing demand continues to grow quite significantly, you could, maybe even estimated to be doubling every single year. And so if we don't have a new approach...

Speaker Change: Computing inflation would be driving up the cost for every company, and it would be driving up the energy consumption of data centers around the world. In fact, you're seeing that.

Jensen Huang: In fact, you're seeing that. And so the answer is accelerated computing. We know that accelerated computing, of course, speeds up applications. It also enables you to do computing at a much larger scale, for example, scientific simulations or database processing. But what that translates directly to is lower cost and lower energy consumed. And in fact, this week, there's a blog that came out that talked about a whole bunch of new libraries that we offer. And that's really the core of the first platform transition going from general purpose computing to accelerated computing. And it's not unusual to see someone save 90% of their computing cost.

Vivek Arya: Is it because all these issues are being solved? I then just help us decide what is the overall impact of any changes in Blackwell timing, what that means to your revenue profile, and how our customers are reacting to it. Thanks, Vivek.

Speaker Change: And so the answer is accelerated computing. We know that accelerated computing, of course.

Speaker Change: Speeds of Applications.

Speaker Change: It also enables you to do computing at a much larger scale, for example, scientific simulations or database processing. But what that translates directly to is lower cost and lower energy consumed.

Jensen Huang: The change to the mask is complete. There were no functional changes necessary. And so we're sampling functional samples of Blackwell, Grace Blackwell, in a variety of system configurations, as we speak. There are something like 100 different types of Blackwell-based systems that are built that were shown at Computext, and we're enabling our ecosystem to start sampling those. The functionality of Blackwell is as it is, and we expect to start production in Q4.

Speaker Change: And in fact, this week, there's a blog that came out that talked about a whole bunch of new libraries that we offer. And that's really the core of the first platform transition going from general purpose computing to accelerate computing.

Speaker Change: And it's not unusual to see.

Speaker Change: O-RAN,

Speaker Change: For someone save 90% of the computing cost.

Jensen Huang: And the reason for that is, of course, you just sped up an application 50x. You would expect the computing cost to decline quite significantly. The second was enabled by accelerated computing because because we drove down the cost of training, large language models are training deep learning so incredibly that it is now possible to have gigantic scale models, multi trillion parameter models and train it on pre train it on just about the world's knowledge corpus. And let the model go figure out how to understand human language representation and how to codify knowledge into its neural networks and how to learn reasoning.

Speaker Change: and the reason for that is of course you just set up an application, 50X.

Toshiya Hari: And your next question comes from the line of Toshia Hari with Goldman Sachs. Your line is open. Hi, thank you so much for taking the question. Jensen, I had a relatively longer term question. As you may know, there's a pretty heated debate in the market on your customers and customers' customers return on investment, and what that means for the sustainability of CapEx going forward. Internally, at Nvidia, what are you guys watching?

Speaker Change: You would expect the computing cost to decline quite significantly. The second was enabled by Excel-rated computing because we drove down the cost of training, large language models, training, deep learning, so incredibly. That is now possible.

Toshiya Hari: What's on your dashboard as you try to gauge customer return and how that impacts CapEx? Is that a quick follow-up maybe for Collette? I think your sovereign AI number for the full year went up, maybe a couple of billion. What's driving the improved outlook, and how should we think about fiscal 26? Thank you. Thanks, Toshia. First of all, when I said ship production in Q4, I mean shipping out. I don't mean starting to ship, but I mean I don't mean starting production by shipping out.

Speaker Change: to have gigantic scale models, multi- trillion parameter models, and train it on pre-training, on just about the world's knowledge corpus.

Speaker Change: And let the model go figure out how to understand human representation, human language representation, and how to...

Jensen Huang: And so, which caused the generative AI revolution?

Speaker Change: codify knowledge into its neural networks and how to learn reasoning.

Speaker Change: and so which caused the generative AI revolution? Now, generative AI...

Jensen Huang: Now, generative AI, taking a step back about why it is that we went so deeply into it, is because it's not just a feature, it's not just the capability, it's a fundamental new way of doing software. Instead of human engineered algorithms, we now have data. We tell the AI, we tell the model, we tell the computer what the expected answers are, whatever, what our previous observations. And then for it to figure out what the algorithm is, what the function, it learns a universal, you know, AI is a bit of a universal function approximator and it learns the function.

Speaker Change: Taking a step back about why it is that we went so deeply into it, is because it's not just a feature, it's not just the capability, it's a fundamental new way of doing software.

Jensen Huang: On the longer term question, let's take a step back. And you've heard me say that we're going through two simultaneous platform transitions at the same time. R&M, the first one is transitioning from accelerated computing to, from general purpose computing to accelerated computing. And the reason for that is because CPU scaling has been known to be slowing for some time and it is, it is slow to a crawl. And yet the amount of computing demand continues to grow quite significantly, you could maybe even estimate it to be doubling every single year.

Speaker Change: Instead of human-engineered algorithms, we now have data, we tell the AI, we tell the model, we tell the computer, what are the expected answers, whatever are our previous observations.

Speaker Change: and then for it to figure out what the algorithm is, what's the function. It learns a universal, you know, AI is a bit of a universal function approximator and it learns the function.

Jensen Huang: And so you could learn the function of almost anything, you know, and anything that you have that's predictable, anything that has structure, anything that you have previous examples of.

Speaker Change: And so you can learn the function of almost anything.

Speaker Change: Anything that's predictable, anything that has structure or anything that you have previous examples of.

Jensen Huang: And so if we don't have a new approach, computing inflation would be driving up the cost for every company and it would be driving up the energy consumption of data centers around the world. In fact, you're seeing that. And so the answer is accelerated computing. We know that accelerated computing, of course, speeds of applications. It also enables you to do computing at a much larger scale, for example, scientific simulations or database processing.

Jensen Huang: And so now here we are with generative AI; it's a fundamental new form of computer science. It's affecting how every layer of computing is done, from CPU to GPU, from human-engineered algorithms to machine-learned algorithms. And the type of applications you could now develop and produce is fundamentally remarkable.

Speaker Change: And so now here we are with General Todayi, it's a fundamental new form of computer science, it's affecting how every layer of computing is done from CPU to GPU.

Speaker Change: From Human Engineering Algorithms to Machine Learn Algorithms, and the type of applications you could now develop and produce is fundamentally remarkable. And there are several things that are happening in generative AI.

Jensen Huang: And there are several things that are happening in generative AI. So the first thing that's happening is the frontier models are growing in quite substantial scale and they're still seeing, we're still all seeing the benefits of scaling and whenever you double the size of a model, you also have more than double the size of the data set that's going to train it and so the amount of flops necessary in order to create that model goes up quadratically and so it's not unexpected to see that the next generation models could take 10, 20, 40 times more compute than the last generation so we have to continue to drive the generational performance up quite significantly so we can drive down the energy consumed and drive down the cost necessary to do it. So the first one is there are larger frontier models trained on more modalities and surprisingly there are more frontier model makers than last year and so you have more on more on more that's one of the dynamics going on in generative AI. At the second is although it's below the tip of the iceberg, you know what we see our chat GPT image generators we see coding we use we use a generative AI for coding quite extensively here at NVIDIA now we of course have a lot of digital designers and things like that but those are kind of the tip of the iceberg what's below the iceberg are the largest systems largest computing systems in the world today which are and you've heard me talk about this in the past which are recommender systems moving from CPUs it's now moving from CPUs to generative AI so recommender systems add generation custom add generation targeting ads at very large scale and quite hyper targeting search and user generated content these are all very large scale applications have now evolved to generative AI of course the number of generative AI startups is generating tens of billions of dollars of cloud renting opportunities for our cloud partners and sovereign AI you know countries that are now realizing that their data is their natural and national resource and they have to use they have to use AI build their own AI infrastructure so that they could have their own digital intelligence enterprise AI as Colette mentioned earlier is starting and you might have seen our announcement that the world's leading IT companies are joining us to take the NVIDIA enterprise platform to the world's enterprises that the company companies that we're talking to so many of them are just so incredibly excited to drive more productivity out of their company and then general robotics the big the big transformation last year as we are able to now learn physical AI from watching video and human demonstration and synthetic data generation from reinforcement learning from systems like omniverse we're now able to work with just about every robotics companies now to start thinking about start building general robotics and so you can see that there are just so many different directions that generative AI is going and so we're actually seeing the momentum of general generative AI accelerating Intescia, to answer your question regarding sovereign AI and our goals in terms of growth and terms of revenue.

Jensen Huang: But what that translates directly to is lower cost and lower energy consumed. And in fact, this week, there's a blog that came out that talked about a whole bunch of new libraries that we offer. And that's really the core of the first platform transition going from general purpose computing to accelerated computing. And it's not unusual to see someone save 90% of their computing cost. And the reason for that is, of course, you just sped up an application 50x.

Speaker Change: So the first thing that's happening is the frontier models are growing in quite substantial scale.

Speaker Change: And they're still seeing, we're still all seeing, the benefits of scaling.

Speaker Change: Whenever you double the size of a model, you also have to more than double the size of the data set that's going to train it. And so the amount of flops.

Speaker Change: Necessary, in order to create that model.

Speaker Change: Go's up quadratically, and...

Speaker Change: And so, it's not unexpected to see that the next generation models could take 20, 10, 20, 40 times more compute than last generation, so we have to continue to drive.

Jensen Huang: You would expect the computing cost to decline quite significantly. The second was enabled by accelerated computing because because we drove down the cost of training, large language models of training deep learning so incredibly that it is now possible to have gigantic scale models, multi trillion parameter models and train it on pre train it on just about the world's knowledge corpus. And let the model go figure out how to understand human represent human language representation and how to codify knowledge into its neural networks and how to learn reasoning.

Speaker Change: the generational performance up quite significantly so we can drive down the energy consumed and drive down the cost necessary to do it. So the first one is there are larger frontier models.

Speaker Change: Trained On More Modalities?

Speaker Change: and surprisingly, they're more frontier model makers.

Speaker Change: Then last year, and so you have more on more, that's one of the dynamics going on in a journey to the AI DeSecond.

Speaker Change: It's all the way below the tip of the iceberg, you know what we see.

Speaker Change: ARM, ChatGPT, Omniverse,

Speaker Change: Image Generators, Image Generators,

Speaker Change: We see coding, we use, we use, during the AI for coding quite extensively here at Nvidia now. We of course have a lot of digital designers and things like that. But those are kind of the tip of the iceberg. What's below the iceberg? Are the largest systems?

Jensen Huang: And so which caused the generative AI revolution. Now, generative AI taking a step back about why it is that we went so deeply into it is because it's not just a feature, it's not just the capability, it's a fundamental new way of doing software. Instead of human engineered algorithms, we now have data, we tell the AI, we tell the model, we tell the computer, what are the expected answers, whatever, what are our previous observations.

Speaker Change: Largest computing systems in a world today, which are, and you've heard me talk about this in the past, which are recommender systems moving from CPUs, it's now moving from CPUs to generative AI.

Speaker Change: So, recommender systems, ad-generation, custom ad-generation targeting ads at very large-scale and quite hyper-targeting search and user-generated content. These are all very large-scale applications have now evolved to generate to the AI.

Jensen Huang: And then for it to figure out what the algorithm is, what's the function, it learns a universal, you know, AI is a bit of a universal function approximator and it learns the function. And so you could learn the function of almost anything, you know, and anything that you have that's predictable, anything that has structure, anything that you have previous examples of.

Speaker Change: Of course, the number of genitive AI startups is generating tens of billions of dollars of cloud renting opportunities for our cloud partners.

Speaker Change: and sovereign AI, you know, countries that are now realizing that their data is their natural and national resource and they have to use AI build their own AI infrastructure so that they could have their own digital intelligence.

Jensen Huang: And so now here we are with generative AI, it's a fundamental new form of computer science, it's affecting how every layer of computing is done from CPU to GPU, from human engineered algorithms to machine learn algorithms. And the type of applications you could now develop and produce is fundamentally remarkable. And there are several things that are happening in generative AI. So the first thing that's happening is the frontier models are growing in quite substantial scale, and they're still seeing, we're still all seeing the benefits of scaling, and whenever you double the size of a model, you also have more than double the size of the data set to go train it.

Colette: EnterPrice AI as Colette mentioned earlier is starting.

Speaker Change: and you might have seen our announcement.

Speaker Change: that the world's leading fight team.

Speaker Change: Companies are joining us to take the Nvidia enterprise platform to the world's enterprises. The companies that we're talking to, so many of them are just so incredibly excited to drive more productivity out of their company. And then general robotics.

Jensen Huang: And so the amount of flops necessary in order to create that model goes up quadratically. So we're not unexpected to see that the next generation models could take 10, 20, 40 times more compute than last generation. So we have to continue to drive the generational performance up quite significantly so we can drive down the energy consumed and drive down the cost necessary to do it. So the first one is there are larger frontier models trained on more modalities, and surprisingly there are more frontier model makers than last year.

Speaker Change: [inaudible]

Speaker Change: Transformation last year as we are able to now learn.

Speaker Change: Physical AI from watching video and human demonstration and synthetic data generation.

Speaker Change: Reinforcement Learning from System like Omniverse.

Speaker Change: We are now able to...

Speaker Change: Work, which is about every robotics company's now to start thinking about start building.

Speaker Change: General Robotics, so you can see that there are so many different directions that General to the AI is going, and so we're actually seeing the momentum of General to the AI accelerating.

Speaker Change: And to see it to answer your question regarding us sovereign AI and our goals in terms of growth in terms of revenue.

Jensen Huang: And so you have more on more on more. That's one of the dynamics going on in generative AI. And the second is although it's below the tip of the iceberg, you know what we see our chat UPT image generators, we see coding, we use, we use a generative AI for coding quite extensively here at NVIDIA now, we of course have a lot of digital designers and things like that. But those are kind of the tip of the iceberg.

Jensen Huang: It certainly is a unique and growing opportunity, something that surfaced with genotype AI and the desires of countries around the world to have their own genotype AI that would be able to incorporate their own language, incorporate their own culture, incorporate their own data in that country. So more and more excitement around these models and what they can be specific for those countries. So yes, we are seeing some growth opportunity in front of us.

Speaker Change: Certainly, is a unique and growing opportunity, something that surfaced with Genota VI and the desires of countries around the world to have their own Genota VI that would be able to incorporate their own language, incorporate their own culture, incorporate their own data in that country.

Speaker Change: So, more and more excitement around these models and what they can be specific for those countries. So, yes, we are seeing some growth opportunity in front of us.

Jensen Huang: What's below the iceberg are the largest systems, largest computing systems in the world today, which are, and you've heard me talk about this in the past, which are recommender systems moving from CPUs, it's now moving from CPUs to generative AI. So recommender systems, add generation, custom add generation targeting ads at very, very large scale and quite hyper targeting search and user generated content. These are all very large scale applications have now evolved to generative AI.

Joe Moore: And your next question comes from the line of Joe Moore with Morgan Stanley. Your line is open. Great. Thank you. Thanks to the impressionally, she talked about Blackwell anticipation being incredible, but it seems like Hopper demand is also really strong. I mean, you're guiding for a very strong quarter without Blackwell and October. So, you know, how long do you see sort of coexisting strong demand for both? And can you talk about the transition to Blackwell? Do you see people intermixing clusters? Do you think most of the Blackwell activity's new cluster is just some sense of what that transition looks like?

Speaker Change: And your next question comes from the line of Joe Moore with Morgan Stanley, your line is open.

Joe Moore: Thank you. Jensen, in the press release, she talked about Blackwell anticipation.

Speaker Change: Being incredible.

Joe Moore: But it seems like copper demand is also really strong, I mean, they're guiding for a very strong quarter without blackwell and October. So, you know, how long do you see sort of coexisting strong demand for both? And can you talk about the transition to blackwell, do you see people, intermixing clusters, do you think most of the blackwell activities, new clusters, just some sense of what that transition looks like?

Jensen Huang: Of course, the number of generative AI startups is generating tens of billions of dollars of cloud renting opportunities for our cloud partners and sovereign AI, you know, countries that are now realizing that their data is their natural and national resource, and they have to use, they have to use AI build their own AI infrastructure so that they could have their own digital intelligence. Enterprise AI, as Colette mentioned earlier, is starting and you might have seen our announcement that the world's leading IT companies are joining us to take the NVIDIA enterprise platform to the world's enterprises, that the companies that we're talking to so many of them are just so incredibly excited to drive more productivity out of their company.

Jensen Huang: Yeah, thanks, Joe. The demand for Hopper is really strong, and it's true. The demand for Blackwell is incredible. There's a couple of reasons for that. The first reason is if you just look at the world's cloud service providers, the amount of GPU capacity they have available, it's basically none. And the reason for that is because they're either being deployed internally for accelerating their own workloads, data processing, for example. Data processing; we hardly ever talk about it because it's mundane. It's not very cool because it doesn't generate a picture or generate words, but almost every single company in the world processes data in the background.

Speaker Change: Yeah, thanks so. The demand for hoppers really strong and it's true that demand for blackwell is incredible.

Speaker Change: Bye!

Speaker Change: There's a couple of reasons for that.

Speaker Change: The first reason is if you just look at the world's cloud service providers in the amount of GPU capacity to have available, it's basically none.

Speaker Change: And the reason for that is because they're either being deployed internally for accelerating their own workloads, data processing, for example.

Speaker Change: Data Processing, you know, we hurriedly ever talked about it because it's mundane.

Speaker Change: It's not very cool because it doesn't generate a picture or generate words, but almost every single company in the world processes data in the background.

Jensen Huang: And Nvidia's GPUs are the only accelerators on the planet that process and accelerate data. Sequel data, pandas data, data science, toolkits like pandas and the new one Pullers. These are the most popular data processing platforms in the world. And aside from CPUs, which, as I've mentioned before, are really running out of steam, Nvidia's accelerated computing is really the only way to get boosting performance out of that.

Jensen Huang: And then general robotics, the big transformation last year as we are able to now learn physical AI from watching video and human demonstration and synthetic data generation from reinforcement learning from systems like omniverse, we're now able to work with just about every robotics companies now to start thinking about start building. And so you can see that there are just so many different directions that general to the AI is going, and so we're actually seeing the momentum of general to the AI accelerating.

Speaker Change: and Nvidia GPUs are the only accelerators on the planet that process and accelerate data.

Speaker Change: SikoDera, [inaudible]

Speaker Change: Pandas, DataDataScience, ToolKits, like Pandas and the new one, Polars. These are the most popular data processing platforms in the world.

Speaker Change: A5 from CPUs, which as I've mentioned before, really running out of steam, Nvidia has accelerated computing as is really the only way to get boosting performance out of that.

Jensen Huang: And so that's number one: the primary, the number one use case, long before generative AI came along, is that the migration of applications, one after another, to accelerated computing. The second is, of course, the rentals. They're renting capacity to model makers or renting it to startup companies. And a generative AI company spends the vast majority of their invested capital into infrastructure. So that they could use an AI to help them create products. And so these companies need it now. Now, they just simply can't afford, you know, to use race money. They want you to put it to use now.

Speaker Change: And so that's number one, is the primary, the number one use case long before JPMorgan was the migration of applications one after another to accelerate computing. The second.

Jensen Huang: Antichia, to answer your question regarding sovereign AI and our goals in terms of growth and terms of revenue. It certainly is a unique and growing opportunity, something that surfaced with genital AI, and the desires of countries around the world have their own genital AI that would be able to incorporate their own language, incorporate their own culture, incorporate their own data in that country. So more and more excitement around these models and what they can be specific for those countries. So yes, we are seeing some growth opportunity in front of us.

Speaker Change: The second is, of course, the rentals. They're renting capacity to model makers, are renting it to startup companies, and a general to the AI company.

Speaker Change: Spence the vast majority of their invested capital into infrastructure so that they could use an AI to help them create products.

Speaker Change: and enter these companies need it now.

Speaker Change: They just simply can't afford, you know, use race money, they want you to...

Jensen Huang: You have processing thing you have to do. You can't do it next year. You have to do it today. And so, so there's a, there's a fair, that's one reason. The second reason for hopper demand right now is because of the race to the next plateau. The revolutionary level of AI. The second person who gets there is incrementally, you know, better or about the same. And so, so the ability to systematically and consistently race to the next plateau and be the first one there is how you establish leadership. You know, NVIDIA is constantly doing that.

Speaker Change: Put it to use now. You have processing and you have to do. You can't do it next year. You have to do it today. So there's one reason. The second reason for Hopper demand right now is because of the race to the next plateau.

Joe Moore: And your next question comes from the line of Joe Moore with Morgan Stanley. Your line is open. Great. Thank you. Vincent, in the press release, you talked about Blackwell anticipation being incredible, but it seems like Hopper demand is also really strong. I mean, you're guiding for a very strong quarter without Blackwell in October. So, you know, how long do you see sort of coexisting strong demand for both? And can you talk about the transition to Blackwell?

Speaker Change: The first person to the next plateau gets to be a revolutionary level of AI.

Joe Moore: Do you see people intermixing clusters? Do you think most of the Blackwell activity's new cluster is just some sense of what that transition looks like? Yeah, thanks. The demand for Hopper is really strong. And it's true. The demand for Blackwell is incredible. There's a couple of reasons for that. The first reason is if you just look at the world's cloud service providers in the amount of GPU capacity they have available, it's basically none.

Speaker Change: The second person who gets there is incrementally better or about the same.

Speaker Change: and so the ability to systematically...

Speaker Change: and consistently race to the next plateau and be the first one there. It's how you establish leadership.

Jensen Huang: And we show that to the world and the GPUs we make and AI factories that we make, the networking systems that we make, the SOCs we create. I mean, we want; we want to set the pace. We want to be consistently the world's best. And that's the reason why we drive ourselves so hard. Of course, we also want to see our dreams come true and all of the capabilities that we imagine in the future and the benefits that we can bring to society. We want to see all that come true. And so, these model makers are the same.

Nvidia: You know, Nvidia is constantly doing that and we show that to the world and the GPUs we make and the AI factories that we make, the networking systems that we make, the SOCs we create. I mean, we want to set the pace, we want to be consistently the world's best. And that's the reason why we drive ourselves so hard.

Nvidia: Of course, we also want to see our dreams come true and all of the capabilities that we imagine in the future and the benefits that we can bring to society. We want to see all that come true.

Joe Moore: And the reason for that is because they're either being deployed internally for accelerating their own workloads data processing, for example. Data processing, you know, we hardly ever talk about it because it's mundane. You know, it's not, it's not very cool because it doesn't generate a picture or, you know, generate words, but almost every single company in the world processes data in the background. And Nvidia's GPUs are the only accelerators on the planet that process and accelerate data.

Jensen Huang: They're, of course, they want to be the world's best. They want to be the world's first.

Nvidia: And so, these model makers are the same, they're, of course, they want to be the world's best, they want to be the world's first.

Jensen Huang: And, and although Blackwell will start shipping out in billions of dollars at the end of this year, the standing up of the capacity is still probably, you know, weeks and a month or so away. And so, between now and then, is a lot of generative AI market dynamic. And so everybody is just really in a hurry. It's either operational reasons that they needed. They need accelerated computing. They don't want to build any more general purpose computing infrastructure. And even Hopper, you know, of course, age 200 state of the art. Hopper, if you have a choice between building CPU infrastructure right now for business or Hopper infrastructure for business right now, that decision is relatively clear.

Nvidia: And...

Blackwell: And although Blackwell will start shipping out in billions of dollars at the end of this year, the standing up at the capacity is still probably weeks and a month or so away.

Blackwell: And so, between now and then, it's a lot of generative AI market dynamic, and so, everybody is just really in a hurry. It's either operational reasons that they need it, they need accelerated computing.

Joe Moore: SQL data, pandas, data data science, tool kits, like pandas and the new one, pullers. These are the most popular data processing platforms in the world. And aside from CPUs, which as I've mentioned before, really running out of steam, Nvidia's accelerated computing is really the only way to get boosting performance out of that.

Blackwell: Um, they don't want to build any more.

Blackwell: General Purpose Computing Infrastructure, and even Hopper, you know, of course, H200s, State of the Art, Hopper.

Blackwell: If you have a choice between...

Jensen Huang: And so that's number one is the primary, the number one use case long before generative AI came along is that the migration of applications one after another to accelerate computing. The second, the second is of course rent the rentals. They're renting capacity to model makers or renting it to startup companies. And a generative AI company spends the vast majority of their invested capital into infrastructure so that they could use an AI to help them create products.

Blackwell: Building.

Blackwell: CPU infrastructure right now for business or copper.

Jensen Huang: And so, I think people are just clamoring to transition the trillion dollars of established installed infrastructure to a modern infrastructure, and hopper saved the art.

Blackwell: Infrastructure for Business Right Now, that decision is relatively clear, and so I think people are just clamoring to transition the trillion dollars of established installed infrastructure to a modern infrastructure in Hopper State of York.

Matt Ramsey: And your next question comes from the line of Matt Ramsey with TD Cowan. Your line is open. And thank you very much. Good afternoon, everybody.

Speaker Change: And your next question comes from the line of Matt Ramsey with TV Cowan. Your line is open.

Matt Ramsey: I wanted to kind of circle back to an earlier question about the debate that investors are having about the ROI on all of this capex. And hopefully this question and the distinction will make some sense. But what I'm, what I'm having discussions about is with like the percentage of folks that you see that are spending all this money. And looking to sort of push the frontier towards AGI convergence. And, as you just said, a new plateau and capability. And they're going to I want to spend regardless to get to that level of capability because it opens up so many doors for the industry and for their company versus customers that are really, really focused today on CapEx versus ROI.

Matt Ramsey: Thank you very much. Good afternoon, everybody.

Jensen Huang: And so these companies need it now. NVIDIA Corp. It's a revolutionary level of AI. The second person who gets there is incrementally better or about the same. The ability to systematically and consistently race to the next plateau and be the first one there is how you establish leadership. Nvidia is constantly doing that and we show that to the world and the GPUs we make and AI factories that we make, the networking systems that we make, the SOCs we create.

Matt Ramsey: And I wanted to kind of circle back to an earlier question about the debate that investors are having about.

Matt Ramsey: The ROI on all of this capex and hopefully this question and the distinction will make some sense. But what I'm having discussions about is with the percentage of folks that you see that are spending all of this money and looking to sort of.

Matt Ramsey: Push the Frontier towards...

Speaker Change: AGI Convergence, and as you just said, a new plateau in Capability, and they're going to spend regardless to get to that level of capability because it opens up so many doors for the industry and for their company. Versus customers that are really, really focused today on…

Jensen Huang: I don't know if that distinction makes sense. I'm just trying to get a sense of how you're seeing the priorities of people that are putting the dollars in the ground on the new technology and what their priorities are and their timeframes are for that investment. Thanks. The people who are investing in Nvidia infrastructure are getting returns on it right away. It's the best ROI infrastructure, computing infrastructure investment you can make today.

Speaker Change: CapEx versus ROI. I don't know if that distinction makes sense. I'm just trying to get a sense of how you're seeing the priorities of people that are putting the dollars in the ground on the new technology and what their priorities are and their timeframes are for that investment. Thanks.

Jensen Huang: We want to set the pace. We want to be consistently the world's best and that's the reason why we drive ourselves so hard. Of course we also want to see our dreams come true and all of the capabilities that we imagine in the future and the benefits that we can bring to society. We want to see all that come true and so these model makers are the same. Of course they want to be the world's best, they want to be the world's first and although Blackwell will start shipping out in billions of dollars at the end of this year, the standing up of the capacity is still probably weeks and a month or so away.

Speaker Change: Thanks, Matt. The people who are investing in Nvidia infrastructure are getting returns on it right away.

Speaker Change: It's the best ROI infrastructure computing infrastructure investment you can make today.

Jensen Huang: One way to think through it is probably the easiest way to think through it is to go back to first principles. You have a trillion dollars with a general purpose computing infrastructure, and the question is: do you want to build more of that or not? For every billion dollars with a general CPU-based infrastructure that you stand up, you probably rent it for less than a billion. Because it's commoditized, there's already a trillion dollars on the ground. What's the point of getting more? The people who are clamoring to get this infrastructure won. When they build out upper based infrastructure and soon black well based infrastructure, they start saving money.

Speaker Change: And so one way to think through it, you know, probably the easiest way to think through it, it's just go back to first principles.

Speaker Change: You have a trillion dollars with a general purpose computing infrastructure, and a question if you want to build more that or not.

Speaker Change: And for every billion dollars with a general CPU based infrastructure, the use stand-up, you probably rented for less than a billion.

Speaker Change: And so, because it's commontized, there's already its trillion dollars on the ground. What's the point of getting more? And so, the people who are clamoring to get the infrastructure at one.

Jensen Huang: Between now and then is a lot of generative AI market dynamic and so everybody is just really in a hurry. It's either operational reasons that they need it, they need accelerated computing. They don't want to build anymore general purpose computing infrastructure and even Hopper, of course H200 state of the art, Hopper, if you have a choice between building CPU infrastructure right now for business or Hopper infrastructure for business right now, that decision is relatively clear and so I think people are just clamoring to transition the trillion dollars of established installed infrastructure to a modern infrastructure in Hopper state of the art.

Speaker Change: When they build out hopper-based infrastructure and, soon, Blackwell-based infrastructure, they start saving the money.

Jensen Huang: That's tremendous return on investment. The reason why they start saving money is because data processing saves money; data processing is just a giant part of it already. So recommender system saves money, so on and so forth. So you start saving money. The second thing is everything you stand up are going to get rented because so many companies are being founded to create generative AI, and so your capacity is rented right away, and the return on investment of that is really good. And the third reason is your own business. You want to either create the next frontier yourself or your own internet services benefit from a next generation ad system or next generation recommender system or next generation search system.

Speaker Change: That's tremendous return on investment and reason why they start saving money in is because data processing saves money, you know, data processing is just a giant part of it already. And so recommender system saves money.

Speaker Change: So, it's over, okay? And so, you start saving money.

Speaker Change: The second thing is, everything you stand up, are going to get rented.

Speaker Change: Because so many companies are being founded to create your energy to the AI, and so your capacity to get rented right away. And the return on investment that that is really good.

Matt Ramsey: And your next question comes from the line of Matt Ramsey with TD Cowan. Your line is open. Thank you very much.

Speaker Change: And on the third reason is your own business.

Speaker Change: You want to either create the next frontier yourself or your own internet services, benefit from a next generation ad system, or next generation recommend their system, or next generation search system.

Jensen Huang: Good afternoon everybody. I wanted to kind of circle back to an earlier question about the debate that investors are having about the ROI on all of this CAPEX and hopefully this question and the distinction will make some sense but what I'm having discussions about is with like the percentage of folks that you see that are spending all this money and looking to sort of push the frontier towards AGI convergence and as you just said a new plateau and capability and they're going to We're going to spend regardless to get to that level of capability because it opens up so many doors for the industry and for their company versus customers that are really, really focused today on CapEx versus ROI.

Jensen Huang: So for your own services, for your own stores, for your own user generated content, social media platforms, for your own services, the generative AI is also a fast ROI. And so there's a lot of ways you could think through it, but at the core, it's because it is the best computing infrastructure you could put in the ground today. The world of general purpose computing is shifting to accelerated computing. The world of human engineered software is moving to generative AI software. If you were to build infrastructure to modernize your cloud in your data centers, build it with accelerated computing and video.

Speaker Change: So, for your own services, for your own stores, for your own user-generated content, social media platforms.

JPMorgan AI: For your own services, JPMorgan AI is also a fast ROI. There's a lot of ways you can think through it, but at the core, it's because it is the best.

Speaker Change: Computing, Infistructure, U-Kid, Put in the ground today.

Speaker Change: The world of general purpose computing is shifting to accelerated computing. The world of human engineered software is moving to generative AI software.

Speaker Change: If you were to build infrastructure.

Jensen Huang: I don't know if that distinction makes sense. I'm just trying to get a sense of how you're seeing the priorities of people that are putting the dollars in the ground on the new technology and what their priorities are and their timeframes are for that investment. Thanks, Matt. The people who are investing in Nvidia infrastructure are getting returns on it right away. It's the best ROI infrastructure, computing infrastructure investment you can make today.

Speaker Change: to modernize your cloud and your data centers, build it with Excel-ready computing Nvidia. That's the best way to do it.

Jensen Huang: That's the best way to do it.

Timothy Arcuri: And your next question comes from the line of Timothy Arkory with UBS. Your line. is open. Thanks a lot.

Speaker Change: And your next question comes from the line of Timothy Arkary with UBS, your line is open.

Timothy Arcuri: I had a question on the shape of the revenue growth, both near and longer term. I know Colette; you did increase OpEx for the year, and if I look at the increase in your purchase commitments and your supply obligations, that's also quite bullish. On the other hand, there's some school of thought that not that many customers really seem ready for liquid cooling, and I do recognize that some of these racks can be air-cooled. But Jensen, is that something to consider on the shape of how Blackwell is going to ramp? And then I guess when you look beyond next year, which is obviously going to be a great year, and you look into 26, do you worry about any other gating factors, like say the power supply chain, or at some point models start to get smaller?

Timothy Arkary: Thanks a lot. I had a question on the shape of the revenue growth both near and longer term. I know, Colette, you did increase OpEx for the year and if I look at the increase in your purchase commitment and your supply obligations.

Jensen Huang: And so one way to think through it, probably the easiest way to think through it is to go back to first principles. You have a trillion dollars with a general purpose computing infrastructure and the question is do you want to build more of that or not? And for every billion dollars with a general CPU based infrastructure that you stand up, you probably rent it for less than a billion. And so because it's commoditized, there's already a trillion dollars on the ground.

Speaker Change: That's also quite bullish.

Speaker Change: On the other hand, there's some school thought that not that many customers really seem ready for liquid cooling and I do recognize that some of these racks can be aircooled.

Speaker Change: Jensen, that's something to consider on the shape of how black was going to ramp. Then I guess when you look beyond next year, which is obviously going to be a great year and you look into 26, do you worry about any other, you know, gaining factors like, say the power?

Jensen Huang: What's the point of getting more? And so the people who are clamoring to get this infrastructure won. When they build out hopper based infrastructure and soon black well based infrastructure, they start saving money. That's tremendous return on investment. And the reason why they start saving money is because data processing saves money, data processing is just a giant part of it already. And so recommender system save money, so on and so forth.

Timothy Arcuri: I'm just wondering if you can speak to that. Thanks.

Jensen: Supply Chain or at some point models start to get smaller, I'm just wondering if you can speak to that thing.

Jensen Huang: I'm going to work backwards. I really appreciate the question, Tim. So remember her, the world is moving from general purpose computing to accelerated computing. And the world builds about a trillion dollars with the data centers; a trillion dollars with data centers in a few years will be all accelerated computing. In the past, no GPUs are in data centers, just CPUs. In the future, every single data center will have GPUs. And the reason for that is very clear because we need to accelerate workloads so that we can continue to be sustainable, continue to drive down the cost of computing so that when we do more computing, we don't experience computing inflation.

Speaker Change: I'm going to work backwards. I really appreciate the question tip. So, remember her.

Speaker Change: The world is moving from general purpose computing to accelerate computing and the world builds about a trillion dollars with the data centers.

Speaker Change: You know, a trillion dollars with data centers in a few years will be all accelerating computing.

Speaker Change: In the past, no GPUs are in data centers, just CPUs, in the future, every single data center, well GPUs.

Jensen Huang: And so you start saving money. The second thing is everything you stand up are going to get rented because so many companies are being founded to create your energy of AI. And so your capacity is rented right away. And the return on investment of that is really good. And the third reason is your own business. You want to either create the next frontier yourself or your own internet services benefit from a next generation ad system or next generation recommender system or next generation search system.

Speaker Change: And the reason for that is very clear, because we need to accelerate workloads so that we can continue to be sustainable, continue to drive down the cost of computing, so that when we do more computing, we don't experience computing inflation.

Jensen Huang: Second, we need GPUs for a new computing model called generative AI that we can all acknowledge is going to be quite transformative to the future of computing. And so I think working backwards, the way to think about that is the next trillion dollars of the world's infrastructure will clearly be different than the last trillion. And it'll be vastly accelerated. With respect to the shape of our RAMP, we offer multiple configurations of Blackwell. Blackwell comes in either a Blackwell classic, if you will, that uses the HGX form factor that we pioneered with Volta. And I think it was Volta.

Speaker Change: Second, we need GPU to use for a new computer model called Generative AI that we can all acknowledge is going to be quite transformative to the future of computing.

Speaker Change: And so, I think working backwards, the way to think about that is the next trillion dollars of the world's infrastructure will clearly be different than the last trillion, and it will be vastly accelerated.

Jensen Huang: So for your own services, for your own stores, for your own user generated content, social media platforms, you know, for your own services, the general to the AI is also a fast ROI. And so there's a lot of ways you could think through it, but at the core, it's because it is the best computing infrastructure you could put in the ground today. The world of general purpose computing is shifting to accelerated computing. The world of human engineered software is moving to generative AI software.

Speaker Change: With respect to the shape of our ramp, we offer multiple configurations of Blackwell.

Speaker Change: Blackwell comes in either a blackwell classic if you will, that uses the HGX form factor that we pioneered with Volta.

Jensen Huang: And so we've been shipping the HGX form factor for some time. It is air-cooled. The gray Blackwell is liquid cooled. However, the number of data centers that want to go liquid cooled is quite significant. The reason for that is because we can, in a liquid cooled data center, in any data-powered limited data center, whatever size data centers you choose, you can install and deploy anywhere from three to five times the AI throughput compared to the past. And so liquid cooling is cheaper. Liquid cooling; our TCO is better. And liquid cooling allows you to have the benefit of this capability we call MV-Link, which allows us to expand it to 72 gray Blackwell packages, which has essentially 144 GPUs.

Speaker Change: and I think it was voltage. And so, we've been shipping the H-CS, H-CS format factor for some time.

Speaker Change: and AirCooled. Bye.

Speaker Change: The Grace Blackwell is liquid-cooled, however, the number of data centers that want to go to liquid-cooled is quite significant and the reason for that is because we can, in a liquid-cooled data center.

Jensen Huang: If you were to build infrastructure to modernize your cloud and your data centers, build it with accelerated computing and video. That's the best way to do it.

Speaker Change: In any data center, power limit of data center, whatever size data centers you choose, you can install and deploy anywhere from 3 to 5 times the AI throughput compared to the past.

Timothy Arcuri: And your next question comes from the line of Timothy Arkory with UBS. Your line, is open. I'm going to work backwards, I really appreciate the question, Tim.

Speaker Change: And so, look at cooling, is cheaper?

Speaker Change: LiquidCooling, OptCO is better, and LiquidCooling allows you to have the benefit.

Speaker Change: of this Capability, we call MBLink.

Speaker Change: which allows us to expand it to 72 great Blackwell packages which has essentially a 144GT.

Jensen Huang: And so imagine 144 GPUs connected in MV-Link. And that when we're increasingly showing you the benefits of that. And the next click is obviously very low latency, very high throughput, large language model inference. And the large MV-Link domain is going to be a game changer for that. And so I think people are we are very comfortable deploying both, and so almost every CSP we're working with are deploying some of both. And so I'm pretty confident that that will ramp it up just fine. Your second question out of the third is that looking forward yet next year is going to be a great year.

Speaker Change: And so imagine a 144 GPU is connected in V-Link, and that, when is, we're increasingly showing you the benefits of that, and the next, you know, the next click is, obviously, very low latency, very high throughput, large language model inference.

Speaker Change: and the large, and beaming domain is going to be a gain changer for that. And so, I think, I think people are...

Jensen Huang: So, remember, the world is moving from general purpose computing to accelerated computing, and the world builds about a trillion dollars with data centers, a trillion dollars with data centers in a few years will be all accelerated computing. In the past, no GPUs are in data centers, just CPUs, in the future, every single data center will have GPUs. And the reason for that is very clear, because we need to accelerate workloads so that we can continue to be sustainable, continue to drive down the cost of computing so that when we do more computing, we don't experience computing inflation. Second, we need GPUs for a new computing model called generative AI that we can all acknowledge is going to be quite transformative to the future of computing.

Speaker Change: are very comfortable at deploying both, and so almost every CSP we're working with are deploying some of both, and so I'm pretty confident that we'll ramp it up just just fine.

Speaker Change: Your second question out of the third.

Jensen Huang: We expect to grow our data center business quite significantly next year. Blackwell is going to be a complete game changer for the industry, and Blackwell is going to carry into the following year. And, as I mentioned earlier, working backwards from first principles.

Speaker Change: And looking forward, yeah, next year is going to be a great year.

Speaker Change: We expect to grow our data center business quite significantly next year, Blackwall is going to be going to be a complete game changer for the industry, and Blackwall is going to carry into the following year, and as I mentioned earlier, working backwards from first principles.

Jensen Huang: Remember that computing is going through two platform transitions at the same time. And that's just really, really important to keep your head on, your mind focused on, which is general purpose computing is shifting to accelerated computing and human engineered software is going to transition to Q&A AI or artificial intelligence learning software.

Speaker Change: Remember that computing is going through two platform transitions at the same time, and that's just really, really important to keep your mind focused on, which is, General purpose computing is shifting to accelerated computing, and human engineer software is going to transition to JPMorgan AI or artificial intelligence-learn software.

Jensen Huang: And so, I think working backwards, the way to think about that is the next trillion dollars of the world's infrastructure will clearly be different than the last trillion, and it will be vastly accelerated. With respect to the shape of our ramp, we offer multiple configurations of blackwell. Blackwell comes in either a blackwell classic, if you will, that uses the HGX form factor that we pioneered with Volta, and I think it was Volta.

Stacy Raskin: Can Jordan's next question come from the line of Stacy Raskin with Bernstein Research? Your line is open. Hi guys, so thanks for taking my questions.

Speaker Change: [inaudible]

Speaker Change: And your next question comes from the line of Stacey Raskin with Bernstein Research. Your line is open.

Stacy Raskin: I have two short questions for Collect. The first, several billion dollars of Blackwell revenue in Q4. Is that additive? You said you expected hopper demand to strengthen in the second half. That mean hopper strength is Q2 to Q4 as well on top of Blackwell added several billion dollars.

Stacey Raskin: Hi guys, thanks for taking my questions. I have two short questions for collect.

Speaker Change: The first, several billion dollars of Black O'Revenue in the Q4.

Speaker Change: I think that additive, you need to stick in the hopper demand to strengthen in the second after that mean hopper strengthens 222-24 as well on top of blackwell.

Jensen Huang: And so, we've been shipping the HGX form factor for some time. It is air cooled. The gray blackwell is liquid cooled, however, the number of data centers that want to go to liquid cooled is quite significant. The reason for that is because we can, in a liquid cooled data center, in any power limited data center, whatever size data centers you choose, you can install and deploy anywhere from three to five times the AI throughput compared to the past.

Stacy Raskin: And the second question on gross margin is if I have mid mid 70s for the year. Where I want to draw that if I have 75th of the year, I'd be something like 71 to 72. For Q4 somewhere knowing, is that the kind of exit rate for gross margins that you're expecting, and how should we think about the drivers of gross margin evolution in the next year as Blackwell ramps. I mean, hopefully, I guess the yields and the inventory was always everything come up.

Speaker Change: AddInCoverBillionDollins, and the second question on gross margin is, if I have mid-70s for the year.

Speaker Change #100: We're going to draw that. If I have 75 to be here, I'd be something like 71 to 72.

Speaker Change #101: For Q4, somewhere in that ring, is that the kind of exit rate for those margins that you're expecting and how should we think about the drivers of gross margin evolution in the next year as blackwell ramps? I mean, hopefully, I guess the yields and the inventory is always in everything come up.

Stacy Raskin: This is Stacy. Let's first take your question that you had about Hopper and Blackwell. So we believe our hopper will continue to grow into the second half. We have many new products for hopper or existing products for hopper that we believe will start continuing to ramp in the next quarters, including our Q3. And those new products moving to Q4. So let's say Hopper there for versus H1 is a growth opportunity for that. Additionally, we have the Blackwell on top of that, and the Blackwell starting of ramping in Q4. So hope that helped you on those two pieces.

Speaker Change #101: no

Speaker Change #102: If you have any questions about Hopper and Blackwell.

Jensen Huang: And so, liquid cooling is cheaper, liquid cooling, our TCO is better, and liquid cooling allows you to have the benefit of this capability we call MV link, which allows us to expand it to 72 gray blackwell packages, which has essentially 144 GPUs. And so, imagine 144 GPUs connected in MV link, and that, when we're increasingly showing you the benefits of that, and the next, you know, the next click is obviously very low latency, very high throughput, large language model inference, and the large MV link domain is going to be a game changer for that.

Speaker Change #103: So we believe our hopper will continue to grow into the second half.

Speaker Change #103: We have many new products for hopper, or existing products for hopper.

Speaker Change #103: that we believe will start continuing to ramp in the next quarters, including our Q3 and those new products moving to Q4. So let's say Hopper there for versus H1 is a growth opportunity for that.

Speaker Change #103: Additionally, we have the blackwell on top of that, and the blackwell starting of ramping into four.

Colette Kress: Your second piece is in terms of our gross margin. We provided gross margin for our Q3. We provided our gross margin on the non-GAAP at about 75. We'll work with all the different transitions that we're going through, but we do believe we can do that 75 in Q3. We provided that we're still on track for the full year, also in the mid 70s or approximately the 75. So we're going to see some slight difference possibly in Q4, again with our transitions and the different cost workers that we have on our new product introduction. However, I'm not in the same number that you are there.

Speaker Change #103: So, hope that helps you on those two pieces.

Speaker Change #103: Your second piece is in terms of on our gross margin. We provided gross margin for our Q3. We provided our gross margin on the non-gap at about 75. We'll work with all the different transitions that we're going through, but we do believe we can do that 75 in Q3.

Jensen Huang: And so, I think, I think people are, Deploying. Our very comfortable deploying both, and so almost every CSP we're working with, are deploying some of both, and so I'm pretty confident that that will ramp it up just fine.

Jensen Huang: Your second question out of the third is that looking forward, yet next year is going to be a great year. We expect to grow our data center business quite significantly next year. Blackwell is going to be a complete game changer for the industry, and Blackwell is going to carry into the following year, and as I mentioned earlier, working backwards from first principles.

Speaker Change #104: We provided that we're still on track for the full year, also in the mid-70s or approximately the 75s. So we're going to see some slight difference possibly in Q4. Again, with our transitions and the different clusters that we have on our new product introductions. However, I'm not in the same number that you are. There, we don't have exactly guidance, but I do believe you're lower than where we are.

Colette Kress: We don't have exactly guidance, but I do believe you're lower than where we are.

Jensen Huang: Remember that computing is going through two platform transitions at the same time, and that's just really, really important to keep your mind focused on, which is general purpose computing is shifting to accelerated computing, and human engineered software is going to transition to Q&A AI, or artificial intelligence learned software.

Benjamin Reitzes: And your next question comes from the line of Ben Ritesus with Melius. Your line is open. Yeah, thanks a lot for the question, Jensen and Collette. I wanted to ask about the geographies. There was the Tanki that came out, and the United States was down sequentially, while several Asian geographies were up a lot sequentially. Just wondering what the dynamics are there, and obviously China did very well. You mentioned your remarks. What are the puts and takes?

Speaker Change #105: And your next question, come to the line of Ben Ritesis with Melias, your line is open.

Ben Ritesis: Yeah, hey, thanks a lot for the question, Jensen and Colette. I wanted to ask about the geographies.

Stacy Raskin: Can Jordan next question come from the line of Stacy Raskin with Bernstein Research? Your line is open. Hi guys, so thanks for taking my questions.

Speaker Change #107: There was the tank you that came out, and the United States was down, scrunchly, wall.

Speaker Change #108: DevRelations, Geographies were up a lot, sequentially, just wondering what the dynamics are there, you know, and...

Stacy Raskin: I have two short questions for collect. The first, several billion dollars of Blackwell revenue in Q4. Is that additive? You said you expected hopper demand to strengthen in the second half. That means hopper strength is Q3 to Q4 as well on top of Blackwell adding several billion dollars.

Speaker Change #109: Obviously, China did very well. You mentioned your remarks. What are the puts in takes? And then, I just wanted to clarify from state, these questions. If that means that this eventual overall revenue growth rates for the company accelerate the fourth quarter, given all those favorable revenue dynamics. Thanks.

Benjamin Reitzes: And then I just wanted to clarify from Stacy's question if that means the substantial overall revenue growth rates for the company accelerate the fourth quarter, given all those favorable revenue dynamics. Thanks.

Colette Kress: And the second question on gross margin is, if I have mid-70s for the year, it's being where I want to draw that. If I have 75 for the year, I'd be something like 71 to 72 for Q4. Is that the kind of exit rate for gross margins that you're expecting? And how should we think about the drivers of gross margin evolution in the next year as Blackwell ramps? I mean, hopefully I guess the yields and the inventory results have everything come up.

Colette Kress: Let me talk about a bit in terms of our disclosure in terms of the Tanki. I require disclosure and a choice of geographies. Very challenging sometimes to create that right disclosure as we have to come up with one key piece pieces in terms of we have in terms of who we sell to and or specifically who we invoice to. And so what you're seeing in terms of there is who we invoice; that's not necessarily where the product will eventually be and where it may even travel to the end customer. These are just moving to our OEMs or ODMs and our system integrators for the most part across our product portfolio.

Speaker Change #110: Let me talk about a bit in terms of our disclosure in terms of the time to a required disclosure and a choice of geographies. Very challenging sometimes to create that right disclosure as we have to come up with one key piece.

Speaker Change #111: This is in terms of we have in terms of who we sell to and or specifically who we invoice to.

Colette Kress: This is Stacy. Let's first take your question that you had about hopper and Blackwell. So we believe our hopper will continue to grow into the second half. We have many new products for hopper, our existing products for hopper that we believe will start continuing to ramp in the next quarters, including our Q3 and those new products moving to Q4. So let's say hopper there for versus H1 is a growth opportunity for that. Additionally, we have the Blackwell on top of that and the Blackwell starting of ramping in Q4. So hope that helped you on those two pieces.

Speaker Change #112: And so what you're saying in terms of there is who we invoice. That's not necessarily where the product will eventually be and where it may even travel to the end customer. These are just moving to our OEMs, our ODMs, and our system integrators for the most part across our product portfolio. So what you're saying there is sometimes just a shift in terms of who they are using to complete their full configuration before those things are going into the data set. Going into notebooks and those pieces of it.

Colette Kress: So what you're seeing there is sometimes just a swift shift in terms of who they are using to complete their full configuration before those things are going into the data center, going into notebooks and those pieces of it, and that shift happens from time to time. But yes, are trying to number there are inverse into China. Keep in mind that is incorporating both gaming also data center also automotive in those numbers that we have going back to your statement in regarding gross margin and also what we're seeing in terms of what we're looking at for hopper and black well in terms of revenue.

Speaker Change #112: and that shift happens all from time to time, but yes, are trying to number there or inversing to China, keep in mind that it is incorporating both gaming, also data center, also automotive, and those numbers that we have.

Colette Kress: Your second piece is in terms of on our gross margin. We provided gross margin for our Q3. We provided our gross margin on the non-gap at about 75. We'll work with all the different transitions that we're going through, but we do believe we can do that 75 in Q3. We provided that we're still on track for the full year also in the mid 70s or approximately the 75. So we're going to see some slight difference possibly in Q4 again with our transitions and the different cost structures that we have on our new product introduction. However, I'm not in the same number that you are there. We don't have exactly guidance, but I do believe you're lower than where we are.

Speaker Change #113: Going back to your statement regarding gross margin and also what we're seeing in terms of what we're looking out for hopper and blackwell in terms of revenue.

C.J. Muse: Hopper will continue to grow in the second half will continue to grow from what we are currently seeing during determining that exact next in each Q3 Q4 we don't have here we are not here to guide yet in terms of Q4 but we do see right now the demand expectations. We do see the visibility that that will be a growth opportunity in Q4. On top of that, we will have our black well architecture. And your next question comes from the line of C.J.

Speaker Change #114: Offer will continue to grow in the second half, will continue to grow from what we are currently seeing.

Speaker Change #115: during determining that exact next.

Speaker Change #115: In each two three and two four, we don't have here, we are not here to guide yet in terms of Q4. But we do see right now the demand expectations we do see, the visibility that that will be a growth opportunity in Q4. On top of that, we will have our BlackRaw architecture.

C.J. Muse: Muse with Cantor Fitzgerald; your line is open. Yeah, good afternoon. Thank you for taking the question.

Speaker Change #116: And your next question comes from the line of CJ Muse with Cantor Fitzgerald. Your line is open.

Benjamin Reitzes: And your next question comes from the line of Ben Ritesus with Melius. Your line is open. Yeah, hey. Thanks a lot for the question, Jensen and Collette. I wanted to ask about the geographies. There was the Tanki that came out and the United States was down sequentially while several Asian geographies were up a lot sequentially. Just wondering what the dynamics are there. You know, and obviously China did very well. You mentioned your remarks.

Jensen Huang: You've been barked on a remarkable annual product cadence with challenges only likely becoming more and more given in a rising complexity in a radical limit advanced package world. So curious you know if you take a step back how does this backdrop all people are thinking around potentially greater vertical integration supply chain partnerships and and then taking through consequential impact to your margin post profile. Thank you. Yeah, thanks. Let's see. I think the first answer to your, the answer to your first question is that the reason why our velocity is so high is simultaneously because the complexity of the model is growing.

CJ Muse: Yeah, good afternoon. Thank you for taking the question. You've embarked on a remarkable annual product cadence with challenges only likely becoming more and more given in a rising complex and in a rather, competitive limit advanced-packed world. So, curious, you know, if you take a step back, how does this background fall for your thinking around potentially greater vertical integration, supply chain partnerships and then taking through consequential impact to your margin post profile. Thank you.

CJ Muse: i

Speaker Change #118: Yeah, thanks, let's see.

Benjamin Reitzes: What are the puts and takes. And then I just wanted to clarify from Stacy's question if that means the substantial overall revenue growth rates for the company accelerate the fourth quarter given all those favorable revenue dynamics.

Speaker Change #119: I think the first answer to your first question is that the reason why our velocity is so high is simultaneously because the complexity of the model is growing.

Jensen Huang: And we want to continue to drive its cost down. It's growing. So we want to continue to increase its scale. And we believe that by continuing to scale the AI models, that will reach a level of extraordinary usefulness. And that it would open up. I realize the next Industrial Revolution. We believe it. And so we're going to drive ourselves really hard to continue to go up that scale. We have the ability fairly uniquely to integrate to design an AI factory because we have all the parts. It's not possible to come up with a new AI factory every year unless you have all the parts.

Colette Kress: Thanks. Let me talk about a bit in terms of our disclosure in terms of the Tanki. I require disclosure and a choice of geographies. Very challenging sometimes to create that right disclosure as we have to come up with one key piece pieces in terms of we have in terms of who we sell to and or specifically who we invoice to. And so what you're seeing in terms of there is who we invoice.

Speaker Change #119: and we want to continue to drive its cost down.

Speaker Change #119: It's growing so we want to continue to increase its scale.

Speaker Change #119: and we believe that by continuing to scale the ARM models that will reach a level of extraordinary usefulness. And that would open up, I realize the next industrial revolution. We believe it.

Speaker Change #119: and so we're going to drive ourselves really hard to continue to go up that scale.

Colette Kress: That's not necessarily where the product will eventually be and where it may even travel to the end customer. These are just moving to our OEMs or ODMs and our system integrators for the most part across our product portfolio. So what you're seeing there is sometimes just a swift shift in terms of who they are using to complete their full configuration before those things are going into the data center, going into notebooks and those pieces of it.

Speaker Change #119: We have the ability.

Speaker Change #119: Fairly uniquely to integrate to design an AI factory because we have all the parts.

Speaker Change #119: It's not possible to come up with a new AI factory every year, unless you have all the parts.

Jensen Huang: And so we have, next year we're going to ship a lot more CPUs than we've ever had in the history of our company, more GPUs, of course. But also MV link switches, CX, DPUs connect XEP for East and West, Bluefield DPUs for North and South, and data and storage processing to InfiniBand for supercomputing centers to Ethernet, which is a brand new product for us, which is well on its way to becoming a multi-billion dollar business to bring AI to Ethernet. And so the fact that we have access to all of this, we have one architectural stack, as you know, it allows us to introduce new capabilities to the market as we complete it.

Speaker Change #119: And so we have next year we're going to ship a lot more CPUs than we've ever had in the history of our company, more GPUs of course.

Colette Kress: And that shift happens from time to time. But yes, our trying to number there are inverse into China. Keep in mind that is incorporating both gaming. Also data center also automotive in those numbers that we have going back to your statement regarding gross margin and also what we're seeing in terms of what we're looking at for hopper and black well in terms of revenue. Hopper will continue to grow in the second half will continue to grow from what we are currently seeing during determining that exact next in each Q3 and Q4.

Speaker Change #119: But also, in V-Link Switches.

Speaker Change #119: CX, DPUs, ConnectEx, EPU for eSum West, BlueField DPUs for North and South and Data and Storage.

Speaker Change #119: Processing, [inaudible]

Speaker Change #119: to InfiniBand for supercomputing centers to Ethernet, which is a brand new product for us, while on its way to becoming a multi-billion-dollar business to bring AI to Ethernet.

Speaker Change #119: And so, the fact that we could build, we have access to all of this, we have one architectural stack as you know.

Colette Kress: We don't have here. We are not here to guide yet in terms of Q4, but we do see right now the demand expectations. We do see the visibility that that will be a growth opportunity in Q4. On top of that, we will have our black well architecture.

Speaker Change #119: It allows us to introduce new capabilities to the market, as we complete it. Otherwise, what happens to you ship these parts, you go find customers to sell it to, and then you've got to build, somebody's got to build up an AI factory, and the AI factory's got a mountain of software.

Jensen Huang: Otherwise, what happens is you ship these parts, you go find customers to sell it to, and then you've got to build. Somebody's got to build up an AI factory. And the AI factory's got a mountain of software. And so it's not about, it's not about who integrates it. We love the fact that our supply chain is disintegrated. In the sense that we could service, you know, Quanta, Foxconn, HP, Dell, Lenovo, SuperMicro, we used to be able to service ZT; they were recently purchased and so on and so forth. And so the number of ecosystem partners that we have, Gigabyte, Assutes, the number of ecosystem partners that we have that allows us to allow us them to take our architecture, which all works, but integrated in a bespoke way into all of the world's cloud service providers' enterprise data centers.

CJ Muse: And your next question comes from the line of C.J. Muse with Cantor Fitzgerald. Your line is open. Yeah, good afternoon. Thank you for taking the question.

Speaker Change #119: And so, it's not about who integrates it, we love the fact that our supply chain is disintegrated. In the sense that we could service...

Jensen Huang: You've been barked on a remarkable annual product cadence with challenges only likely becoming more and more given in a rising complexity and a radical limit advance package world. So curious, you know, if you take a step back, how does this backdrop all people are thinking around potentially greater vertical integration supply chain partnerships and then taking through consequential impact to your margin post profile. Thank you. Yeah, thanks. Let's see, I think the first answer to your, the answer to your first question is that the reason why our velocity is so high is simultaneously because the complexity of the model is growing and we want to continue to drive its cost down.

Speaker Change #119: You know, Quanta, FoxCon, HP, Dell, Lenovo, SuperMicro, we used to be able to serve as the ET. They were recently purchased.

Speaker Change #119: and so on and so forth. And so the number of ecosystem partners that we have, Gigabyte, the number of ecosystem partners that we have, that allows them to take our architecture which all works.

Speaker Change #119: but integrated in a bespoke way into all of the world's cloud service providers and appliance data centers. The scale and reach necessary from our O-D-M, and our integrators, integrators' pledging, is vast and gigantic because the world is huge.

Jensen Huang: The scale and reach necessary from our ODMs and our integrators integrator supply chain is vast and gigantic because the world is huge. And so that part we don't we don't want to do, and we're not good at doing, and but we know how to design the infrastructure provided the way that customers would like it and lets the ecosystem integrated. Well, yeah, so anyways, that's the reason why.

Speaker Change #119: And so that part, we don't want to do and we're not good at doing, but we know how to design the AI infrastructure, provided the way that customers would like it and let the ecosystem integrate it. So anyways, that's the reason why.

Jensen Huang: It's growing, so we want to continue to increase its scale and we believe that by continuing the scale, the AI models, that will reach a level of extraordinary usefulness and that it would open up, realize the next industrial revolution, we believe it, and so we're going to drive ourselves really hard to continue to go up that scale. We have the ability fairly uniquely to integrate, to design an AI factory because we have all the parts.

Aaron Rakers: And your final question comes from the line of Aaron Rakers with Wells Fargo. Your line is open. Yes, thanks for taking the question. I wanted to go back into the Blackwell Products cycle. One of the questions that we tend to get asked is how you see the RAC scale system mix dynamic as you think about leveraging N.D. Link, you think about GB, you know, N.V.L. 72, and how that go to market, you know, dynamic looks, you know, as far as the Blackwell products cycle. I guess that puts us deeply. How do you see that mix of RAC scale systems as we start to think about the Blackwell cycle playing out?

Speaker Change #120: And your final question comes from the line of Aaron Rakers with Wells Fargo. Your line is open.

Aaron Rakers: Yes, thanks for taking a question.

Aaron Rakers: I wanted to go back into the Blackwell product cycle. One of the questions that we tend to get asked is how you see...

Aaron Rakers: The RACS scale system mix dynamic, as you think about leveraging ND links, you think about GB, NVL72, and how that go to market dynamic looks as far as the blackwall product cycle. I guess I've put this thing, how do you see that mix of RACS scale systems as we start to think about the blackwall cycle playing out?

Jensen Huang: It's not possible to come up with a new AI factory every year unless you have all the parts. Next year, we're going to ship a lot more CPUs than we've ever had in the history of our company, more GPUs of course, but also NVLink switches, CX, DPUs, ConnectX, EPU for East and West, BlueField DPUs for North and South and Data and Storage processing to InfiniBand for supercomputing centers to Ethernet, which is a brand new product for us.

Jensen Huang: Yeah, Aaron, thanks. The Blackwell RAC system is designed and architected as a RAC, but it's sold in this aggregated system components. We don't sell the whole RAC. And the reason for that is because everybody's RAC is a little different, surprisingly. You know, some of them are OCP standards; some of them are not; some of them are enterprise. And the power limits for everybody could be a little different; choice of CDUs, the choice of power bus bars, the configuration and integration into people's data centers, all different. And so the way we designed it, we architected the whole RAC.

Speaker Change #122: Yeah, everything's the black-wall rack system. It's designed and architected as a rack, but it's sold in this aggregated system components.

Speaker Change #123: We Don't Sell The Whole Rack.

Speaker Change #124: And the reason for that is because everybody's racks a little different, surprisingly.

Jensen Huang: We just dwell on its way to becoming a multi-billion dollar business to bring AI to Ethernet. And so the fact that we have access to all of this, we have one architectural stack as you know, it allows us to introduce new capabilities to the market as we complete it. Otherwise, what happens is you ship these parts, you go find customers to sell it to, and then you've got to build somebody's got to build up an AI factory.

Speaker Change #125: You know, some of them are OCP standards, some of them are not, some of them are enterprise.

Speaker Change #125: and the power limits for everybody could be a little different choice of CDUs, the choice of power bus bars, the configuration and integration into people's data centers, all different.

Jensen Huang: The software is going to work perfectly across the whole RAC. And then we provide the system components. Like, for example, the CPU and GPU compute board is then integrated into an MGX. It's a modular system architecture. MGX is completely ingenious. And we have MGX ODMs and integrators and OEMs all over the planet. And so just about any configuration you would like, where you would like that 3,000 pound RAC to be delivered, it has to be integrated and assembled close to the data center because it's fairly heavy. And so everything from the supply chain, from the moment that we ship the GPU CPUs, the switches, the next, from that point forward, the integration is done quite close to the location of the CSPs and the locations of the data centers.

Speaker Change #125: And so the way we designed it, we architected the whole rack, the software is going to work perfectly across the whole rack.

Jensen Huang: And the AI factory's got a mountain of software. And so it's not about, it's not about who integrates it. We love the fact that our supply chain is disintegrated. And in the sense that we could service, you know, Quanta, Foxconn, HP, Dell, Lenovo, SuperMicro, we used to be able to service ZT. They were recently purchased and so on and so forth. And so the number of ecosystem partners that we have, Gigabyte, Assutes, the number of ecosystem partners that we have that allows us to allow us them to take our architecture, which all works, but integrated in a bespoke way into all of the world's cloud service providers enterprise data centers.

Speaker Change #125: and then we provide the system components. Like for example, the CPU and GPU compute board is then integrated into an MGX. It's a modular...

Speaker Change #125: System Architecture, NGX is completely ingenious.

Speaker Change #125: And, uh, we have MGX.

Speaker Change #125: O-D-M's and integrators and O-M's all over the planet. So just about any configuration you would like, where you would like that 3,000 pound rack to be delivered. You know, it's got to be close to it.

Speaker Change #125: It has to be integrated and assembled close to the data center because it's fairly heavy.

Speaker Change #125: And so everything from the supply chain, from the moment that we ship the GPU CPUs, the switches, the next.

Jensen Huang: The scale and reach necessary from our ODMs and our integrators integrator supply chain is vast and gigantic because the world is huge. And so that part we don't we don't want to do and we're not good at doing. And but we know how to design the infrastructure provided the way that customers would like it and let the ecosystem integrate it. So anyways, that's the reason why.

Speaker Change #125: From that point forward, the integration is done quite close to the location of the CSPs and the locations of the data centers. And so you can imagine how many data centers in the world there are and how many logistics hubs we've scaled out to.

Jensen Huang: And so you can imagine how many data centers in the world there are and how many logistics hubs we've scaled out to with our ODM partners.

Jensen Huang: And so I think because we show it as one RAC and because it's always rendered that way and shown that way, we might have left the impression that we're doing the integration. Our customers hate that we do integration. The supply chain hates us doing integration. They want to do the integration. That's their value added. There's a final design in, if you will, you know, it's not quite as simple as shimmy into a data center, but the design fit in is really complicated. And so the installation, the design fit in, the installation, the bring up, the repair and replace, that entire cycle is done all over the world.

Speaker Change #126: With our O-DM partners. And so, I think, because we, we show it as one rack, and because it's always, you know, rendered that way, and, and shown that way, we, we might have left the impression that we're doing the integration. Our customers hate that we do integration.

Aaron Rakers: and your final question comes from the line of Aaron Rakers with Wells Fargo. Your line is open. Yes, thanks for taking the question. I wanted to go back into the Blackwell products cycle. One of the questions that we tend to get asked is, how do you see the RAC scale system mix dynamic? As you think about leveraging NV Lake, you think about GB, you know, NVL 72, and how that go to market dynamic looks, you know, as far as the Blackwell products cycle. I guess I put this simply, how do you see that mix of RAC scale systems as we start to think about the Blackwell cycle playing out?

Speaker Change #126: The supply chain hates us doing integration.

Speaker Change #126: They want to do the integration. That's their value added. There's a final design in, if you will, you know, it's not quite as simple as shimmy into a data center. But that design fit in is really complicated.

Speaker Change #126: And so, the design fit in, the installation, the bring up, the repair and replace, that entire cycle is done all over the world.

Jensen Huang: Yeah, Aaron, thanks. The Blackwell RAC system, it's designed and architected as a RAC, but it's sold in a disaggregated system components. We don't sell the whole RAC. And the reason for that is because everybody's RAC is a little different, surprisingly. Some of them are OCP standards, some of them are not, some of them are enterprise, and the power limits for everybody could be a little different choice of CDUs, the choice of power bus bars, the configuration and integration into people's data centers, all different.

Jensen Huang: We have a sprawling network of ODM and OEM partners that does this incredibly well. So, integration is not the reason why we're doing RACs. It's the anti-reason of doing it. The way we don't want to be an integrator; we want to be a technology provider.

Speaker Change #126: We have a sprawling network of ODM and OEM partners that does this incredibly well.

Speaker Change #127: So, integration is not the reason why we're doing racks, it's the anti-reason of doing it. The way we don't want to be an integrator, we want to be a technology provider.

Jensen Huang: And I will now turn the call back over to Jensen Huang for closing remarks. Thank you. Let me make a couple more, make a couple of comments that I made earlier again, that data centers worldwide are in full steam to modernize the entire computing stack with accelerated computing and generative AI. Hopper demand remains strong, and the anticipation for black oil is incredible.

Speaker Change #127: and I will now turn the call back over to Jensen Huang for closing remarks.

Jensen Huang: Thank you, let me make a couple more comments that I made it earlier again, that data center worldwide are in full steam to modernize the entire computing stack with accelerated computing and gender to BI.

Jensen Huang: And so the way we designed it, we architected the whole RAC. The software is going to work perfectly across the whole RAC. And then we provide the system components. Like for example, the CPU and GPU compute board is then integrated into an MGX. It's a modular system architecture. MGX is completely ingenious. And we have MGX ODMs and integrators and OEMs all over the planet. And so just about any configuration you would like, where you would like that 3000 pound RAC to be delivered, it has to be integrated and assembled close to the data center because it's fairly heavy.

Jensen Huang: Let me highlight the top five things, the top five things of our company. Accelerated computing has reached the tipping point; CPU scaling slows, developers must accelerate everything possible. Accelerated computing starts with CUDAX libraries; new libraries open new markets for Nvidia. We released many new libraries, including CUDAX, celebrated Polars, Pandas, and Spark, the leading data science and data processing libraries, CUVS for vector databases. This is incredibly hot right now. Aerial and Chionna for 5G wireless base station, a whole suite of a whole world of data centers that we can go into now. Parabrics for gene sequencing and AlphaFold 2 for protein structure prediction is now CUDAX already.

Speaker Change #129: Hopregamand remains strong in the anticipation for Blackwellers Infredible.

Speaker Change #130: Let me highlight the top five things.

Speaker Change #130: The top five things of our company, accelerated computing has reached a tipping point, CPU scaling slows, developers must accelerate everything possible. Accelerating computing starts with CUDAX libraries.

Speaker Change #130: New Libraries.

Speaker Change #130: Open New Markets for Nvidia.

Speaker Change #130: We released many new libraries, including Codex-Albert Polars, Pandas, and Spark, the leading data science and data processing libraries.

Tov-S: Tov-S, Thanks for watching!

Speaker Change #132: From VectorDataBases, this is incredibly hot right now.

Speaker Change #133: Aerial, and Giona for 5G Wireless Vastation, a whole...

Speaker Change #133: Sweet of a whole world of data centers that we can go into now.

Speaker Change #134: Parabrix for Gene Sequencing, and AlphaFull2 for Protein Structure Prediction, is now KuTX alright.

Jensen Huang: And so everything from the supply chain, from the moment that we ship the GPU CPUs, the switches, the next, from that point forward, the integration is done quite close to the location of the CSPs and the locations of the data centers. And so you can imagine how many data centers in the world there are and how many logistics hubs we've scaled out to with our ODM partners.

Jensen Huang: We are at the beginning of our journey to modernize a trillion dollars' worth of data centers from general purpose computing to accelerator computing. That's number one. Number two, black oil is a step function leap over hopper. Black Well is an AI infrastructure platform, not just the GPU. Also happens to be in the name of our GPU, but it's an AI infrastructure platform. As we reveal more of Black Well and sample systems to our partners and customers, the extent of Black Well's leap becomes clear. The black CPU, the black well dual GPU and a co-auth package, connect X-DPU for e-swash traffic, blue field GPU for north north south and storage traffic, NB link switch for all GPU communications, and quantum and spectrum x for both Infiniband Ethernet can support the massive burst traffic of AI.

Speaker Change #135: We are at the beginning of our journey to modernize a trillion dollars worth of data centers from general purpose computing to accelerate computing.

Speaker Change #135: Let's number one, number two. Blackwell is a step-function leap over hopper. Blackwell is an AI infrastructure platform, not just the GPU.

Speaker Change #135: Also happens to be in the name of our GPU, but it's an AI infrastructure platform, as we reveal more of Blackwell.

Jensen Huang: I think because we show it as one RAC, and because it's always rendered that way, and shown that way, we might have left the impression that we're doing the integration. Our customers hate that we do integration. The supply chain hates us doing integration. They want to do the integration. That's their value added. There's a final design design in, if you will, you know, it's not quite as simple as shimmy into a data center, but that the design fit in is really complicated.

Speaker Change #135: and sample systems to our partners and customers, the extent of black-less leak becomes clear.

Speaker Change #135: The Blackwell Vision took nearly five years, and seven, one of a kind ships to realize. The great CPU?

Speaker Change #135: The Blackwell Dual GPU and a Colossed Package

Speaker Change #135: ConnectXDPU for East West Traffic.

Speaker Change #135: BlueField-DPU for north-south and storage traffic, Envealing Switch.

Speaker Change #135: For Alt, All GPU Communications,

Speaker Change #135: and Quantum Inspector Max, but for both InfiniBand and Ethernet, can support the massive burst traffic of AI.

Jensen Huang: And so the install, the design fit in, the installation, the bring up, the repair, repair and replace, that entire cycle is done all over the world. And we have a sprawling network of ODM and OEM partners that does this incredibly well. So integration is not the reason why we're doing racks, it's the anti-reason of doing it. The way we don't want to be an integrator, we want to be a technology provider.

Jensen Huang: Black well AI factories are building size computers and video design and optimized the black well platform full stack end to end from chips systems networking even structured cables tower and cooling and mounds of software to make it fast for customers to build AI factories. These are very capital intensive infrastructures. Customers want to deploy it as soon as they get their hands on the equipment and deliver the best performance and TCO. Black well provides three to five times more AI throughput in a power-limited data center than Hopper.

Speaker Change #136: Blackwall AI factories are building size computers, Nvidia designed and optimized the Blackwall platform full stack, end-end, from chips, systems, networking.

Speaker Change #136: Even structured cables.

Speaker Change #136: Power and cooling, and Mounds of software to make it.

Speaker Change #136: Fast for customers to build AI factories. These are very capital intensive infrastructures. Customers want to deploy it as soon as they get their hands on the equipment. And deliver the best performance and TCL.

Jensen Huang: And I will now turn the call back over to Jensen Huang for closing remarks. Thank you, let me make a couple more, make a couple of comments that I made earlier again, that data center worldwide are in full steam to modernize the entire computing stack with accelerated computing and generative AI. Hopper demand remains strong and the anticipation for black oil is incredible.

Speaker Change #136: Blackwell provides 3-5 times more AI throughput in a power limit of data centered in copper.

Jensen Huang: Webber. The third is NVLink; this is a very big deal with all GPUs. Switch is game-changing. The Blackwell system lets us connect 144 GPUs in 72 GB-200 packages into one NVLink domain, with an aggregate NVLink bandwidth of 259 terabytes per second and one rack. Just put that in perspective, that's about 10 times higher than Hopper, 259 terabytes per second. Kind of makes sense because you need to boost the training of multi-trillion parameter models on trillions of tokens, and so that natural amount of data needs to be moved around from GPU to GPU. For inference, NVLink is vital for low latency, high throughput, large language model, token generation.

Speaker Change #137: The third ZenB link, this is a very big deal with its all GPUs switch, is game-changing. The Blackwell system lets us connect 144 GPUs.

Speaker Change #137: In 72 GB200 packages into 1 MBD, with an aggregate, aggregate, and VLINK bandwidth of 2569 terabytes per second in 1 rack. Just put that in perspective, that's about 10 times higher than Hopper.

Jensen Huang: Let me highlight the top five things, the top five things of our company, accelerated computing has reached the tipping point, CPU scaling slows, developers must accelerate everything possible. Accelerating computing starts with CUDAX libraries, new libraries, open new markets for NVIDIA. We released many new libraries including CUDAX celebrated Polars, pandas, and Spark, the leading data science and data processing libraries, QVS for vector databases, this is incredibly hot right now, Ariel and Chiona for 5G wireless base station, a whole suite of a whole world of data centers that we can go into now, Parabrics for gene sequencing, and AlphaFull 2 for protein structure prediction is now CUDAX celebrated.

Speaker Change #137: 259 TeraBites per second.

Speaker Change #138: Kind of makes sense because you need to boost the training of multi-

Speaker Change #138: Trillion Parameter Models on Trillions of...

Speaker Change #138: Tokens, Thanks for watching!

Speaker Change #138: And so that natural amount of data needs to be moved around from QQ to QQ.

Speaker Change #138: For inference, MBLink is vital for low-lane C-high throughput, large-language model, token generation. We now have three networking platforms.

Jensen Huang: We now have three networking platforms: NVLink for GPU scale-up, quantum, InfiniBand for supercomputing, and dedicated AI factories, and Spectrum X for AI on Ethernet. NVIDIA's networking footprint is much bigger than before. Viewer to AI momentum is accelerating. Generative AI frontier model makers are racing to scale to the next AI plateau to increase model safety and IQ. We're also scaling to understand more modalities from text, images, and video to 3D, physics, chemistry, and biology. Chatbots, coding AI, and image generators are growing fast, but it's just the tip of the iceberg. Internet services are deploying generative AI for large scale recommenders, add targeting and search systems. AI startups are consuming tens of billions of dollars yearly of CSP cloud capacity, and countries are recognizing the importance of AI and investing in sovereign AI infrastructure.

Speaker Change #138: NVLink for GPU scale up.

Speaker Change #138: Quantum InfiniBand for supercomputing and dedicated AI factories, and Spectrum X for AI on Ethernet. And we guess networking footprint is much bigger than before.

Jensen Huang: We are at the beginning of our journey to modernize a trillion dollars worth of data centers from general purpose computing to a solidarity computing, that's number one, number two. Blackwell is a step function leap over Hopper, Blackwell is an AI infrastructure platform, not just the GPU. It also happens to be an email of our GPU, but it's an AI infrastructure platform. As we reveal more of Blackwell and sample systems to our partners and customers, the extent of Blackwell's leap becomes clear.

Speaker Change #138: Viewer today, Momentum is accelerating, Junitive AI Frontier model makers are racing to scale to the next AI plateau to increase model safety and IQ.

Speaker Change #138: We're also scaling to understand more modalities from text, images, and video to 3D, physics, chemistry, and biology.

Speaker Change #138: ChatBots, Coding AI, and image generators are growing fast, but it's just a tip of the iceberg. Internet services are deploying genitive AI for large-scale recommenders, ad targeting and search systems.

Jensen Huang: The Blackwell vision took nearly five years and seven, one of a kind ships to realize. The gray CPU, the Blackwell dual GPU in a co-off package, ConnectX DPU for e-swash traffic, Bluefield DPU for north south and storage traffic, NB link switch for all GPU communications, and Quantum and Spectrum X for both Infiniband Ethernet can support the massive burst traffic of AI. Blackwell AI factories are building size computers, Nvidia design and optimized the Blackwell platform full stack, end to end, from chips, systems, networking, even structured cables, power and cooling, and mountains of software to make it fast for customers to build AI factories.

Speaker Change #138: AI startups are consuming tens of billions of dollars, yearly of CSP Cloud Capacity, and countries are recognizing the importance of AI, and investing in sovereign AI infrastructure.

Jensen Huang: NVIDIA Omniverse is opening up the next era of AI, general robotics. And now the enterprise AI wave has started, and we're poised to help companies transform their businesses. The NVIDIA AI Enterprise platform consists of NeMo, NIMS, NIM Agent Blueprints, and AI Foundry that our ecosystem partners, the world-leading IT companies, used to help customers customize AI models and build the spoke AI applications. Enterprises can then deploy on NVIDIA AI Enterprise Runtime, and at $4,500 per GPU per year, NVIDIA AI Enterprise is an exceptional value for deploying AI anywhere, and for NVIDIA's software TAMM can be significant as the CUDA compatible GPU installed base grows from millions to tens of millions.

Speaker Change #138: And Nvidia AI, Nvidia Omniverse is opening up the next era of AI, General Robotics.

Speaker Change #139: And now, the Enterprise AI Weave has started, and we're pleased to help companies transform their businesses. The Nvidia AI Enterprise Platform consists.

Speaker Change #139: of NeMo.

Speaker Change #139: LEMS.

Speaker Change #139: Nam Agent BluePrinz, and AI Foundry

Speaker Change #139: That our ecosystem partners, the world-leading IT companies.

Speaker Change #139: Use to help customer companies customize AI models and build the spoke AI applications. Enterprises can then deploy on Nvidia AI Enterprise runtime.

Jensen Huang: These are very capital intensive infrastructures, customers want to deploy it as soon as they get their hands on the equipment and deliver the best performance and TCO. Blackwell provides three to five times more AI throughput in a power limited data center than Hopper. Server. The third is NVLink. This is a very big deal with all GPUs switch is game-changing. The Blackwell system lets us connect 144 GPUs in 72 GB-200 packages into one NVLink domain, with an aggregate NVLink bandwidth of 259 terabytes per second and one rack.

Speaker Change #139: and at $4500 per GPU per year, Nvidia Enterprise is an exceptional value for deploying AI anywhere.

Speaker Change #140: and for Nvidia's software tam can be significant as the CUDA compatible GPU install base grows from millions to tens of millions. And as Colette mentioned, Nvidia software will exit the year at a $2 billion run rate.

Colette Kress: And as Collette mentioned, NVIDIA software will exit the year at a $2 billion run rate.

Jensen Huang: Thank you all for joining us today.

Colette: Thank you all for joining us today.

Speaker Change #141: And Ladies and gentlemen, this concludes today's call and we thank you for your participation. You may now disconnect.

Jensen Huang: Just put the 9 terabytes per second. It kind of makes sense because you need to boost the training of multi-trillion parameter models on trillions of tokens. And so that natural amount of data needs to be moved around from GPU to GPU. For inference, NVLink is vital for low latency, high throughput, large language model, token generation. We now have three networking platforms. NVLink for GPU scale-up, quantum infiniBand for supercomputing and dedicated AI factories, and SpectrumX for AI on Ethernet.

Jensen Huang: NVGIS networking footprint is much bigger than before. GeoGVAI momentum is accelerating. Generative AI frontier model makers are racing to scale to the next AI plateau to increase model safety in IQ. We're also scaling to understand more modalities from text, images, and video to 3D, physics, chemistry, and biology. Chatbots, coding AI's and image generators are growing fast, but it's just a tip of the iceberg. Internet services are deploying generative AI for large scale recommenders, add targeting and search systems. AI startups are consuming tens of billions of dollars yearly of CSP's cloud capacity, and countries are recognizing the importance of AI and investing in sovereign AI infrastructure.

Speaker Change #142: Thanks for watching!

Jensen Huang: And NVGIS and NVGIS omniverse is opening up the next era of AI, general robotics. And now the enterprise AI wave has started and we're poised to help companies transform their businesses. The NVGIS enterprise platform consists of Nemo, NIMs, NIM agent Blueprints, and AI Foundry that our ecosystem partners the world leading IT companies used to help customers customize AI models and build the spoke AI applications. Enterprises can then deploy on NVGIS enterprise runtime and at $4,500 per GPU per year NVGIS enterprise is an exceptional value for deploying AI anywhere and for NVGIS software tam can be significant as the CUDA compatible GPU install base grows from millions to tens of millions and as Collette mentioned, NVGIS software will exit the year at a $2 billion run rate.

Jensen Huang: Thank you all for joining us today.