From Anthropic to Waymo: Kleiner Perkins Bets on AI's Next Act

Kleiner Perkins partner Ilya Fushman sees AI as being in the early stages of reshaping the economy, highlighting opportunities in frontier models, consumer agents, autonomous vehicles and robotics. He describes physical AI as a potential source of the next trillion-dollar companies and says he is increasingly optimistic about IPO and M&A markets. No specific company results or market moves were reported.

Analysis

The investable signal is not the breadth of AI use cases; it is the prospect of liquidity returning to venture portfolios. A durable IPO/M&A reopening could improve fundraising and capital recycling for venture firms, but also create a supply overhang: new listings compete with public AI names for risk capital and provide harder valuation marks for late-stage holdings. Treat optimism from a venture investor as a sentiment indicator, not evidence that exits or deal volumes have turned.

Near term, the main risk is narrative spillover into public software valuations. Agents may compress pricing power for some application vendors, while gains accrue to infrastructure providers and firms that can show measurable labor or throughput savings. Physical AI is a longer-dated thesis: deployment depends on hardware costs, reliability, safety approval and customer payback, so model progress alone does not establish attractive returns for robotics or autonomous-vehicle businesses.

Over 1–3 months, verify IPO filings, completed M&A and secondary-market discounts rather than relying on private-company growth claims. Over 6–18 months, watch deployment economics and customer adoption for physical AI. The contrarian risk is that a perceived IPO reopening could be premature; conversely, successful exits could validate private marks but draw capital away from already crowded public AI trades. No broad directional trade is justified from this interview alone.

Key Decisions for Investors

• Avoid adding broad AI exposure solely on the interview’s optimism. Track actual IPO filings, deal closings and private-market secondary discounts as confirmation of an exit-cycle turn.
• Build a conditional relative-value watchlist: favor AI infrastructure and automation exposure over application software only where earnings evidence shows durable demand or measurable productivity gains; do not initiate a sector pair without valuation and estimate data.
• For physical AI, wait for evidence on unit economics, deployment volumes, reliability and regulatory approvals before treating robotics or autonomous-vehicle enthusiasm as a near-term earnings catalyst.
• Falsify the IPO/M&A recovery thesis if announced deals fail to close, listings are repeatedly postponed, or secondary discounts widen; revisit the physical-AI thesis if customer deployments and payback periods improve materially over the next 6–18 months.