The world's most valuable AI company just lost its visionary founder, and now the question is whether his $50 trillion bet on physical AI survives him.
The Summary
- Nvidia interim CEO Colette Kress held a town hall Wednesday reiterating confidence after Jensen Huang's sudden departure, while NVentures contacted at least three robotics startups last Thursday to reassure them of continued support
- Kress is expected to pursue deals more slowly than Huang and has previously pushed back on investments he championed, raising questions about Nvidia's investment strategy
- Huang backed more than 35 robotics and physical AI companies and positioned physical AI as the industry's "first opportunity to address a $50 trillion industry"
The Signal
When a founder leaves, the first question isn't what changes. It's what dies. Jensen Huang didn't just run Nvidia, he shaped an entire category. He coined "physical AI." He put humanoid robots onstage at developer conferences. He framed robotics as a $50 trillion opportunity that manufacturing, logistics, and human labor had somehow missed. Now he's gone, and the robotics startups he backed are getting reassurance calls.
The outreach started last Thursday, right after Huang resigned. At least three portfolio companies heard the same message: business as usual. But usual for whom? Huang moved fast, saw robotics as existential, and reportedly overrode internal skeptics. Kress, now interim CEO, is known to pursue deals more slowly and has pushed back on investments Huang championed. That's not a leadership transition. That's a philosophy shift.
"Physical AI" wasn't just marketing. It was Huang's bet that the same chips powering language models could power machines that move, build, and work.
The scale of that bet matters. More than 35 robotics and physical AI companies took Nvidia's money under Huang's watch. Some got direct investments. Others got NVentures backing. All got something harder to quantify: validation from the man who turned GPUs into the infrastructure layer of the AI economy. When Huang said robotics would industrialize the physical world, capital followed. When he appeared onstage with humanoid robots, competitors took note.
Now those startups are asking what Kress thinks of physical AI. And the answer isn't clear yet. She held a town hall Wednesday and expressed confidence in the company's direction. But confidence in what? Nvidia's core business, selling chips to hyperscalers training frontier models, doesn't need robotics to work. Physical AI was Huang's expansion play. It was the pitch that Nvidia wouldn't just power cloud AI, but edge AI, embodied AI, the AI that assembles your car and stocks your shelves.
Here's the tell: startups don't get reassurance calls unless someone thinks they need reassurance. If Kress planned to double down on Huang's robotics vision, she'd announce it. Instead, NVentures is quietly telling portfolio companies the support continues. That's damage control, not conviction.
The Implication
Watch what Nvidia does in the next 90 days. If Kress shows up at a robotics conference or announces a new physical AI initiative, Huang's vision survives. If she stays quiet and focuses on datacenter chips and AI training infrastructure, the robotics bet gets quietly deprioritized. For the 35+ startups in Nvidia's portfolio, that means hedging. Find other chip partners. Prove your robots work on non-Nvidia silicon. The reassurance calls bought you time, not certainty.
For everyone else building in the agent economy, this is a reminder that bets on embodied AI are still speculative. The infrastructure is there. The models are improving. But the capital and conviction behind physical AI just took a hit. If you're building agents that live in software, you're fine. If you're building agents that need to pick things up and move them around, your next funding round just got harder.