technologyScore 45/100Watch

NVIDIA scaling competitive open-source physical AI models (Nemotron, Cosmos, Root, Dream Zero) as existential hedge against closed-source model dominance

Andrew Kang· Mechanism Capital· Robotics· —· pub 2026-07-29· about NVIDIA, Anthropic (NVDA)
Nvidia is also a very big player in open-source model development... It is an almost an existential threat for closed-source models to win because, as you saw with Anthropic starting to train on Google TPUs, if companies decide to optimize for and train on other hardware, Nvidia starts to lose their business... And so that's why they're putting so much effort into developing their own open-source models like Nemotron, like the autonomous vehicle models... that's something that I feel like can't be understated because it's if you're building just physical AI models, you have to think about I'm competing with one of the best AI companies in the world.

Why it matters

NVIDIA is proactively developing open-source physical AI models to prevent hardware lock-in by closed-source competitors. This diversification of AI model control limits NVIDIA's downside if training migrates to non-NVIDIA hardware (e.g., TPUs).

Investment implication

NVIDIA's vertical integration into physical AI model development reduces concentration risk in its GPU business. However, it signals intensifying competition for compute from large, well-resourced players like Google, Anthropic, and OpenAI.

Source

What's Actually Holding Humanoid Robots Back Isn't the AI | Andrew Kang, RoboStrategy (YouTube)
← All signals