The Chip King Who Almost Owned the Labs
Nvidia built the infrastructure the AI economy runs on. Its CEO says missing the OpenAI and Anthropic seed rounds was his biggest mistake — and he is now spending $40 billion to fix it.
Nvidia CEO Jensen Huang publicly regretted missing the seed rounds of OpenAI and Anthropic, revealing that a $5-10 billion compute threshold—rather than funding gaps—locked traditional venture capital out of transformative AI before the field existed. Google and Amazon captured early AI lab relationships by structuring investments as compute credits tied to their own custom silicon (TPUs and Trainium), creating hardware lock-in that persists today. Nvidia, which built the GPU infrastructure these labs depend on, is now buying equity at valuations set by competitors—investing $40 billion across OpenAI and Anthropic—though it remains unclear whether this ownership translates to actual influence over chip selection.
- •The AI lab investment landscape was defined by compute access thresholds of $5-10 billion, which excluded traditional VCs before the field existed as an investment category
- •Cloud providers structured AI lab investments as compute credits, creating hardware dependency: Anthropic runs substantially on Google TPUs and Amazon Trainium rather than Nvidia GPUs
- •Nvidia, despite powering the AI boom with its GPU infrastructure, entered AI lab equity investing late and at valuations determined by its cloud competitors
