Nvidia confirmed Wednesday it will acquire Hugging Face for $12.9 billion, the chip maker’s largest acquisition ever and a clear signal that open-source AI infrastructure has become too valuable to leave independent.
The deal gives Nvidia control of a platform hosting over 3 million AI models and used by more than 18 million developers. That’s not just a repository. It’s the central distribution hub for open-source AI, the place where researchers share models, developers find pre-trained tools, and the entire open ecosystem coordinates.
For Nvidia, this isn’t about killing open source. It’s about owning the highway. The company already dominates AI compute with its GPUs. Now it owns the platform where most of those models actually live and get shared. Nvidia can promote its own tools, steer developers toward its hardware, and make sure every popular open model runs best on its chips.
It’s a $12.9 billion bet that open source matters more than most people realize. While the conversation focuses on closed models from OpenAI and Anthropic, a huge portion of production AI runs on open models. Hugging Face is where those models come from.
The Hugging Face deal landed the same day two other massive funding rounds surfaced. Accel is reportedly leading a $1 billion round for Thinking Machines at a $40 billion valuation. The company’s already doing over $100 million in annual revenue, which puts it on track to join the small club of AI companies with real business models.
Meanwhile, Crusoe reportedly raised $3 billion at a $30 billion valuation after securing a $13 billion contract with Jane Street. That’s a data center company landing one of the largest infrastructure contracts in recent memory, a sign that AI compute demand isn’t just hype.
Put these three deals together and you’re looking at nearly $17 billion changing hands in a single day. The money’s real, the valuations are staggering, and the underlying message is clear: infrastructure is where the value is concentrating.
In a separate report, WIRED learned that OpenAI recently cut off its partnership with Cursor, the AI coding startup, despite internal estimates showing the relationship would generate more than $1 billion in annual revenue. The reason? SpaceX, controlled by Elon Musk, acquired Cursor.
OpenAI walked away from $1 billion to avoid doing business with Musk. That’s not a normal commercial decision. It’s a sign of just how toxic the relationship between OpenAI and its co-founder has become.
It also reveals something about Cursor’s scale. A billion dollars in annual revenue from a single partnership means Cursor was consuming an enormous amount of API capacity. That’s not a side project. That’s a major customer.
For OpenAI, losing Cursor to a Musk-controlled entity stings twice: once because of the revenue, and again because it strengthens a competitor. SpaceX now owns a tool that could help developers move away from OpenAI’s models entirely.
Nvidia’s Hugging Face acquisition reshapes the open-source landscape. The largest distributor of open models is now owned by the company that makes the chips those models run on. That vertical integration gives Nvidia enormous leverage.
The funding rounds for Thinking Machines and Crusoe show that investors are still willing to deploy billions into AI infrastructure, even at eye-watering valuations. These aren’t speculative bets on research labs. They’re bets on companies with contracts and revenue.
And OpenAI’s decision to sacrifice a billion-dollar customer rather than work with Musk shows that the personal conflicts in AI are affecting billion-dollar decisions. That’s unusual even for an industry this competitive.
The pattern across all three stories is consolidation. Nvidia’s buying distribution. Thinking Machines and Crusoe are raising enough capital to dominate their niches. OpenAI’s choosing strategic purity over revenue. The AI industry’s still growing fast, but the window for new players is closing.
One email at dawn. The five stories that mattered, with the bits removed and the meaning kept. Free, for now.