OpenAI has raised $110 billion in private funding, the company announced Friday morning, opening one of the largest private rounds in history. The deal puts the AI pioneer at a $730 billion pre-money valuation and marks a sharp escalation in the AI infrastructure arms race. The money is not just capital — it is a chain of compute commitments that lock OpenAI, Amazon and Nvidia into each other for years. Here is how the deal is actually built, and why the structure matters more than the headline number.
Key takeaways
- The round totals $110 billion at a $730 billion pre-money valuation, and it remains open, with OpenAI expecting more investors to join. - Amazon wrote the largest check at $50 billion, of which $35 billion is contingent on OpenAI reaching AGI or completing an IPO — an unusual, milestone-gated structure. - The deal expands OpenAI's AWS commitment from $38 billion to $138 billion, with at least 2 gigawatts of Trainium compute, and moves models onto a new stateful runtime on Amazon Bedrock. - Nvidia committed $30 billion tied to 3 gigawatts of dedicated inference capacity and 2 gigawatts of Vera Rubin training; SoftBank added $30 billion. - For GEO, the takeaway is durability: the engine most people query is being wired into the industry's deepest compute pipeline, so optimizing for ChatGPT visibility is a long-term bet, not a fad.
The funding breakdown
The $110 billion rests on three big commitments, and each one is really a compute deal wearing an investment's clothes.
Amazon's $50 billion is the largest single check in the round — a bet on OpenAI's future that also secures Amazon's own. As part of the deal, OpenAI will build a new stateful runtime environment for its models to run on Amazon Bedrock, and the two expand OpenAI's previously announced AWS commitment from $38 billion to $138 billion in total compute services, including at least 2 gigawatts of AWS Trainium. The most striking detail: $35 billion of Amazon's investment carries performance conditions. Per The Information, that portion arrives only when specific milestones are met — namely, if OpenAI achieves artificial general intelligence or completes its IPO.
Nvidia committed $30 billion, deepening the symbiosis between the AI chip leader and its largest customer. The investment is tied to OpenAI consuming 3 gigawatts of dedicated inference capacity and 2 gigawatts of Vera Rubin system training. CEO Jensen Huang, who earlier denied reports that Nvidia was pulling back, said in January: "We're going to invest a huge amount of money. I believe in OpenAI. Their work is amazing." SoftBank rounded out the announced commitments with $30 billion, and the round stays open for more.
Why the structure matters
Read past the number and the shape of the deal tells the real story. This is not cash handed over for equity; it is a set of interlocking compute obligations. OpenAI gets capital and guaranteed capacity. Amazon and Nvidia get a locked-in, multi-gigawatt customer. The milestone-gated $35 billion effectively ties a large slice of the raise to outcomes — AGI or an IPO — that are years out and far from certain.
