OpenAI closed a $110 billion private round at a $730 billion pre-money valuation, dwarfing every previous private financing on record. The named investors were Amazon ($50B, including AWS compute commitments), Nvidia ($30B) and SoftBank ($30B), according to TechCrunch.
The money is pointed almost entirely at infrastructure: roughly 2GW of AWS Trainium capacity for training and 3GW of Nvidia capacity for inference. As OpenAI put it, "Frontier AI moves from research into daily use at global scale." At the time of the first close, the round was explicitly still open for more.
What is unusual here is not only the size but the composition. Two of the three anchors — Amazon and Nvidia — are also the suppliers the capital will be spent with. Amazon's cheque comes bundled with AWS compute commitments; Nvidia's arrives alongside the inference capacity OpenAI is buying. The round is as much a supply agreement denominated in equity as it is a financing, which is why the gigawatt figures are quoted next to the dollar figures rather than buried in a footnote.
It is also a round that only a handful of balance sheets on earth could have anchored. At $110B the cheque sizes stop resembling venture capital and start resembling industrial project finance: three strategic investors, each committing sums normally associated with building refineries or fabs, against a company whose product did not exist a decade ago. The "still open for more" line at the first close was not boilerplate — it signalled that demand for allocation exceeded what had already been papered.
The round did take more. OpenAI closed on March 30 at $122 billion, at an $852 billion post-money valuation, still anchored by Amazon, Nvidia and SoftBank.
The two sets of numbers are consistent rather than contradictory, and the arithmetic is worth spelling out because the headline figures were reported in different terms: a $730B pre-money valuation plus $122B of new capital produces the $852B post-money figure. The first reports caught the round mid-raise at $110B; the final close added roughly $12B on the same terms.
One counterpoint arrived alongside the close: on annual recurring revenue, Anthropic passed OpenAI. The largest private raise in history and a loss of the revenue lead landed in the same window, which is a reminder that the round buys capacity, not customers. Capital at this scale is a bet on demand that has not been booked yet — the gigawatts are being contracted years ahead of the revenue meant to fill them.
Three things follow from a raise structured this way.
First, concentration. When the same firms are both the funders and the vendors, the valuation and the compute contracts move together, and a slowdown at one end is felt immediately at the other. Second, the pre-money/post-money gap is going to keep generating confused headlines: $730B, $110B, $122B and $852B are all correct numbers describing one round at different moments, and they will be quoted against each other for months. Third, the ARR crossover. Revenue leadership and capital leadership have now separated, and the gigawatt commitments here only make sense if that separation is temporary.
For now, the number to hold onto is the last one: $122 billion raised, $852 billion post-money, and several gigawatts of contracted capacity that has to be filled with paying work.
TechCrunch — OpenAI raises $110B in one of the largest private funding rounds in historyTweet — final close at $122BTweet — ARR chart