OpenAI: Capital, Compute, and the Valuation Story. Part 1.

Part 1. Follow the money: who funded OpenAI, who promised what, and what is actually committed?

OpenAI is often discussed in numbers so large they start to blur together and look like one giant pile of capital.

There is Microsoft’s investment, giant funding rounds, a half-trillion-dollar infrastructure project, huge long-term commitments linked to cloud and compute, and strategic partnerships that define who funds what, who controls what, and who gets paid along the way. These figures are often mentioned together, but they describe different types of commitments. To understand OpenAI’s financial position, it helps to categorise them first.

The first categoory is money raised by OpenAI.
The second is infrastructure commitments around OpenAI.
The third is OpenAI’s own long-term obligations and economic-sharing arrangements.

Read separately, they reveal a company with unusually strong backing and, just as importantly, very heavy external dependencies.

1. Direct capital into OpenAI

The most straightforward part of the story is OpenAI’s funding history.

Microsoft’s began backing OpenAI in 2019 with an initial $1 billion investment, which grew in 2021 and 2023. Reports indicate that Microsoft has invested a total of$13 billion since 2019.

Then came a series of very large public funding rounds:

1. In October 2024, OpenAI announced it raised $6.6 billion at a valuation of $157 billion after the investment.
2. In March 2025, it disclosed a much larger round: $40 billion at a $300 billion valuation after the investment. That round was led by SoftBank, which contributed $30 billion, along with $10 billion from a group of investors that included Microsoft, Coatue, Altimeter, and Thrive.
3. In March 2026, OpenAI announced am even larger round: $122 billion in committed capital at an $852 billion valuation after the investment. This round was mainly backed by Amazon, NVIDIA, and SoftBank, with continued support from Microsoft. Public reports indicated up to $50 billion from Amazon (with $35 billion of that contingent on OpenAI going public or achieving AGI), $30 billion from NVIDIA, and $30 billion from SoftBank. Plus, over $3 billion came from individual investors through bank channels .

So, the funding story is fairly clear: Microsoft backed OpenAI early and continued its backing as the company scaled. OpenAI then completed three giant public rounds in 2024, 2025, and 2026, each at a much higher valuation than the previous one.

Table 1. Funding history of OpenAI

2. Infrastructure built around OpenAI

In January 2025, OpenAI announced The Stargate Project, describing it as a new initiative that will invest $500 billion over four years in AI infrastructure for OpenAI in the United States. OpenAI stated the project would start deploying $100 billion immediately, and named SoftBank, OpenAI, Oracle, and MGX as the initial equity funders. SoftBank would take financial responsibility while OpenAI would manage operations. This makes Stargate part of OpenAI’s economic story, even if it isn’t reflected in their balance sheet like a traditional funding round.

Later updates clarified the scale.

In July 2025, OpenAI reported that Oracle and OpenAI had reached an agreement to develop an additional 4.5 gigawatts of Stargate capacity in the U.S., bringing Stargate’s total development to over 5 gigawatts of AI data centre capacity and over 2 million chips in conjunction with the Abilene, Texas site.
In September 2025, SoftBank, OpenAI, and Oracle reported that Stargate had nearly 7 gigawatts of planned capacity and over $400 billion in investment over the next three years.

In April 2026, OpenAI stepped back from plans to rent compute directly at the planned “Stargate Norway” site, while Microsoft took the spare capacity instead. OpenAI mentioned it was discussing accessing that same compute through Microsoft rather than renting it directly. Around this time, OpenAI also paused a similar Stargate-branded project in the U.K.

So, Stargate now looks more partner-mediated and more flexible than the original plan from January 2025 implied, though some infrastructure may still get built. Microsoft’s amended agreement with OpenAI, announced later in April 2026, also supports this idea: Microsoft remains OpenAI’s primary cloud partner, while OpenAI now has more flexibility to work across other cloud providers as well.

The infrastructure story around OpenAI is no longer just about a single giant project in the abstract. It is about how OpenAI secures compute in practice: sometimes through dedicated buildouts, sometimes through partner-backed capacity, and increasingly through a mix of both.

3. What OpenAI is committed to pay or share

The third layer focuses on what OpenAI is committed to spending or sharing.

In April 2026, OpenAI’s commitment to use at least $250 billion in Azure services by 2032 remained in place. Also, Microsoft would received a guaranteed 20% cut of OpenAI revenue until 2030, subject to an undisclosed cap, and Microsoft would continue as OpenAI’s primary cloud partner with a license to OpenAI intellectual property through 2032.

These long-term commercial agreements influence how OpenAI scales, how much it is committed to spend, and how much of that growth is already accounted for.

That is the basic structure. OpenAI has raised enormous amounts of capital, but not all of the large figures around it belong to the same category. Some money went directly to the company, some is tied to infrastructure being build around it, and some reflect long-term commitments that OpenAI will need to fulfill over time.

Separating these aspects does not tell us whether the economics work. But it does make the next question much clearer: how much compute is actually being built, who controls it, and how heavily OpenAI’s growth depends on sustaining that expansion. That is where Part 2 begins.