How OpenAI Was Funded: From Nonprofit Donations to Billions in Investments and Global Partnerships (2015–2025)

Graziano Stefanelli
2 days ago
6 min read

Nonprofit Beginnings (2015–2018)

OpenAI started in late 2015 as a nonprofit. This means it didn’t sell any part of itself and didn’t have owners or shares. It was built only to do research that could help everyone, not to make anyone rich.

The first funding—about $130 million—came as donations from well-known tech founders and investors.

Elon Musk pledged about $10 million, but he later stepped away and may not have paid all of it.
Sam Altman gave around $10 million and became a major leader at OpenAI.
Peter Thiel and Reid Hoffman each gave several million dollars, while other people like Jessica Livingston, Greg Brockman, and companies like AWS and Infosys added smaller amounts.
All this money went directly into hiring researchers, paying salaries, and buying computing hardware, not to profits.
There was no way for donors to get their money back or make money from OpenAI.

Because these were pure donations, OpenAI could move quickly and test out lots of different projects, like robotics, games, and language models. There was no need for “board approval” from investors, which let the team change direction easily. For example, they could shift focus from teaching robots to play games to building powerful text generators when that seemed most promising.

Capped-Profit Transition (2019)

By 2018, OpenAI realized it needed much more money and computing power than normal donations could provide, especially to compete with companies like Google and Microsoft.

In 2019, OpenAI created a new company (OpenAI LP) with a special rule:

Investors and employees could get paid back if OpenAI made money, but their profit was “capped”—they could only get up to 100 times what they put in.
If OpenAI ever made more than that, the extra would go back to the nonprofit, not the investors.
Khosla Ventures was the first big outside investor, putting in about $50 million, which gave them a right to a capped share of profits but not true “ownership.”
Early employees and founders also got these “profit units,” a special kind of promise for future rewards that kept everyone motivated.

This structure let OpenAI attract real investors and reward its team, but still kept control in the hands of the nonprofit board. The board could block any decision that went against OpenAI’s original mission. For example, if an investor pushed for risky or unsafe AI projects, the board could say no. This protected OpenAI’s mission, even as it brought in more money.

Microsoft’s Initial Investment (2019)

In July 2019, OpenAI got its biggest boost yet—$1 billion from Microsoft.

About $300 million was in Azure “credits,” meaning OpenAI could use Microsoft’s cloud computers for free, and the rest was in cash.
In exchange, Microsoft became OpenAI’s preferred partner for cloud computing and got to use OpenAI’s newest AI technology before anyone else.

This partnership helped both sides. OpenAI could suddenly afford to train much larger models, using massive numbers of GPUs that would have been out of reach otherwise. Microsoft, in turn, could build its own AI-powered products using OpenAI’s models, starting with Bing and later expanding to things like Copilot for coding.

Microsoft didn’t get “shares” or control, but it did get special deals:

Early access to new AI models
A tight technical relationship, with OpenAI staff working closely with Microsoft on shared projects

With this investment, OpenAI jumped ahead of almost every other AI lab except Google in terms of resources and capability.

Additional Venture Capital (2019–2022)

After Microsoft’s deal, other investment firms wanted to be part of OpenAI’s growth.

Thrive Capital, Founders Fund, and several others joined in, adding about $100–200 million between 2019 and 2022.
These new investors also got “profit units,” which gave them a chance for future profits, but only within the capped limits.
OpenAI employees hired during these years were sometimes paid partly with these “units,” which made working there more attractive even compared to Google or Facebook.

These investments let OpenAI expand its staff rapidly and keep up in the “AI talent war.” The company was able to hire top researchers and engineers who wanted to work on the world’s most exciting AI projects, even if they could have earned more at big tech companies. Many of the investor deals were “convertible,” meaning they would turn into profit units at the next big funding event.

Microsoft’s Mega-Investment (2023)

In January 2023, Microsoft made a huge new deal, promising over $10 billion in a combination of cash, more cloud credits, and a bigger share of OpenAI’s future profits.

Microsoft could now get up to 49% of OpenAI’s profits, but only until they got back what they put in—then their share would drop to under 20%.
Microsoft became OpenAI’s exclusive cloud provider, so all of OpenAI’s big projects would run on Azure.
In return, Microsoft got to use OpenAI’s technology in its own products, including Bing (for search), Office (like Word and Excel), and GitHub Copilot (coding help).

This let OpenAI hire hundreds of new people and move even faster, launching new versions of GPT and rolling out AI features in real-world products. For Microsoft, the deal meant it could offer smarter AI features than competitors, making Bing and Office more appealing to millions of users.

Major Funding Rounds (2024–2025)

By late 2024, OpenAI had become one of the world’s most valuable tech companies.

In October 2024, it raised $6.6 billion from companies like Microsoft, NVIDIA, SoftBank, and Thrive Capital. This made OpenAI worth about $157 billion.
At the same time, OpenAI took $4 billion in loans from major banks, so it didn’t have to give up as much ownership.
Then, in March 2025, OpenAI raised an incredible $40 billion from SoftBank, Microsoft, and others—one of the biggest single rounds of funding in tech history. After this, OpenAI was valued at about $300 billion.

These investments meant that companies like NVIDIA, SoftBank, and Thrive became special partners. For example, NVIDIA got early access to new AI models (to help design better chips), while SoftBank could test OpenAI’s tools in its other businesses.

This much money meant OpenAI could afford even more advanced computers, hire more people, and build new products, keeping it ahead of rivals in the AI race.

Infrastructure and Compute Deals

Besides getting money from investors, OpenAI needed more and better computers than ever.

In 2025, OpenAI signed a deal to spend almost $12 billion with CoreWeave for high-powered GPU cloud computing.
CoreWeave would supply thousands of the world’s fastest graphics chips (GPUs), which are essential for training and running huge AI models.
OpenAI also partnered with Oracle, SoftBank, and others to build new, custom data centers, known as “Project Stargate,” to make sure it always had enough computing power.

These arrangements meant OpenAI didn’t have to rely on just one supplier for computing power and could keep growing even if demand for AI hardware increased worldwide. Having more control over its own data centers also meant lower costs and more security for its models and data.

Capitalization and Summary Table

Below is a summary table showing how OpenAI’s funding grew over time and what each stage allowed it to do:

Funding Phase	Amount & Structure	What It Made Possible
Nonprofit Origins	~$130 million in donations, no ownership	Built team, tried new research ideas, full flexibility
Capped-Profit Formation	$50M+ Khosla, units for founders/staff	Attracted real investment, kept mission control
Microsoft Anchor Deal	$1B (cash + Azure credits)	Paid for huge computing power, close partnership with Microsoft
Venture Capital Infusions	~$100–200M more from VCs and angels	Hired top talent, stayed competitive with Big Tech
Microsoft Expansion Commitment	$10B+ (cash, credits, profit share)	Rolled out new AI to millions, expanded team and infrastructure
Series E & Debt (Oct 2024)	$6.6B new investment, $4B in loans	Funded fast growth, valued at $157B, added more big partners
Series F (Mar 2025)	$40B, led by SoftBank, others join	Valued at ~$300B, let OpenAI build more products, partner globally
Compute & Data Center Contracts	~$12B to CoreWeave, Project Stargate	Guaranteed the biggest, fastest computers for AI, not dependent on just one supplier

This is how OpenAI went from a small nonprofit with no owners to one of the world’s top tech companies in under 10 years. The structure—blending donations, careful investor deals, and massive hardware partnerships—allowed OpenAI to get huge amounts of money and computing power while still protecting its core goal: making sure AI benefits everyone and stays safe. The original nonprofit board still controls the mission, so no outside company can take over or force OpenAI to work only for profit.

_______

DATA STUDIOS

datastudios.org