OpenAI is launching a new $400 million venture fund, more than twice the size of its original $175 million vehicle, according to a new U.S. Securities and Exchange Commission filing reported by The Wall Street Journal. But the size of the fund is only part of the story.
Unlike OpenAI’s first venture fund, which was backed by outside investors including Microsoft, the new fund is being financed entirely from OpenAI’s own balance sheet. That gives the company something it did not have before: direct economic exposure to the startups building on top of the AI ecosystem it is trying to dominate. And OpenAI already has a track record worth paying attention to.
The first fund found some serious winners
OpenAI launched its Startup Fund in 2021, initially targeting $100 million before ultimately raising around $175 million from external investors. The structure was unusual. OpenAI itself was not the fund’s investor, and Sam Altman initially held legal control of the vehicle. That changed in 2024, when control was transferred to Ian Hathaway, who had been helping run the fund since its launch.
The fund subsequently backed a growing list of AI startups across developer tools, healthcare, enterprise software and consumer applications. Among them were Anysphere, the company behind Cursor; Harvey, the legal AI platform; Descript; Speak; Figure AI; 1X; Physical Intelligence; and Ambience Healthcare. OpenAI also raised additional capital through a series of special-purpose vehicles, taking its total venture activity beyond the original fund.
Some of those bets now look particularly interesting in hindsight. Anysphere, for example, received an $8 million seed investment from the OpenAI Startup Fund in 2023. The company went on to build Cursor into one of the breakout products of the AI software wave, before Cursor was ultimately acquired by SpaceX in 2026.
Harvey followed a similarly aggressive trajectory, becoming one of the best-known AI legal-tech companies in the market. The lesson for OpenAI is obvious: getting into promising AI companies early can create returns far beyond the original equity investment. But financial returns are only one part of the equation.
OpenAI is buying more than equity
This is where the new $400 million fund becomes much more interesting. For OpenAI, investing in an AI startup can potentially create several different returns at once. There is the obvious one: equity appreciation.
But there can also be strategic value. A startup backed by OpenAI may become an OpenAI API customer. It may build its product around OpenAI models. It may distribute OpenAI technology to thousands or millions of users. It may eventually become an acquisition target.
In other words, OpenAI can potentially make money from the startup and make money from the startup’s growth through its own ecosystem. That creates a powerful flywheel: OpenAI provides capital → startup builds on OpenAI → startup grows → OpenAI earns equity upside and potentially more model usage → the ecosystem gets stronger.
This is not entirely new. Google, Microsoft, Amazon and Nvidia have all used strategic investments to strengthen their positions across technology ecosystems. But in AI, the stakes are unusually high. The competition is no longer just about which company has the best foundation model. It is increasingly about who owns the applications, workflows, distribution and customer relationships built on top of those models.
$400 million gives OpenAI a much bigger seat at the table
The new fund is roughly 2.3 times larger than the original $175 million fund. According to The Wall Street Journal, the fund is expected to focus on early-stage AI companies and invest in roughly 8–10 startups a year, with individual investments ranging from several million dollars to as much as $100 million. Ian Hathaway is continuing to lead the fund. That is enough firepower to become a meaningful player in the venture market rather than simply a strategic side project.
And there is another important difference: OpenAI is now putting its own capital at risk. With external limited partners, investment losses and gains are shared according to the fund structure. With the new vehicle, OpenAI takes the economic exposure itself, but also retains the upside. If one of its portfolio companies becomes the next Cursor or Harvey, OpenAI gets to capture the investment return directly.
The bigger question: where does OpenAI want its ecosystem to go?
There is a strategic question underneath all of this. OpenAI is not just building an AI model company anymore. It is increasingly trying to shape an entire technology ecosystem around its models. That means having influence over: developer tools, enterprise software, healthcare, legal technology, robotics, AI agents, consumer applications, or infrastructure.
A venture fund is a relatively cheap way to get exposure to all of those markets simultaneously. OpenAI doesn’t have to build every application itself. It can fund the companies that do.
That could become increasingly important as competition intensifies between OpenAI, Anthropic, Google and a growing number of open-source and Chinese AI players. The foundation models may eventually become more interchangeable. If that happens, the real competitive advantage could move further up the stack – toward the companies that own the actual products and workflows customers use every day.
There is also a potential conflict
The strategy is powerful, but it comes with an obvious tension. If OpenAI becomes both a dominant model provider and a major investor in the companies building on top of AI, its incentives become more complicated. Startups may benefit enormously from OpenAI’s capital, technology and distribution.
But what happens when one of those startups wants to use a competing model? Or when an OpenAI-backed company starts building something that could eventually compete with OpenAI itself? And what happens if startups begin choosing their technology stack partly because of access to OpenAI’s investment network?
These questions become more important as the fund grows. The first OpenAI Startup Fund was relatively small. Its portfolio could be viewed as a collection of early bets. A $400 million fund backed directly by OpenAI starts to look more like an ecosystem-building strategy.
OpenAI is turning venture capital into another AI weapon
The most interesting thing about the new fund may therefore not be the $400 million itself. For OpenAI, $400 million is relatively small compared with the billions of dollars it is raising and spending to build models and computing infrastructure. But venture capital gives the company something its enormous infrastructure spending cannot: optionality.
Instead of trying to predict which AI applications will win, OpenAI can place dozens of early bets and let the market decide. If a handful become category leaders, OpenAI can benefit through its ownership stake, its model usage, its distribution relationships, and potentially future acquisitions.
The first fund showed that the strategy can work. Now OpenAI is putting more of its own money behind it. The next phase of the AI race may not just be about building the best model. It may be about owning the ecosystem that grows around it. And OpenAI has just committed another $400 million to make sure it has a seat at that table.
