The most active investor in the AI industry is not a venture firm, a sovereign wealth fund or a hedge fund. It is the company that makes the chips. Nvidia has invested close to $99 billion in AI-related companies over the past year, according to a tally by Quartz, with more than $40 billion of that committed in the current year alone. The recipients read like a map of the industry’s center: OpenAI, Anthropic, SpaceX, xAI and a constellation of smaller labs and infrastructure firms.
The numbers describe a strategy that goes well beyond passive portfolio building. Nvidia has been converting its position as the dominant supplier of AI chips into a position as the dominant investor in AI itself, writing checks to the very companies whose computing needs its products serve. The pattern is visible across the industry: startups that announce partnerships with Nvidia for computing capacity frequently announce investments from Nvidia in the same breath, and the two announcements have become a single genre of corporate news.
The logic is straightforward, and Nvidia has been explicit about it. Every dollar invested in an AI company is a dollar that tends to come back as orders for Nvidia hardware, because the companies that train and run frontier models have few alternatives to its chips. The investments lock in demand, build loyalty and give Nvidia a seat at the table when its customers make their own strategic decisions. From selling shovels to holding equity in the miners, the company has moved to occupy both positions at once.
The scale of the program is what makes it unusual. A $99 billion investment portfolio would be enormous for any institution, and the fact that it belongs to a company whose core business is selling chips means the portfolio has a dual purpose: financial return and strategic control. Nvidia’s own valuation, which has made it one of the most valuable companies in the world, increasingly rests on the story that it is not merely a hardware supplier but the financial backbone of the AI industry.
The investments also serve as a hedge against the industry’s own volatility. AI companies raise and spend money in cycles, and a downturn in funding would hurt the labs Nvidia depends on for demand. By investing directly, Nvidia gains some influence over its customers’ survival, and its checks have helped keep the AI ecosystem financed through periods when venture capital alone might not have been sufficient.
Regulators have begun to notice. The concentration of the AI industry, with one company supplying the dominant chips and simultaneously owning stakes in the firms that buy them, has attracted scrutiny from antitrust authorities in the United States and Europe. The concern is not the investments themselves but their cumulative effect: a chip maker that controls the hardware, invests in the customers and could one day influence the competitive outcomes among them occupies a position that competition law was designed to question.
There is a precedent in technology history for companies that used investment to extend their reach, but few have done it at this scale or with this much of their own revenue at stake. Nvidia’s approach differs from the classic platform investments of the past because its core product is in such short supply: the companies it backs are competing for chips that Nvidia allocates, and the combination of allocation decisions and ownership stakes gives it influence over the ecosystem that no earlier chip company possessed.
The portfolio is not without risk. AI valuations have reached levels that would make any investor nervous, and Nvidia’s holdings include companies whose revenue does not yet justify their prices. A correction in AI valuations would hit Nvidia twice: once through its investment portfolio and once through the slowdown in chip demand that would follow if its customers’ access to capital dried up. The company’s exposure to the industry’s fortunes has become total, in both directions.
For the companies that receive Nvidia’s money, the investment is usually welcome and sometimes complicated. Nvidia’s checks come with its hardware allocation, a combination that startups find hard to refuse, but they also come with an investor whose interests may not always align with theirs. As the portfolio grows, some of Nvidia’s portfolio companies compete with one another, and the company’s role as a common investor across the industry’s rival camps is a fact that will shape how those rivalries play out.
The $99 billion figure, if accurate, is likely to grow. Nvidia has shown no sign of slowing its investment pace, and the list of companies seeking its money appears to lengthen with every funding round in the industry. The company has effectively become the AI industry’s central bank, distributing capital alongside compute, and its decisions about which companies to back will help determine which laboratories, which models and which business models survive. For a company that began as a graphics-chip maker, the transformation is complete: Nvidia now sells the industry its machines and holds the industry’s paper, and the two roles reinforce each other at every turn.
