Nvidia has committed more than $40B in AI equity deals by 2026, raising dot-com era comparisons.

Nvidia has invested more than $40 billion in AI and technology companies by 2026, a figure that surpasses its total investment activity in 2025. The GPU giant isn't just selling shovels during the gold rush. He is also buying a stake in the mine.

The centerpiece of this spending spree is an estimated $30 billion investment in OpenAI, the largest single equity deal in Nvidia's history. The rest of the billions are spread across a portfolio of AI infrastructure plays, including a significant deal with IREN, a company that used Bitcoin before bringing it to AI computing.

Financing your own needs

The thing about Nvidia's investment strategy is that the companies it supports are also its biggest customers. OpenAI runs massive GPU clusters to train and serve its models. IREN is building AI compute infrastructure. These components require Nvidia hardware and more.

In English: Nvidia is writing checks to companies that go around and using that money to buy more Nvidia chips.

The IREN Agreement clearly illustrates the pattern. The arrangement includes a $2.1 billion equity guarantee along with a $3.4 billion AI computing infrastructure deal. IREN, which previously operated as a bitcoin mining company, is effectively banking on building data centers that run on Nvidia silicon.

No one saw the crypto crossover coming.

IREN's transformation from a Bitcoin miner to an AI infrastructure provider deserves its own attention. Bitcoin mining operations have three things that AI data centers need most: power capacity, cooling infrastructure, and facilities designed to run GPUs around the clock.

For Nivea, supporting these conversions serves two purposes. It expands the physical infrastructure that can run the chips while simultaneously reducing the GPU capacity available for competitive use.

Dot-com déjà vu

Commentators have begun to draw parallels between Nvidia's circular financial approach and the dynamics that characterized the technology bubble of the late 1990s. In the dot-com era, companies like Cisco and Sun Microsystems provided supplier financing to start-ups who used the capital to buy networking equipment and servers from Cisco and Sun Microsystems. The income looks amazing until the customers lose their money and the loans default.

The bull case at this point is that the underlying technology creates real economic value. OpenAI is reported to generate meaningful revenue. The demand for AI computing seems real and growing.

The bear case is that $40 billion in equity deals creates a web of financial interdependence in which a fall in AI spending will go through Nvidia's portfolio and revenue simultaneously. If your clients are your portfolio companies, an economic downturn in your sector will hit you twice: once on the income statement, once on the balance sheet.

Navia is becoming an effective centralized distributor of capital for the entire AI ecosystem, choosing which companies get funding and which projects get built. That's an unusual influence for a company whose main business is designing chips.