NVIDIA Is Betting Billions on AI Data Centers. Here's Why That Matters

NVIDIA, the company that makes the computer chips powering artificial intelligence, is doing something unusual: it's spending over $40 billion buying stakes in other companies and infrastructure projects. These aren't just any investments. NVIDIA is putting money into the data centers—the massive buildings full of computers—that will use its chips. In a sense, the company is now both the supplier of the tools and a financial investor in the places that will use them.

The biggest deal is a $40 billion purchase of Aligned Data Centers, done together with Microsoft, the investment firm BlackRock, and Elon Musk's AI company xAI. This data center will house clusters of NVIDIA chips. The deal is expected to close by mid-2026.

In another major move, Oracle is buying $40 billion worth of NVIDIA's chips to power OpenAI's new data center in the United States. This is one of the largest single orders of computer chips in business history—a sign of just how much computing power is needed to train and run advanced AI systems.

Expanding Beyond Chip Sales

NVIDIA is also building partnerships in new ways. The company announced a collaboration with Intel, its long-time competitor, to develop AI equipment together. That kind of partnership would have been unthinkable a few years ago.

NVIDIA is also investing in AI research labs. The company is involved in discussions to invest in a $60 billion funding round for OpenAI, the company behind ChatGPT. This comes after SoftBank invested over $40 billion in OpenAI, making it one of the most well-funded private companies in the world.

NVIDIA also committed money to more specialized suppliers. It committed up to $2.1 billion in IREN, a company that builds power-efficient data centers, and up to $3.2 billion in Corning, a glass maker that supplies fiber optics and materials used in high-performance computer systems.

How Institutional Money Is Getting Involved

Brookfield, a major investment firm, recently created a new fund targeting $10 billion to invest in AI infrastructure. When combined with other investors and financing, this fund could control $100 billion in AI infrastructure assets. The fund focuses on large-scale data centers built to a common design that NVIDIA has endorsed, creating a standardized way to build these massive facilities.

This represents a shift in how infrastructure gets built. Instead of relying only on the large tech companies' own budgets, specialized investment funds are now emerging to funnel capital to the scale these AI projects require.

There is historical precedent for this. In the late 1990s, when the internet was expanding rapidly, Cisco—a major networking equipment company—mixed chip sales with strategic investments in the service providers and companies that would use its products. That created a feedback loop: Cisco's success as an investor drove more demand for its gear. NVIDIA appears to be following the same playbook, but at much larger financial scales.

The Returns Have Been Real

NVIDIA's earlier equity bets have paid off alongside its core chip business. The company's investment in Intel is now worth over $25 billion, up from its initial $5 billion stake. That shows how investing in other companies can generate outsized returns.

In practical terms, NVIDIA is earning money from multiple sources tied to the same AI infrastructure expansion: it sells the chips, it holds stakes in the companies that buy those chips, and it receives licensing fees when infrastructure funds use its reference designs. Think of it like a real estate developer who builds the building, invests in the shopping mall inside it, and collects rent from the stores.

The broader context here is worth noting. NVIDIA's strategy suggests the company believes that building out AI infrastructure at the required scale cannot happen through chip sales alone. The sheer amount of capital required—tens and hundreds of billions of dollars—may demand that the chip maker become a financial partner, not just a supplier. This is a departure from how the semiconductor industry has typically operated, but it reflects the unprecedented scale of the current AI buildout.