Nvidia's investment portfolio hit $99 billion. Here is what it is buying and why.

Key points

• Nvidia's equity investments, the shares it holds in other companies, were valued at $99 billion as of 26 July 2025, up from roughly $7 billion a year earlier.
• The company committed more than $40 billion to financing rounds across the AI industry in 2026.
• Nvidia invested nearly $50 billion specifically into frontier AI labs, the research companies building the most advanced AI systems.
• Its $5 billion stake in Intel has grown to a market value of $30 billion, while its SpaceX holding was worth $21 billion as of June 2025.
• Nvidia also announced plans to acquire AI startup Hugging Face for $12.9 billion.

Nvidia makes the chips that power almost every serious AI system on the planet. Now it is writing cheques for the companies that buy those chips, too.

The company's portfolio of equity investments, meaning shares it holds in other businesses, reached $99 billion as of 26 July 2025. Two years ago that figure was $2.2 billion. CNBC Tech first reported the scale of the holdings alongside Nvidia's latest earnings.

What is Nvidia actually buying?

Three broad groups of companies have received the bulk of the money.

First, frontier AI labs: the research firms building the most capable AI models. Nvidia says it has invested nearly $50 billion into this group. In February, it committed $30 billion to OpenAI as part of that company's $110 billion funding round.

Second, neoclouds: specialised data-centre businesses that buy Nvidia's GPUs (the chips that do the heavy computing AI needs) in bulk and then rent access to smaller companies. Nvidia invested $2 billion in CoreWeave in January 2025 and another $2 billion in Nebius in March 2025.

Third, optical and photonics companies: firms using light instead of electricity to move data inside computers, which can be faster and more energy-efficient. Since March 2025, Nvidia has committed at least $6.5 billion to this area. Lumentum, Coherent and Marvell each received $2 billion.

Nvidia also announced plans to acquire Hugging Face, an AI startup that hosts and shares AI models, for $12.9 billion.

Why would a chip company invest in its own customers?

The short answer: Nvidia needs those customers to survive and grow.

Frontier AI labs were growing faster than their finances could support, Nvidia's chief financial officer Colette Kress told analysts. They struggled to fund the data centres they needed. If those labs can't afford the hardware, Nvidia doesn't sell chips.

Analyst Ian Fogg at CCS Insight put it plainly: "Nvidia has a clear interest in ensuring that its customers and partners prosper to provide future business for Nvidia."

There is a second motive. By investing in companies that build software and hardware around its chips, Nvidia encourages them to keep optimising for its architecture rather than switching to chips from rivals like AMD. Naveen Chhabra of Forrester described it as protecting Nvidia's "software moat," the deep layer of tools and code that makes switching away from Nvidia expensive and painful.

The Intel stake offers a different kind of insurance. Nvidia's $5 billion investment, now worth $30 billion on paper, helps secure access to Intel's manufacturing capacity and reduces Nvidia's reliance on Asian chip factories at a time when global supply chains face pressure.

What does this mean for ordinary people?

Directly, very little changes today. But Nvidia's spending shapes which AI companies get funded, which technologies get built, and which chip architecture the industry standardises on. The companies receiving this money build the AI tools that show up in healthcare software, customer service bots and productivity apps that millions of people already use.

If Nvidia's bet pays off, a stable, well-funded AI ecosystem could mean more reliable, cheaper AI services over time. If the strategy concentrates too much of the industry around a single chip maker's preferences, regulators may eventually take notice.