Nvidia Has Quietly Become the AI Industry's Biggest Investor

Nvidia is best known as the company that sells the chips behind almost every frontier AI system. In the first months of 2026, it has also become one of the most aggressive equity investors in the same companies that buy those chips. New disclosures put the total at over $40 billion in committed AI investments this year alone.

The single largest cheque is the previously reported $30 billion stake in OpenAI. The remainder is spread across a roster of customers and infrastructure partners, including up to $3.2 billion in Corning, $2.1 billion in data-centre operator IREN, and additional commitments to CoreWeave and Nebius. The company has also taken positions in roughly two dozen private AI rounds.

The strategy

The pattern is consistent. Nvidia is financing the AI supply chain at every layer it touches. The labs building frontier models have received cheques. So have the cloud operators that rent out the GPUs, the data-centre developers laying the fibre, and the materials suppliers that produce the glass and substrates inside the chips. The unstated logic is that every dollar of investment increases demand for Nvidia hardware. The company gets to set the pace of its own market expansion.

That strategy has been working. Nvidia's market capitalisation has held near the top of the global rankings for most of the past year, even as the wider AI trade has wobbled. The equity stakes give it leverage in supply negotiations and a financial cushion against the cyclicality of chip sales.

The questions getting louder

Critics have begun to call the arrangement what it visibly is: capital recycling. As TechCrunch noted, Nvidia is in many cases investing in the same companies it then sells to, which has the effect of inflating revenue figures that depend on those very customers having the cash to spend.

The concern is not only optical. If a customer like CoreWeave or IREN is partly funded by Nvidia equity, and then spends most of that equity on Nvidia chips, the consolidated economic picture is a single company moving money around its own perimeter. Regulators in the United States and Europe have begun asking for more detail. Antitrust scholars have raised similar questions about Nvidia's investments in lab-side customers, an issue that surfaced again with Anthropic's Colossus lease from a SpaceX subsidiary earlier this month.

None of this is illegal, and none of it has slowed Nvidia's growth so far. But the $40 billion figure is hard to read as anything other than an admission that the company is no longer simply riding the AI wave. It is now one of the largest forces creating it. For investors, the practical question is whether the value Nvidia is generating in its own ecosystem reflects real end-customer demand, or whether some portion of it is the result of money making one short loop and being counted as revenue along the way.

The answer probably involves a bit of both. The harder question is how much.