How Nvidia is backstopping America’s AI boom

Chief Executive Jensen Huang has repeatedly sought to leverage the enormous confidence investors have in Nvidia’s future to help strengthen the company’s supply chain partners. It has used its balance sheet clout to keep the AI boom humming through deals, partnerships and investments in companies that are among its top customers, including cloud-computing provider CoreWeave, rival chip designer Intel and xAI.

The deals highlight an issue that some investors are calling “circularity" in Nvidia’s prospects, whereby the company takes steps to boost or shore up demand for its AI chips by supporting startups and other companies. Those companies can then use those funds or new liquidity to buy Nvidia chips.

For every $10 billion Nvidia invests in OpenAI, the startup will spend $35 billion on Nvidia chips, according to an analysis from NewStreet Research. That arrangement reduces Nvidia’s typical margins for cutting-edge chips, but ensures continued demand and offers a lifeline to cash-strapped AI companies. It effectively amounts to a discount for OpenAI.

NewStreet analysts said they expect Nvidia to offer similar deals to xAI and other “cash-constrained" players.

The OpenAI announcement alone on Monday added nearly $160 billion to Nvidia’s market value, cementing its ties to the leading software company of the artificial intelligence era. It also helped to silence those skeptical of how OpenAI would pay for the many multibillion-dollar commitments it has made in recent months.

Nvidia’s investment amounts to a major vote of confidence in OpenAI, which has seen its growth reach 700 million monthly users but struggled to chart a path to profitability.

Sam Altman, OpenAI’s co-founder and CEO, told investors last fall that the company is set to lose $44 billion through 2029, the year it expects to turn its first profit. That was before OpenAI signed pricey commitments to buy chips and rent data center capacity from the likes of Broadcom and Oracle.

What all these deals have in common is that Nvidia—the world’s most-valuable public company and the most powerful munitions dealer in the global AI arms race—is lending its name and financial firepower to smaller partners.

AI executives expect the crucial development to give OpenAI access to much cheaper sources of capital than it had before. In the past, when OpenAI has needed access to thousands of Nvidia’s chips, it has done so through cloud service providers and companies known as neo-clouds, which act as middlemen, paying for the development of data center clusters, buying the chips, and then renting them out at a premium.

The debt used to underwrite deals for data centers often depends in part on the perceived creditworthiness of money-losing AI companies such as OpenAI. Interest rates on data centers closely linked to such startups have come in as high as 15%, a proxy in the debt market for the risk investors see in AI business models, according to people familiar with the financing of such deals.

Interest rates for projects backed by the likes of Microsoft, for example, can range from 6% to 9%, the people said. Last week, credit rating agency Moody’s noted risks to Oracle’s balance sheet due to how much of its future AI data centers rely on OpenAI. The agency moved Oracle’s outlook from stable to negative in July.

Nvidia, on the other hand, enjoys near-unshakeable faith from the market, which has sent its share price skyrocketing. That market confidence allows it to finance those infrastructure costs from its balance sheet or issue new equity to pay for them, a much less expensive financing option for OpenAI.

Executives in the AI infrastructure space say that Monday’s deal will likely reduce the credit risk of lending to the company, giving the firm access to lower-interest rate loans.

In addition to OpenAI, Nvidia has made strategic investments in a number of large tech companies over the past year, often in ways that strengthen its edge.

Nvidia owns a 7% stake in CoreWeave, a cloud services provider that rents out its massive data clusters to users like OpenAI, Microsoft and others.

Earlier this month, the two companies signed a $6.3 billion pact for Nvidia to buy back any unused cloud capacity CoreWeave has through April, 2032.

And last week, Nvidia announced it would invest $5 billion in Intel, a rival chip designer and manufacturer that has fallen badly behind in the global AI race. The partnership involves developing new products that will allow Nvidia to more easily connect its GPUs to processors designed by Intel, giving Nvidia a deeper footing in the personal computer market.

In December, Elon Musk’s xAI said Nvidia was a “strategic investor." In March, Nvidia also joined a global AI partnership that includes xAI and plans to spend billions on AI data centers and energy infrastructure.