Nvidia has spent the past few years selling the chips that power artificial intelligence. Now it is also acting like a bank for some of its customers.
The company has taken on financing obligations that could total around $230 billion. This includes lease backstops and residual-value support tied to deals across the AI industry.
One of the largest pieces is a $105 billion backstop connected to an OpenAI lease deal in Ohio. Nvidia has also agreed to provide up to $125 billion in residual-value support for financing deals involving major Wall Street firms.
Big Financing Deals
Residual-value support generally means Nvidia is promising that certain assets backing a loan won't fall below an agreed price. This kind of guarantee gives lenders more confidence to extend credit.
Nvidia also recently struck deals with two young cloud computing companies in Australia. Under these deals, Nvidia agreed to act as a customer of last resort if the companies can't find other buyers for their computing power.
In return, Nvidia gets a share of revenue once those companies pass a certain earnings level. This arrangement helps smaller firms secure financing to purchase Nvidia chips.
These numbers don't include a separate $6.3 billion deal from last year, where Nvidia agreed to buy unused cloud computing capacity from CoreWeave. They also don't include Nvidia's equity stakes in many of the companies it supports.
At the end of its last fiscal quarter, Nvidia held $72.5 billion in public and private company stakes. Many of these are tied to firms that also receive financial backstops from Nvidia.
Nvidia has strong finances to support these commitments. The company had more than $80 billion in cash and marketable securities as of its last fiscal quarter.
Nvidia is not required to cover its full lease backstops. Those obligations shrink over time as leases progress.
Still, Nvidia described its Australian cloud deals as a "new business model." That suggests the company may take on more financial guarantee deals going forward.
Part of the reason for this shift is that large tech companies are becoming more financially stretched. Firms like Amazon and Microsoft already carry huge spending commitments tied to AI infrastructure.
Their financial statements show $904 billion in leases that have not yet started. They also show $1.52 trillion in purchase commitments largely tied to AI projects.
Borrowing costs for AI projects are also rising. Yields on AI-linked tech bonds have been increasing as more debt gets issued to fund data centers.
A Warning From the Past
History offers a caution for companies that extend financing to less established customers. During the dot-com era, telecom equipment maker Lucent Technologies offered loans and guarantees to smaller phone companies to boost sales.
Lucent later faced an SEC complaint over how it recognized revenue tied to those deals. The company eventually shrank to about a third of its original workforce before merging with France's Alcatel.
Wolfe Research analyst Chris Caso raised questions about whether chip companies backing data center financing could become a lasting trend. He said that outcome would be worth watching closely.
Nvidia is currently in a much stronger financial position than Lucent was at the time. Demand for AI chips remains high, and supply is not expected to catch up for another year or more.
Investors and analysts are expected to ask Nvidia about these financing arrangements when the company reports earnings on Wednesday.
