On August 27, Nvidia (NASDAQ:NVDA) confirmed it had paused parts of a financing program built to help AI cloud companies afford its chips, according to a Wall Street Journal report cited by Reuters. The timing stands out. Days earlier, Nvidia posted one of the biggest quarters in corporate history, with revenue jumping 106% and profit surging 126%. Investors now have to weigh a company still growing at a pace few businesses ever reach against fresh questions about how it finances the customers buying its chips.
Records That Keep Piling Up
The scale is the headline. Revenue for Nvidia’s fiscal second quarter, which ended July 26, came in at $96.2 billion, up 106% from a year earlier and ahead of both Nvidia’s own targets and Wall Street’s estimates. Gross margin expanded to 75% from 72.4% a year ago, operating income climbed 124% to $63.7 billion, and adjusted earnings per share rose 120% to $2.22. Guidance for the current quarter points to $108 billion in revenue, an 89.5% year-over-year increase.
The bigger reassurance for shareholders is that the hyperscalers building their own AI chips are not walking away. Hyperscale revenue hit $48.7 billion in the quarter, up 102% from a year ago, even as Amazon (NASDAQ:AMZN) and Alphabet (NASDAQ:GOOGL) executives said they will keep buying Nvidia hardware alongside their custom silicon. Amazon Web Services agreed to deploy an additional 2 million Nvidia GPUs over the next two years, a deal CEO Jensen Huang said could also include millions of CPUs. Nvidia is leaning into that opening, projecting $20 billion in stand-alone CPU revenue this year and more than double that next year, while also committing to return at least half its free cash flow to shareholders.
The Deal Nvidia Just Shelved
The paused financing initiative is where the bear case sharpens. Announced less than two months ago, the program offered credit support to small AI cloud firms in exchange for a cut of their revenue, with Nvidia agreeing to rent back compute capacity those firms couldn’t sell and collecting 50% of revenue above a certain threshold. Nvidia stepped back from some of those deals last week, though a spokesperson said the broader model “is still in place and continues to evolve due to high demand.” The report noted some Nvidia employees had privately worried the arrangement could draw antitrust scrutiny, and that the company had told cloud providers they could only rent chips to approved customers, a level of control that reportedly irritated potential partners early on.
