How Nvidia Tried to Profit Twice From Every Chip Sale—and Why It Backfired

NVDA's plan to collect 50% of cloud revenue above a guaranteed floor, which would have let it earn twice on each chip sold, was paused within weeks after internal antitrust warnings.

NVDA mobilized over $500 billion in third-party capital with Apollo, BlackRock, and Goldman Sachs, with balance-sheet-supported labs projected to reach 25% of business next year.

Don't wait: the analyst who called NVIDIA in 2010 just revealed his top 10 AI stocks. See the full list FREE now .

NVIDIA's ( NASDAQ:NVDA ) $96.22 billion quarter and 105.85% revenue growth would normally end the conversation. Instead, management disclosed a plan to earn a second time on every chip sold to smaller cloud providers. Days later, parts of the initiative had been paused less than two months after its announcement, with some employees warning internally about potential antitrust scrutiny.

NVIDIA still runs the most profitable franchise in semiconductors, but the speed of the retreat is the story worth examining.

How the Same Chip Was Supposed to Pay Twice

The mechanism is unusual. NVIDIA would guarantee or rent unused capacity from a smaller cloud provider, which gave lenders the certainty needed to finance the hardware purchase.

CFO Colette Kress described it directly on the call: "NVIDIA provides a take or pay commitment on a portion of the facility's capacity, a minimum revenue guarantee that gives lenders the confidence to underwrite the project, and in exchange, we share in a portion of the Neocloud's revenue earned above that floor."

Above the guaranteed floor, NVIDIA would collect 50% of cloud revenue. Management summarized the economics without euphemism: "In this model, we get paid twice, once on the hardware sale and again through the share of rental revenue."

A capital-starved cloud provider gets financeable, NVIDIA books the sale, and then rides the utilization curve on hardware it already sold. The structure effectively converts a one-time transaction into a recurring revenue stream tied to compute usage, without requiring NVIDIA to operate the infrastructure itself.

NVIDIA does not need this program to justify its $5.25 trillion market cap. Data Center revenue was $89.023 billion last quarter, up 117%.

But the second revenue stream would layer recurring economics onto a transactional business. Management said it could "drive billions in revenue over the medium to long term."