The Wall Street Journal, citing people familiar with the matter, reports NVIDIA
(NVDA.O) has paused part of its AI compute partner program that provided credit
support to cloud providers in exchange for revenue sharing, less than two months
after the plan’s July announcement. NVIDIA disclosed this week the proposed,
typically six‑year agreements involve about $36 billion in commitments. Under
the proposal, if a cloud provider cannot find customers NVIDIA would lease its
GPUs to them to guarantee revenue and help secure financing; once provider
revenue exceeds a base threshold covering chip depreciation, data‑center and
personnel costs, NVIDIA would receive 50% of the excess. Some NVIDIA employees
fear the program could prompt antitrust scrutiny over how much the company could
shape customer operations. NVIDIA may adjust the program or fold it into other
initiatives.
