NVIDIA Projects Double AI Sales

GPU prices are on the move—Nebius just notified customers of on-demand rate hikes up to 21% for NVIDIA’s high-end chips, citing tight supply as AI demand surges. With heavyweights like Huawei and NVIDIA CEO Jensen Huang both signaling that demand is outpacing supply, the core issue isn’t just higher costs—it’s whether providers can keep pushing prices without losing customers. The dynamic is clear: as long as utilization rates stay high and queues are long, capacity owners have the upper hand. But if customers start balking, or shift to long-term contracts to lock in better rates, the balance could tip fast.

Here’s the catch: AMD is reportedly raising prices too, but theirs looks more like a cost pass-through than true pricing power. The looming risk is that visible hikes might push major buyers toward custom silicon and in-house solutions, accelerating a shift away from the big incumbents. Meanwhile in China, Huawei’s launch of the Atlas 960E SuperPoD ramps up pressure, but export controls and production bottlenecks mean NVIDIA’s real challenge is mostly regional—for now.

NVIDIA’s counter? Expand its software moat by bringing Rust into CUDA and beefing up GPU fleet management with Run:ai. Whether these moves lock in more developers and squeeze more revenue from every GPU-hour is a trend to watch. Based on reporting from Reuters, The Straits Times, and Tom’s Hardware.

Powered by Apisod.com