The scaling economics of large AI workloads have yet to find a ceiling that slows procurement. Nvidia, the chip giant driving the current AI infrastructure cycle, has projected sales growth of around 70%, citing sustained AI demand, and has described financing deals extended to customers as excellent investments carrying limited risk.
The demand dynamic behind the projection runs across two phases, both drawing on the same hardware stack. Model training requires dense, parallel compute over extended periods, and that requirement scales with model complexity. Commercial inference adds a persistent second layer: every processed query is a compute event, and in aggregate the operational load of a widely deployed model can approach the demands of its original training run. Neither phase has a near-term substitute that bypasses high-end GPU capacity. That is the supply constraint Nvidia's 70% growth projection rests on.
The figure is a forward estimate attributed by Nvidia to ongoing AI demand. The specific revenue base or reporting period the forecast applies to was not included in the available disclosure.
Nvidia went beyond a revenue outlook in addressing the financing arrangements it extends to hardware buyers. The company described those deals as excellent investments and explicitly characterized the risk as limited. Extending credit to customers creates balance sheet exposure that ordinary component sales do not, which gives the framing weight. Calling the positions excellent and the risk limited is a direct public claim about the creditworthiness and capital durability of Nvidia's largest customers.