Nvidia's AI Business Faces New Risks Amid Frontier Model Slowdown
Nvidia's AI business has been affected by recent calls from high-profile AI leaders to slow down frontier AI model development. Dario Amodei, CEO of Anthropic, is among those who have urged a slower pace for these models.
This has raised questions about how any future guardrails might affect Nvidia's core AI data center business and long pipeline of infrastructure projects. However, major customers report no evidence of cancelled chip orders or paused buildouts.
Nvidia's investment narrative hinges on the scarcity and value of AI compute capacity. The company uses its full stack, from GPUs to software, to supply that capacity. Hyperscalers and sovereign buyers are expected to continue building AI factories at the current clip.
The biggest operational risk for Nvidia is not demand, but constraints such as chip supply and memory availability, power, and regulatory scrutiny of large data centers. A sustained hit to China access or a faster shift by big customers to in-house silicon could start to bite into the multi-year AI infrastructure story.
Nvidia's attempt to address one of these bottlenecks is through its participation in the AI Energy Management Alliance, which aims to make AI data centers more grid-friendly. The alliance involves Nvidia, Google, and Emerald AI, and targets one of the practical bottlenecks in scaling AI: access to reliable, affordable power.