Nvidia CEO Says Telecom Operators Will All Offer AI Compute

Huang's pitch reframes telecom operators as prospective compute hosts, which would put Nvidia's hardware inside infrastructure that regulators and carriers already control — a different sales channel than hyperscalers. The $50 billion per-gigawatt figure is the tell: capital intensity at that scale favors operators with balance sheets and spectrum, not startups, and the five-layer framing implicitly positions Nvidia as the toll point beneath models and applications. Whether carriers actually convert base stations into AI nodes, rather than reselling centralized capacity, is the open question worth watching.