Nvidia projects 70 percent sales growth on AI chip demand

Nvidia’s latest forecast makes clear that the AI build‑out is still in a full acceleration phase, not tapering off. A planned 70 percent revenue jump on top of already historic growth means hyperscalers, sovereign funds and major AI labs are still signing multi‑year capacity deals and building ever‑larger clusters. With Vera Rubin now in full production and memory shortages the main brake on shipments, the constraint on AGI‑class experiments remains physical infrastructure, not model ideas.

This has two big implications for the AGI race. First, it entrenches Nvidia’s role as a central gatekeeper: whoever can lock in long‑term access to its highest‑end accelerators and memory‑rich platforms will enjoy a clear runway to scale models, context windows and agent deployments. Second, it increases the risk of systemic over‑investment and circular financing, as Nvidia both supplies chips and helps bankroll AI companies that are its customers. If demand holds, this will compress timelines for trying very large‑scale training runs and multi‑agent systems; if it wobbles, the unwind could be messy. Either way, R&D agendas at OpenAI, Anthropic, Google and emerging players will be shaped by Nvidia’s capacity, pricing and willingness to back their data center plans.