News

Nvidia CEO Jensen Huang says company is ramping capacity for Amazon Web Services to support OpenAI

Wednesday, February 25, 2026 at 11:08 PM

Nvidia CEO Jensen Huang confirmed that the company is aggressively increasing supply and capacity for Amazon Web Services (AWS) to accommodate expanding infrastructure needs from OpenAI. This ramp-up is intended to support growing compute demand, including inference needs for companies like Anthropic and the rise of agentic AI.

Context

At the Morgan Stanley TMT conference in March 2026, Nvidia CEO Jensen Huang confirmed a major strategic shift in the AI infrastructure landscape, stating that the company has "expanded OpenAI's reach of capacity to AWS" and is currently "ramping AWS like mad." This follows a series of massive investment announcements in early 2026, including a $50 billion commitment from Amazon into OpenAI. Under this partnership, AWS has become the exclusive third-party provider for OpenAI Frontier, a new platform for managing autonomous AI agents. This ramp-up is critical for the semiconductor supply chain as OpenAI is expected to consume 2 gigawatts of Trainium capacity while utilizing hundreds of thousands of Nvidia GPUs. The collaboration introduces a Stateful Runtime Environment on Amazon Bedrock, moving beyond isolated API calls to persistent AI memory. For investors, this signifies a multi-cloud expansion for OpenAI—previously tied closely to Microsoft Azure—and a significant acceleration of high-end chip deployment to support agentic AI workloads through 2026.

Related Companies

M
Morgan Stanley
MS
Nvidia
Nvidia
NVDA
US