Skip to content
AI Buildoutby AlphAI
Dealschip supply

NVIDIA × OpenAI · Vera Rubin compute

Amended since Agent-reviewed
Chips & networks

OpenAI announced an expanded NVIDIA collaboration covering dedicated Vera Rubin inference and training capacity. This record uses the February 2026 compute scope rather than treating the earlier letter of intent as a finalized deployment.

Participants & roles

Terms as stated

Contract duration and delivery terms are not recorded.

Announced

Figures as stated

it capacity mwcommitted
3 GW

As stated by OpenAI ·

Source-described dedicated Vera Rubin inference scope only; the separately mentioned 2 GW of training is not summed here.

OpenAI source

Status history

  1. Amended

    OpenAI announced an expanded NVIDIA collaboration covering dedicated Vera Rubin inference and training capacity. This record uses the February 2026 compute scope rather than treating the earlier letter of intent as a finalized deployment.

    OpenAI ↗Agent review