NVIDIA announced that its Groq 3 LPX artificial intelligence inference accelerator has entered full production, positioning the hardware for high-speed processing in agentic AI systems.

The company said Groq 3 LPX is an extension of its Vera Rubin platform and is designed to accelerate AI inference—the process of generating responses from trained models. NVIDIA said the accelerator enables ultrafast token generation to support more responsive AI applications.

The announcement was issued through the NVIDIA Newsroom on Aug. 24, 2026. The company also released images of the Groq 3 LPX rack alongside the production update.