NVIDIA Grok 3 LPX enters mass production as Samsung yields exceed 80%

Nvidia has announced the start of volume production for its Grok 3 LPX AI inference accelerator, manufactured by Samsung Electronics on its 4nm process with yields exceeding 80%. The platform packages 256 LPUs into a rack for up to four times faster inference in agentic AI workloads. Nebius is the first customer for its Nebius Token Factory deployment at year-end. The collaboration now includes HBM4 supply for Nvidia's Rubin GPU and other components like SOCAMM2. OpenAI's Habanero chip is also using Samsung's HBM4. Samsung has raised prices by 10-15% and the 4nm line is at near 100% utilization, with analysts expecting the foundry to return to profitability in the third quarter.

The information on this website is generated using AI and we cannot guarantee its accuracy. Please use it as reference information only.