Samsung Foundry Begins Mass Production of Nvidia's Groq3 AI Inference Chip

Samsung Foundry has commenced mass production of Nvidia's Groq3 LPX AI inference accelerator, a development announced at Hot Chips 2026. This move leverages Samsung's 4-nanometer process expertise, built through high-bandwidth memory (HBM) base die production, to manufacture core AI chips, signaling a crucial step in expanding AI infrastructure and compute capacity.