Samsung Starts Mass-Producing Nvidia Inference Chips
Samsung Electronics’ foundry (contract manufacturing) division has secured a new production base within Nvidia’s artificial intelligence (AI) ecosystem. As the center of gravity in the AI market shifts from training models to inference that performs actual answers and tasks, Nvidia’s inference-dedic
Samsung Electronics' foundry has begun mass-producing Nvidia's inference chip, 'Groq 3 LPX,' as the AI industry shifts from model training to real-time tasks. Nvidia's 'Groq 3 LPX' accelerates AI responses by using high-speed LPUs, which are produced by Samsung Electronics. The LPU chip focuses on reducing AI response times using SRAM instead of general DRAM.
Nebius, an AI cloud company, is Samsung Electronics' first customer for the 'Groq 3 LPX.' Nvidia CEO Jensen Huang emphasized that inference is the growth engine of AI, promising increased throughput, efficiency, and responsiveness. This development expands Samsung Electronics' involvement with Nvidia beyond memory production, opening new avenues in the foundry business as AI semiconductors diversify.
SpaceXAI, led by Elon Musk, has also secured the 'Vera' CPU, which will run on the 'Vera Rubin' system, with plans to use it in both terrestrial and orbital data centers. The competition in the AI semiconductor market is intensifying, with AMD introducing its 'Helios' system and Nvidia securing additional customers to strengthen its position.
Written by urgent.news from BusinessKorea's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
