Wire · technology
NVIDIA Mass Produces Groq 3 LPX via Samsung Foundry
◆ Sectors
◆ Geography
◆ Source
◆ Verified
Fusion42 · 25 August 2026 · Fusion42 review
NVIDIA has started mass production of its AI inference chip Groq 3 LPX using Samsung Electronics' 4-nanometer foundry process, targeting AI inference tasks with faster speeds. The chip is already secured by AI cloud operator Nebius, and this marks a strategic move to separate inference from training workloads in AI computing.
This Wire brief sits within Fusion42's coverage of AI Infrastructure, and 3 sources have reported it between 24 Aug 2026 and 25 Aug 2026.
◆ ◆ The Wire takeaway
NVIDIA's use of Samsung's advanced foundry for Groq 3 LPX chips opens a clear path for faster AI inference hardware supply. If you build AI deployment tools or run AI cloud services, this expands your options to optimise inference efficiency now.
◆ Coverage
3 sources · first reported 24 Aug 2026 · latest 25 Aug 2026
◆ Related on Wire
◆ Topics