Nvidia taps Samsung to boost Groq AI chip production

Nvidia has asked Samsung Electronics’ foundry business division to boost production of Groq-designed inference AI chips as demand for energy-efficient inference processors rises, industry sources said.

Groq is a US AI chip startup backed by Nvidia through a technology licensing partnership.

Samsung Electronics’ foundry division will raise wafer volumes for Groq from 9,000 units last year to 15,000 as mass production begins. The company will manufacture the chips on its 4-nanometer process and produce all processors for HyperAccel, a South Korean AI chip startup.

Reports say Nvidia may unveil a Groq-designed inference chip at GTC 2026 using SRAM instead of high-bandwidth memory.

🔗 Source: Chosun Daily

🧠 Food for thought

Implications, context, and why it matters.

Groq’s core strategy aims to avoid GPU supply-chain bottlenecks

Groq markets its Language Processing Unit (LPU) for AI inference, while GPUs tend to fit training workloads better 1.
Groq says the chip uses on-chip static RAM (SRAM) memory and reaches over 80 terabytes/second of bandwidth. The company adds that, at the architecture level, it uses about 10 times less energy than GPUs 2.
Groq says it avoids tight supply by sticking with a 14-nanometer process and skipping High Bandwidth Memory (HBM), which remains in extremely limited supply 1.
Groq says the current LPU is designed, engineered, and manufactured entirely in North America. It contrasts this with the global sourcing needed for today’s advanced GPUs 1.

Record AI-driven chip profits are squeezing smartphones

Demand for AI chips has triggered a memory shortage that reaches parts used in personal computers and mobile devices 3.
That shortage has lifted results for memory makers such as Samsung, whose operating profit rose over 200% year over year in Q4 2025, partly due to its memory business 3.
The same forces have weighed on other Samsung units. Samsung’s mobile business operating profit fell to 1.9 trillion won, down 9.5% year over year, as rising costs are expected to hit its smartphone business 3.
Analysts say higher component prices could squeeze device margins and push electronics prices up if companies pass those costs along 4.