According to Korean media reports, Samsung’s foundry business is staging a strong rebound. After securing orders from Tesla and Qualcomm, Samsung has now won the contract to produce NVIDIA’s next-generation AI inference chip, the Grok 3 LPU, and is currently in talks with AMD for future chip collaborations.
At GTC 2026, NVIDIA CEO Jensen Huang announced that Samsung Electronics will manufacture the Grok 3 LPU, the first product launched after NVIDIA’s $20 billion acquisition of Groq. The chip targets the surging demand for inference processors driven by the shift toward agent-based AI.
Samsung has already begun mass production of Grok 3 LPU, with shipments expected to start in the third quarter of this year. Industry insiders revealed that Samsung Foundry has set Q4 as its break-even milestone, with analysts predicting a sharp reduction in losses from the second half of 2026 and possible profitability by year-end.
On the technical side, Samsung’s 2nm process yield has stabilized, and its new Taylor, Texas fab is scheduled to begin operations in the second half of the year, boosting production capacity.
Previously, Samsung Foundry was considered the “problem business” within its Device Solutions division due to heavy upfront investments and lower-than-expected yields. Its global market share had slipped into single digits. However, with the maturity of its 2nm process and the return of major clients, analysts expect this downward trend to reverse in 2026.
Must log in before commenting!
Sign In Sign Up