Samsung to make AMD and Nvidia HBM parts at second Texas fab, report says
Samsung plans to use its second Texas fab to produce AI memory parts for AMD and Nvidia, according to a report. The plant, set for 2030 production, is booked at full capacity, with most orders from AMD. Samsung declined to comment. The first plant, investing $37B, will start production in 2027, supplying Tesla. Samsung's foundry unit expects profitability by Q4, with 2nm orders projected to grow over 130% this year.
How this was made

The 30-second read
Why it matters
The announcement secures a steady stream of high‑margin orders, improving the outlook for Samsung's foundry earnings and may reduce reliance on TSMC capacity.
Market read
The fab expansion signals a strategic win for Samsung in the AI memory market, with potential upside for its stock and broader implications for the US chip ecosystem.
What to watch
Potential delays in equipment delivery or regulatory hurdles could postpone the 2030 production start.
Background
Samsung's foundry arm has struggled with profitability; the second Texas fab aims to capture AI-driven memory demand.
Market effects
Strengthens the AI‑memory supply chain, may ease capacity constraints for AMD and Nvidia.
Boosts US semiconductor manufacturing footprint and could attract further CHIPS Act incentives.
Highlights continued shift of advanced chip production to the United States, affecting global fab competition.
Counterpoint
If demand for AI memory softens, the fab could become underutilized, pressuring Samsung's margins.
Key entities
- CompanySamsung Electronics
South Korean semiconductor giant launching a second Texas fab.
- CompanyAMD
Primary customer for the new HBM production.
- CompanyNvidia
Secondary customer for the new HBM production.

