Skip to Content

Samsung’s $44 B Taylor, Texas Fab: Capacity, EUV Pellicles, and Market Impact

Samsung’s new 5‑million‑sq‑m Texas fab aims for 50,000 wafers per month, introduces EUV pellicles, and could reshape the foundry market against Intel and TSMC.
29 January 2026 by
TechStora Editorial Board

Overview

Samsung announced a $44 billion, 5‑million‑square‑meter semiconductor complex near Taylor, Texas. Construction began in 2022 with production slated for 2024. The site will be the company’s largest logic fab and its first to employ EUV pellicles.

Facility Size and Capacity

The Texas campus exceeds 1,235 acres—significantly larger than Intel’s 700‑acre Silicon Desert campus in Arizona. Samsung targets 50,000 wafer starts per month (WSPM) using the SF2/SF3P process, surpassing Intel Fab 52 (40,000 WSPM) and typical TSMC modules (≈20,000 WSPM).

  • Site area: >5 M m² (1,235 acres)
  • Total investment: $44 billion
  • Initial capacity: 50,000 WSPM
  • Key process: SF2/SF3P for advanced logic chips

Why EUV Pellicles Matter

EU‑V pellicles are ultra‑thin membranes that protect photomasks from particle contamination during exposure. Their adoption has been limited because of high cost, stringent transmission, thermal, and mechanical requirements. Samsung previously avoided pellicles, relying on ultra‑clean handling, frequent mask inspection, and short‑reuse cycles.

Recent advances—ASML’s Twinscan NXE:3600D/3800E tools, metal‑silicide (MeSi) pellicles with superior transmittance, and higher‑throughput EUV sources—make pellicles economically viable.

  • Pellicle cost: $10‑$100 k each
  • Contract: 25 billion won (~$17.5 M) to FST for attachment, detachment, and inspection systems
  • Benefit: Reduces stochastic mask‑borne defects, crucial for 2 nm nodes and large‑die chips like Tesla’s AI5

Economic and Technical Implications

Introducing pellicles improves yield stability, a known weakness of Samsung’s EUV nodes. Stable yields are essential for high‑volume customers (e.g., Tesla) that demand consistent performance across large wafers and many EUV layers.

While pellicle costs increase per‑mask expense, the reduction in defect‑related yield loss and the ability to run higher‑layer EUV designs offset the added overhead, aligning with Samsung’s margin targets.

Competitive Landscape

Samsung’s Taylor fab now competes directly with:

  • Intel Fab 52 (Arizona) – larger footprint but lower capacity (40,000 WSPM)
  • TSMC’s Arizona modules – lower capacity (≈20,000 WSPM) and no pellicle deployment yet

By offering 50,000 WSPM with EUV pellicle protection, Samsung positions itself as the most capable pure‑play logic foundry in the United States.

Outlook

With Tesla’s AI5 and other advanced chips committing to SF2/SF3P, Samsung’s Texas fab is poised to attract additional high‑volume customers seeking reliable EUV production. Successful pellicle integration could also resolve historic yield variability, strengthening Samsung’s market share in the 2 nm and beyond era.