Samsung Foundry 2nm Orders to Double Next Year as Broadcom Negotiations Underway
Miles Bennett
Samsung expects its 2nm design-win count to more than double year-on-year in 2026 and is negotiating with Broadcom and other major clients — the clearest commercial signal yet after years of trailing TSMC in advanced nodes.
2nm wins doubling — what has Samsung actually landed?
Samsung has secured 2nm projects from major cloud providers and AI/HPC customers; design work is already under way.
This means → 2nm is no longer a lab-stage promise — paying customers are committing design resources.
Advanced-node utilisation (8nm and below) has hit full loading, and profitability improved sharply before incentive-related charges.
AI/HPC share jumping from the teens to 30-plus — how is the mix shifting?
Samsung targets double-digit-or-higher foundry revenue growth in 2026, with advanced nodes exceeding 50% of total foundry sales.
AI/HPC's share of the application mix is expected to rise from the high teens in 2025 to over 30% in 2026.
In plain terms = Samsung's foundry revenue is pivoting fast — from "mostly mobile chips" to "mostly AI chips."
The Broadcom talks — why is this the make-or-break proof point?
Samsung is negotiating with Broadcom and other major clients on multiple 2nm engagements, while ramping second-gen 2nm for new mobile products in H2.
Broadcom is one of the world's largest custom-AI-chip designers; which foundry it picks is read by the market as an endorsement of process capability.
This means → whether these talks convert into volume orders is the watershed for Samsung's 2nm commercialisation story.
U.S. fabs and 1.4nm — how is capacity being laid out?
Taylor Fab 1 in Texas remains on track for 2026 operations, with 2nm capacity ramping gradually; Fab 2 breaks ground before year-end, targeting volume production by 2030.
Customer inquiries for 1.4nm are rising; Samsung is evaluating additional fab capacity, with plans to be set in phases tied to order progress.
This reflects Samsung playing two cards — domestic U.S. capacity + next-gen nodes — to win clients sensitive to geopolitical risk.
What makes Samsung claim its model is "unreplicable"?
Samsung highlights its integrated model spanning design, foundry, memory, and advanced packaging, calling this end-to-end capability unmatched by any other chipmaker today.
On mature nodes, strategy is shifting to higher-margin specialty applications: 8nm, 17nm, CMOS image sensors — the core chip in phone cameras — embedded non-volatile memory, and reserved capacity for silicon photonics (next-gen tech that moves data with light instead of electricity).
In plain terms = Samsung's logic is "compete on process at the high end, cherry-pick margins at the low end" — no more chasing volume on price.
Content is for reference only, not financial advice.