Samsung's $17 billion semiconductor plant in Taylor, Texas, has cleared its most significant credibility hurdle since construction began in early 2022. On May 28, Margaret Han, vice president of Samsung Electronics Foundry's North American operations, told attendees of the Samsung Advanced Foundry Ecosystem (SAFE) Forum 2026 at the company's U.S. headquarters: "Customers are scheduled to begin production at the Taylor fab starting next year. We are ready." That two-word declaration, the clearest commitment Samsung has made publicly about the plant's production timeline, landed three months after unnamed industry sources told South Korean media that even Samsung's internal schedule for full-scale production had not been finalized.
For chipmakers evaluating their next manufacturing partner, the statement changes the calculus. After several delays that stretched the plant's launch window from the originally planned second half of 2024 to the present, a named executive confirming a 2027 start on the record is a different kind of signal than quarterly earnings-call hedges.
Samsung Taylor Fab 2nm Production Timeline: What Changed
The plant's history reads as a case study in the gap between infrastructure ambition and yield reality. Samsung announced the Taylor facility in November 2021 with a $17 billion initial investment, promising production by the second half of 2024. That date slipped to 2025, then to the second half of 2026, and finally to early 2027 — the last shift driven partly by Samsung's decision to abandon the originally planned 4nm process and jump directly to 2nm, a transistor architecture that relies on Gate-All-Around technology Samsung has been refining since its 3nm production runs began.
The yield numbers tell the more honest story. When Samsung began mass-producing Exynos 2600 chips on its first-generation SF2 process in September 2025, analysts pegged the yield at roughly 50 percent — meaning half of the wafers produced met spec. Industry analysts who track foundry economics generally set 60 percent as the floor for commercially viable runs. By early 2026, yield had reportedly climbed toward 55–60 percent on the first-generation SF2 node and higher on the improved SF2P variant.
What SF2P+ Brings to Taylor Fab
The process Han referred to at the SAFE Forum is SF2P+, the third generation in Samsung's 2nm roadmap. Samsung's 2nm node family runs from SF2 (first generation) to SF2P (second generation) to SF2P+ (third generation) to SF2X, a node aimed at AI and high-performance computing that sits further out on the roadmap. Each generation improves on what engineers call power, performance, and area — the three levers that chip designers pull when choosing a process.
SF2P+ applies an Optic Shrink technique to the SF2P design. Samsung demonstrated the technology at a SAFE Forum for partners in July 2025, where the third-generation node was reported to deliver a 20–30 percent performance improvement over its predecessor. Samsung's proprietary Multi-Bridge Channel FET architecture, which it uses instead of the FinFET transistors favored by TSMC through its 3nm generation, enables the nanosheet-width adjustments that underpin these efficiency gains. The result is a process designed for workloads where power draw per operation is as critical as raw throughput — particularly the autonomous-driving and AI inference chips that dominate Samsung Foundry's confirmed customer base.
The foundation for that customer base is a $16.5 billion contract with Tesla signed in July 2025, under which the Taylor facility will manufacture Tesla's AI5 and AI6 automotive chips through 2033. Tesla's AI5 production is split between Samsung and TSMC, while the AI6 chip — covering autonomous driving, robotics, and AI data center workloads — is allocated entirely to Samsung's 2nm process at Taylor.
When Will Samsung Foundry Next Share Process Details With Partners?
Samsung plans to hold the Seoul edition of the SAFE Forum in July 2026 and is expected to convene its Samsung Foundry Forum — the event at which it lays out the full technology roadmap for foundry partners — in the same month. Han said at the May 28 forum that the company will share detailed information on leading-edge processes with partners in July. Those two events will be the next opportunity for the industry to cross-check the Taylor readiness claim against hard process specifications.
Counterpoint Research director Kyeong-su Kang noted earlier in May 2026 that Samsung is positioned to expand foundry volume as 2nm capacity at Taylor comes online, though Counterpoint estimates 2nm wafer costs run approximately 30 percent above 3nm production — a pricing premium that puts pressure on fabless customers such as MediaTek and Qualcomm, which are still weighing whether Samsung's yield recovery is durable enough to justify the higher cost.
The Taylor fab has now passed the equipment installation phase that once seemed uncertain. An April 16, 2026 report confirmed a ceremony scheduled for April 24 at Taylor Fab 1, drawing Foundry Business President Han Jin-man and a cluster of global equipment and materials partners. Samsung's Korean supply-cluster firms — including Dongjin Semichem, Soulbrain, FST, and Hanyang ENG — had established production and office facilities in Taylor through 2025, converting a single-plant story into a long-term U.S. foundry play. The remaining question is not whether the facility exists, but whether 2nm yields will cross the 60 percent threshold that Samsung's own foundry analysts identify as the line between demonstration-scale and commercially viable mass production.
Frequently Asked Questions
When will Samsung Taylor fab start production?
Samsung Foundry Vice President Margaret Han confirmed at the SAFE Forum on May 28, 2026, that customers are scheduled to begin production at the Taylor, Texas plant starting in 2027. Samsung expects to have 2nm production capacity established at Taylor Fab 1 within 2026, with customer production runs following in 2027.
What chips will Samsung make at Taylor Texas?
The confirmed anchor customer is Tesla, whose AI5 and AI6 autonomous-driving processors are covered by a $16.5 billion deal signed in July 2025 running through 2033. Tesla's AI5 production is split between Samsung and TSMC, while the AI6 chip is allocated entirely to Samsung's 2nm Gate-All-Around process at Taylor.
What is the SF2P+ process node?
SF2P+ is the third generation of Samsung's 2nm process family. It applies an Optic Shrink technique to the second-generation SF2P design and delivers a 20–30 percent performance improvement over SF2P. The node uses Samsung's Multi-Bridge Channel FET architecture and is tuned for AI workloads and high-performance computing. Samsung demonstrated it at a SAFE Forum in July 2025 and plans to install it at Taylor Fab 1 in 2026.
Why has Samsung Taylor fab been delayed?
The Taylor fab missed its original late 2024 production target for two main reasons: Samsung switched from a 4nm process to a 2nm Gate-All-Around process, which carries more manufacturing complexity, and 2nm yields took longer than projected to reach commercially viable levels. Yields on the first-generation SF2 node were estimated at approximately 50 percent when mass production began in September 2025, below the 60 percent threshold that most foundry customers consider a minimum for stable supply.
ⓒ 2026 TECHTIMES.com All rights reserved. Do not reproduce without permission.
