Samsung Taylor Fab 2 Construction Starts November, Targets 2030

A second fab before the first one is finished

Samsung Electronics says it will start building a second semiconductor fab at its Taylor, Texas campus next month. Its first plant there is still working through qualification and has not reached full operation. Jonathan Taylor, who runs Samsung's Austin and Taylor sites as head of infra technology, made the announcement at an October 15 event. It was also the first time reporters had been allowed inside the facility.13 He said the first fab remains "on track" to come online by the end of 2026, about five years after Samsung first announced the Taylor project at $17 billion.1

The second building, called Fab 2, will go up next to the first one and is expected to be finished in 2030.4 Samsung describes it as half again as large as Fab 1.3 Industry reporting puts it at roughly 2.7 million square feet.59 The local business press compares its production floor to about nine football fields, against six for the first building.52 Samsung says the campus should support about 3,500 direct jobs once both plants are running.3

Samsung also pledged $22.84 million for nonprofits, schools and workforce training in Central Texas. That includes about $10.7 million for roughly 14 organizations in Taylor and nearby Manor, plus $10.64 million to extend its workforce partnerships with local school districts and Texas universities.18 A separate $1.5 million will fund a skilled-trades program with the Taylor school district covering electrical, HVAC, welding and plumbing certifications.1 Those are the trades a fab construction site and its later operations need most.

Why now: demand, not just ambition

Building a second fab before the first one has run at volume looks aggressive, but it follows a pattern Samsung has been describing since summer. On its second-quarter 2026 earnings call in July, the company said its leading-edge foundry capacity was not keeping up with demand. It also confirmed then that Fab 2 construction would start this year, with mass production by 2030.7 Coverage from Korea adds that Samsung's advanced lines at Hwaseong and Pyeongtaek, which make chips at 2 to 4 nanometers, are running at capacity. That puts pressure on the company to speed up its timetable in Texas.19

The larger competitive picture points the same way. TSMC has reportedly committed most of its advanced capacity as AI chip orders pile up, and Samsung sees a chance to take the overflow.19 One commentator argues that grabbing land and supply-chain commitments now is a deliberate hedge, and that Samsung treats its U.S. base as essential to answering TSMC's growth in Arizona.6 The site leaves room for more: Samsung owns over 1,200 acres near Taylor and could eventually build as many as 10 fabs there.1

Preparation is visibly underway. Samsung has reportedly asked several equipment suppliers to get SEMI safety certifications ahead of time for tools bound for Fab 2, even though the building's detailed specifications are not final.15 Much of the groundwork was finished months ago. In March, the Taylor city council extended its contract with an engineering firm to handle permit reviews and construction inspections.7

Where the reporting diverges: how far along is Fab 1?

The biggest disagreement in the coverage is over the first fab's status, and it matters because Fab 2's case rests on Fab 1 succeeding.

Samsung's own description to Texas reporters is cautious. Executives said installed tools are starting up and running processes. The plant is entering a qualification sequence that covers the tools first, then the factory, then the product, and Taylor himself called that a lengthy process.1 Industry outlets describe the plant the same way: it is going through staged qualification, aims to be operating by the end of 2026, and expects volume production in 2027.9 DigiTimes reported in mid-September that trial production of Tesla's AI5 processor had started, which puts the fab in 2nm wafer production and yield verification.14

Seoul Economic Daily went further. It reported that Taylor had entered partial mass production on 2nm, that utilization had passed 20% in August and neared 30% in September, and that Samsung wanted to get close to 100% by year-end.19 That account is much more bullish than anything Samsung said on the record in Texas. One summary of the September reports also says the wafers were prototypes intended to confirm process stability, and that final mass-production verification is targeted for the end of 2026.16

The more reliable reading is the conservative one. On-the-record statements from the site point to qualification work now and customer volume in 2027. The Korean utilization figures most likely describe early wafer runs, not true high-volume output. That fits what Samsung Foundry vice president Margaret Han said in May, that customers would start production at Taylor in 2027.12

The customer question

Customers have been Taylor's weak point for years. The original target was around 2024 output. It then moved to 2025, then to the second half of 2026, and then to early 2027. Analysts note that the delays followed customer commitments more closely than construction progress.13 Samsung also dropped its original 4nm plan for the site and went straight to 2nm gate-all-around technology, which made the job harder.12

The situation changed with Tesla's $16.5 billion contract, signed in July 2025 and running through 2033.512 Tesla is still the only customer Samsung has officially named.1 Samsung is expected to build both Tesla's AI6 and its AI5 on 2nm in Texas.9 Samsung declined to name anyone else, saying only that many companies have shown interest.1 Korean press reports go further: one says Fab 1 is already fully booked for 2nm, with orders linked to Tesla, Broadcom communications chips and Arm AI chips for smart devices. It cites an initial capacity of 50,000 wafer starts per month.15 None of those names beyond Tesla has been confirmed, so they should be treated as unverified until a customer or Samsung says so.

Tesla's plans add a complication. Elon Musk has announced Terafab, a chip fab to be developed jointly by Tesla and SpaceX. Samsung's site chief says that does not affect the Taylor deal, arguing that demand is high enough for Samsung to try to deliver chips sooner.1 Earlier reporting said that when Tesla approached Samsung about Terafab, Samsung offered extra capacity at Taylor rather than joining the project directly.18 Fab 2 looks partly designed to make that offer believable.

Money, process and what remains unknown

Reported investment figures vary. Several outlets cite more than $37 billion across Central Texas, covering two advanced fabs, an R&D center in Taylor and expansion of Samsung's existing Austin operations.79 That spending is backed by up to $4.745 billion in CHIPS Act funding.9 Other coverage calls Taylor a $44 billion megafab.13 The gap probably comes from what each figure includes, but it shows how much the stated cost has grown from the original $17 billion.1

Samsung has not said which process Fab 2 will run.9 Korean industry watchers expect it to use a next-generation 1.4nm node.19 That fits reporting that customer inquiries about 1.4nm have been increasing, even though Samsung was reportedly thinking about shelving the node last year.9 For now, the 1.4nm plan is an expectation, not a commitment.

Yield is the main risk. Samsung's 2nm process held up Tesla's roadmap earlier this year, when AI6 was reported to have slipped about six months.11 Samsung's foundry also still trails TSMC by a wide margin overall.17 Counterpoint estimates that 2nm wafers cost around 30% more than 3nm. That premium makes fabless customers such as Qualcomm and MediaTek cautious until they are sure Samsung's yield recovery will last.12

The bottom line

Starting Fab 2 now is a reasonable bet, not a reckless one. Samsung holds a large anchor contract, its Korean advanced lines are full, its main rival is reportedly sold out at the leading edge, and its Texas site has room to grow. Even so, the 2030 date should be read as a goal. Fab 1 took several years longer than first promised, and one industry analysis expects Fab 2's schedule to depend heavily on demand.9 The steel going up next month shows Samsung is serious about Texas. Whether that pays off will depend on Fab 1 reaching high-volume 2nm output in 2027 at yields that attract customers beyond Tesla.