Taiwan Semiconductor Manufacturing Co. is evaluating a large new facility in Texas, signalling an expansion driven by soaring AI chip demand and ongoing efforts to diversify manufacturing amid industry capacity shortages and geopolitical pressures.
Taiwan Semiconductor Manufacturing Co. is weighing a large new manufacturing campus in Texas, according to Bloomberg, in a move that would deepen its expansion in the US and underline how strongly AI demand is shaping semiconductor investment. The plan is still under review and has not been approved publicly, but the fact that TSMC is even studying a second American mega-site points to a capacity crunch that has already pushed its Arizona programme far beyond its original scale. Bloomberg reported that the Texas idea would involve multiple fabrication plants on one site rather than a single standalone fab.
Arizona remains the clearest example of how quickly that programme has expanded. TSMC’s US commitment reached $265 billion in July 2026 after the company announced an additional $100 billion investment in the state, including plans for four more advanced chip plants, according to the City of Phoenix and TSMC. Tom’s Hardware reported that chief executive C.C. Wei framed the move as an expansion of the company’s US manufacturing footprint, not a replacement for earlier plans. The Arizona site now includes multiple fabs already in production or under construction, with the first fab having begun high-volume output on N4 technology in late 2024, the second fab finished in 2025 and scheduled for N3 production in 2027, and a third fab under construction for N2 and A16 technologies later this decade, according to TSMC’s own Arizona overview.
That timeline matters because it shows TSMC is building far ahead of current demand. The company’s Arizona programme began as a $12 billion pledge in 2020, then moved through successive increases as orders for advanced logic chips continued to climb. A second Arizona fab was added in 2022, and a third followed in 2024. By March 2025, TSMC and the White House announced another major expansion, bringing the total much higher before the July 2026 update lifted the figure to $265 billion. In practical terms, each new commitment has reflected the same problem: demand for leading-edge capacity has continued to outrun the industry’s ability to deliver it.
The customers driving that demand remain crucial. Nvidia and Apple are among TSMC’s most important clients, and both depend on the foundry’s most advanced manufacturing nodes rather than general-purpose chip capacity. That makes TSMC’s location decisions unusually sensitive to the roadmaps of a small number of large buyers. If those customers want more wafers, TSMC has to secure more advanced production space, which explains why a Texas campus, if approved, would almost certainly be designed around the same sophisticated process technology being deployed in Taiwan and Arizona.
A second US site would also fit TSMC’s broader strategy of reducing geographic concentration. For most of its history, the company kept its most advanced manufacturing in Taiwan. That model is now being tested by customer pressure, US industrial policy and the need for more resilient supply chains. A Texas campus would spread risk across two states, but it would also be expensive and slow to execute. Advanced fabs require long construction schedules, extensive utility build-outs and specialised equipment, so even a formal decision would not translate into new output for several years.
TSMC would not be entering an empty field. Samsung Foundry has been building in Taylor, Texas, while Intel has committed large sums to projects in Ohio and Arizona. Yet both rivals have faced slower progress than TSMC, which has already put its first Arizona fab into production. Samsung’s Taylor project covers more than a single plant and broader regional development, but it still has not matched the scale of orders that TSMC has secured from leading-edge customers. Intel’s Ohio project, meanwhile, has been delayed and is now generally expected much later than originally planned.
The timing of any Texas decision is also shaped by federal support. The US CHIPS Act remains active, but funding has moved more slowly than political messaging suggested, with the Government Accountability Office saying in August 2026 that the Commerce Department had awarded about $31.7 billion across 24 companies and 49 projects, while only about $13.1 billion had been disbursed by April. The Congressional Research Service said in July that several companies with preliminary memoranda still had not received final awards. Any TSMC incentive package for Texas would probably combine federal support, state tax breaks and direct negotiation, much as previous Arizona expansions did.
For markets, the immediate significance is not cheaper chips but a clearer signal that tight supply is likely to persist. A Texas campus would not ease conditions this year or next, because even TSMC’s Arizona build-out is still moving through phased production. But the report suggests the company expects AI-related demand to remain elevated well into the next decade. For buyers of high-end GPUs, processors and AI accelerators, that means the real constraint is still advanced manufacturing capacity, not a temporary shortage. In that sense, a possible Texas mega-site is less a near-term fix than another sign that the semiconductor industry is now planning around sustained AI-driven demand.
Disclaimer: This content is intended for informational purposes only. Readers are advised to exercise their own judgement, conduct due diligence, or consult a qualified expert before acting on any information provided.





