- The CHIPS and Science Act’s $52.7 billion in public funding has attracted over $450 billion in private investment across 25 states, with major fab construction now underway at Intel, TSMC, Samsung and Micron sites.
- TSMC’s three Arizona fabs target 4nm production in 2025 and 2nm by 2028, putting leading-edge logic capacity on US soil for the first time at scale, directly relevant to Nvidia and Apple, whose chips are fabbed on TSMC processes.
- Micron’s New York and Idaho buildout aims to lift the US share of advanced memory production from under 2% to around 10% by 2035, with the first Idaho DRAM output scheduled for 2027, a timeline that matters as High Bandwidth Memory demand accelerates.
The US produced close to 40% of the world’s semiconductors in 1990. By 2022, that share had fallen to 12%. The CHIPS and Science Act has since pulled more than $450 billion in private investment into domestic fab construction across 25 states, and the biggest commitments are now moving off the announcement slide and into the ground.
Intel’s $100 Billion US Fab Expansion
The Ohio build is the centrepiece of Intel‘s US expansion: two new leading-edge fabs, each covering around 1.4 million square feet, expected to come online in 2030 and 2032 respectively, producing chips for applications ranging from vehicles to data centres.
TSMC’s Arizona Advanced Chip Hub
Taiwan Semiconductor Manufacturing Company (TSMC)the world’s largest dedicated independent foundry, has committed over $65 billion to three leading-edge fabs in Phoenix, Arizona. The US Department of Commerce awarded TSMC Arizona up to $6.6 billion in direct funding and up to $5 billion in loans. The first fab is on track for 4nm production in the first half of 2025; the second targets 2nm by 2028; a third, also targeting 2nm or more advanced nodes, is expected before the end of the decade. Each fab’s cleanroom area is reportedly double the size of an industry-standard logic facility. The three sites together are expected to create around 6,000 direct manufacturing jobs and more than 20,000 construction roles. The most advanced AI accelerators, from Nvidia to Apple, are fabbed on TSMC processes, so getting even a portion of that capacity on US soil carries real strategic weight. For more on the broader infrastructure investment picture, see our coverage of what’s driving the $5 trillion global AI hardware build-out.
Samsung’s Texas Logic Chip Ecosystem
Samsung Electronics is putting over $40 billion into Central Texas, split between a new complex in Taylor and an expansion of its existing Austin facility, with up to $4.745 billion in direct CHIPS Act funding. The Taylor site is planned to include two leading-edge logic fabs and an R&D facility, targeting advanced logic for mobile, 5G, high-performance computing and AI workloads. Operations are anticipated to begin in 2026, with Samsung targeting 2nm-class manufacturing at the new facility. The expanded Texas footprint is projected to create over 15,000 jobs. Samsung’s foundry business competes directly with TSMC for advanced logic orders, so its US capacity matters to any customer looking to diversify beyond a single supplier.
Micron’s Memory Chip Powerhouse
Memory is where the US supply gap is most acute. Micron Technology has secured up to $6.4 billion in direct CHIPS Act funding to support approximately $150 billion in manufacturing investment and $50 billion in R&D over two decades, split between New York and Idaho. The stated goal is to lift the US share of advanced memory manufacturing from under 2% to around 10% by 2035.
In Clay, New York, Micron plans up to four high-volume manufacturing fabs, each with 600,000 square feet of cleanroom space. In Boise, Idaho, two leading-edge fabs will sit alongside Micron’s existing R&D operations, with DRAM output from the first Idaho facility scheduled for 2027. The Idaho proximity to R&D is deliberate: it’s designed to accelerate time-to-market for High Bandwidth Memory, the stacked DRAM that every major AI accelerator depends on. Across both sites, the investments are projected to create around 23,000 direct jobs. The inference scaling push is intensifying HBM demand further, making Micron’s domestic capacity timeline genuinely consequential.
GlobalFoundries Fortifies Essential Chip Production
Not every critical chip runs on a cutting-edge node. GlobalFoundries (GF) has secured up to $1.5 billion in direct CHIPS Act funding, part of an aggregated investment of over $13 billion, rising to over $16 billion with state support, across its New York and Vermont facilities over the next decade. The focus is on mature-node and speciality processes: automotive, IoT, aerospace and defence, smart mobile devices and data centres.
In Malta, New York, GF is expanding its fab to add process technologies currently produced only overseas, targeting supply security for the US auto industry. In Essex Junction, Vermont, the facility is being modernised to expand production of gallium nitride (GaN) semiconductors, a wide-bandgap material increasingly used in EV power electronics and data centre power conversion. The two projects are expected to generate close to 1,000 direct manufacturing jobs and over 9,000 construction roles.
Texas Instruments Boosts Analog and Embedded Processing
Texas Instruments (TI) is expanding its US manufacturing base with up to $1.6 billion in direct CHIPS Act funding, alongside an expected $6 billion to $8 billion from the US Department of the Treasury’s Investment Tax Credit. The combined support feeds into TI’s plan to invest over $60 billion across seven new 300mm wafer fabs at three sites in Texas and Utah. The direct CHIPS Act funding covers three fabs under construction in Sherman, Texas (SM1 and SM2) and Lehi, Utah (LFAB2), with investment continuing through 2029.
Analog and embedded processing chips are unglamorous but foundational: they handle power management, sensing and motor control in everything from automotive systems to industrial equipment. TI’s facilities are designed to manufacture tens of millions of these chips daily. The programme is projected to create 2,000 company jobs and thousands of indirect roles. Sourcing this category domestically reduces a supply-chain dependency that became painfully visible during the 2020-2021 semiconductor shortage.
The US is projected to reach around 14% of overall global chip manufacturing capacity by 2032, up from 10% in 2022, according to a joint report by the Semiconductor Industry Association and Boston Consulting Group. For more coverage of AI chips and infrastructure, visit our AI Hardware section.
