Records Detail Terafab’s Conditional $119 Billion, Four-Phase Buildout Through 2036

Public filings reveal that Terafab could grow through four construction phases into a $119 billion semiconductor complex by 2036 far beyond the confirmed $16.8 billion first phase announced for Grimes County, Texas. The newly documented scale, detailed by Engineering News-Record, describes a possible decade-long buildout rather than one fully committed project.

Each phase would be able to manufacture semiconductors independently, and the applicant would decide after completing one phase whether to build the next. The later investments therefore remain conditional. The filings also expand the known construction scope to include onsite power generation, water treatment and recycling, industrial wastewater treatment, and obligations to maintain and repair county roads used by construction traffic.

That distinction matters when evaluating both the project’s industrial scale and its public infrastructure burden. The $119 billion figure represents the upper boundary if all four phases proceed, not capital that has already been committed to a single continuous construction program. A Grimes County notice estimated $55 billion for the initial phases and as much as $119 billion with the additional expansion.

Four separate investment decisions

Phase 1 is projected for construction from 2026 through 2028 and accounts for approximately $16.8 billion in eligible investment across two local school districts. Phase 2, projected for 2029 through 2031, would add about $38.6 billion. Phase 3, listed for 2032 through 2034, carries another $38.6 billion, while Phase 4 is projected for 2035 through 2036 at approximately $25.3 billion.

Those figures explain how the program reaches its headline scale, but the authorization structure sets an important boundary. An independently operable fab can add capacity without requiring the full campus to be completed. It also allows the companies behind Terafab to stop, defer or modify subsequent expansion after assessing demand, process performance and the output of an earlier phase.

The filings describe a campus intended to combine integrated-circuit design, photomask production, wafer fabrication, advanced packaging and system-level integration. SpaceX, Tesla and xAI are identified as likely users of its output. Terafab plans to employ Intel’s 14A manufacturing technology, a 1.4-nanometer-class process using conventional and next-generation extreme-ultraviolet lithography systems.

The fab extends well beyond its cleanrooms

The disclosed property schedules show why a semiconductor campus cannot be evaluated only by the size or cost of its fabrication buildings. Listed systems include cleanrooms, sub-fabs, a central utility plant, substations, cooling equipment, transformers, backup generation, battery storage and infrastructure for compressed gases and chemicals.

Under the county’s executed tax-abatement agreement, the facilities are expected to use electricity from onsite power plants rather than the grid. Plans call for surface water from Gibbons Creek Reservoir instead of groundwater, along with onsite water reuse and wastewater treatment. A separate agreement designates six county roads for construction traffic and makes the company responsible for maintaining those routes during construction and repairing related damage.

These systems are not secondary conveniences. Advanced chip production depends on stable electrical power, tightly controlled process water, thermal management and waste handling. Building them alongside the fab may give Terafab greater control over critical inputs, but it also increases the commissioning scope that must be completed before useful manufacturing output can begin.

Construction progress is not production readiness

The public filing schedule says first-phase commercial operations are expected in 2028, but an outside forecast from SemiAnalysis places initial operation in late 2029 and volume output around mid-2030. That analysis allocates 20 months to building construction, 17 months to equipment installation and six months to ramping production.

The gap illustrates why visible construction does not translate quickly into chips. Completing the building shell is only one stage; lithography and other process tools must then be delivered, installed, connected, calibrated and qualified before production can ramp. An ASML extreme-ultraviolet scanner is reportedly scheduled to arrive in mid-2027, potentially before the projected late-2027 completion of the development-fab shell, which could leave the machine temporarily unused.

Intel is targeting high-volume manufacturing with 14A in the second half of 2028 if its schedule holds, while TSMC intends to begin volume production of its 1.2-nanometer-class A12 process in late 2029. Terafab’s value proposition is therefore not simply being first to a leading process. It is the prospect of bringing design, fabrication and packaging closer to the companies expected to consume the chips.

For now, the firm commitment remains Phase 1. The $119 billion campus depends on three later construction decisions, and even the first useful volume output may remain several years beyond the earthwork now underway.

By Thomas Caldwell — AMI’s senior editor for mechanical and mobility engineering, covering vehicle electronics, systems integration, electrification, chassis systems, propulsion, and safety policy.