SpaceX Terafab Could Become a New Variable in Semiconductor Equipment Demand

Overview

This report discusses the potential impact of SpaceX's proposed vertically integrated semiconductor complex, Terafab, on the US semiconductor equipment, foundry, logic, memory, and packaging and testing supply chains. UBS compares Terafab to Tesla's Gigafactory: when external supply chains cannot meet long-term capacity requirements, SpaceX may choose to build integrated capabilities spanning design, manufacturing, and testing iteration to support terawatt-scale AI compute.

Core views

UBS's core view is that the market may be underestimating the actual progress of the Terafab plan and its potential contribution to WFE demand. If SpaceX invests approximately $1.1 trillion in AI business capital expenditure over the next five years, allocates 20% of it to Terafab, and 60% of capital expenditure flows to WFE, Terafab could contribute approximately $135 billion in WFE demand. The report believes the project could begin with a pilot line of approximately $5 billion in 2027, rise to approximately $10 billion in 2028, and reach more than $50 billion annually around 2030/2031, approaching TSMC's WFE spending levels in the coming years. Even if only part of the plan is realized, it would reinforce UBS's long-term view that global WFE could approach $300 billion in 2029.

Analysis framework

The report uses a top-down capital expenditure decomposition, comparisons of capacity and investment intensity across comparable projects, cross-validation of supply-chain orders and hiring signals, and scenario analysis of potential Intel cooperation models and memory supply-chain constraints. UBS derives Terafab capital expenditure from its forecast for SpaceX's AI business capital expenditure and then estimates equipment demand using a typical 60% WFE share. It also references the investment scales of Intel's Ohio project and Boise ID1 fab to infer that the initial Terafab fab cluster could include approximately 80k wsm of memory capacity, two foundry/logic fabs of approximately 20k wsm each, as well as a mask shop and back-end testing and assembly capabilities.

Methodology notes

• Capital Expenditure DecompositionWFE spend bridge

  Estimate WFE demand based on AI business capital expenditure, the Terafab allocation, and the equipment capital expenditure ratio

  UBS starts with approximately $1.1 trillion in SpaceX AI business capital expenditure over the next five years, assumes approximately 20%, or about $225 billion, is allocated to Terafab, and then assumes a typical 60% of capital expenditure flows to WFE, resulting in approximately $135 billion of five-year WFE demand.

• Comparable Project AnalysisGreenfield fab comparison

  Infer Terafab's potential capacity configuration using the investment scales of greenfield fabs such as Intel Ohio and Boise ID1

  The report compares SpaceX's disclosed initial $55 billion semiconductor fab investment in Grimes County, potentially expanding to $119 billion, with projects such as Intel Ohio at approximately $28 billion and Boise ID1 at approximately $15 billion, and infers that Terafab could be a large fab cluster.

• Valuation methodsP/E, EV/FCF, SOTP

  The report discloses the use of valuation techniques including P/E, EV/FCF, and SOTP for relevant companies

  The valuation and risk disclosures state that relevant company valuations use methods including P/E and EV/FCF. Intel Corp.'s target price is based on a SOTP valuation of Intel Products, Intel Foundry, Altera, Mobileye, and All Other, with a customized P/S multiple applied to F2026E revenue.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Semiconductor equipment suppliers

  Potential largest direct beneficiaries

  Strengths

  Terafab could generate substantial incremental WFE demand and support an upside scenario in which global WFE approaches $300 billion in 2029.

  Weaknesses

  Demand realization depends on SpaceX project progress, financing, and implementation of the process roadmap.

  Comparison

  If annualized spending exceeds $50 billion, the incremental scale could be comparable to TSMC's WFE spending over the next several years.

  Risks

  Order delays, project downsizing, changes in the technology roadmap, or supply-chain execution failure.

• Intel Corp.

  Potential technology and manufacturing partner

  Strengths

  May provide process flows, manufacturing IP, PDKs, design rules, tool recipes, and operational know-how, and benefit through licensing or royalties.

  Weaknesses

  The cooperation structure has not been confirmed, while Intel itself continues to face competitive pressure from AMD and NVDA.

  Comparison

  The report compares the potential technology licensing model with the historical IBM-AMD technology transfer framework and mentions Ohio One site as a potential Terafab campus scenario.

  Risks

  Terafab pilot failure, unfavorable cooperation terms, weak execution of Intel's capacity expansion, and insufficient competitiveness in advanced processes.

• Memory suppliers

  Affected by medium- to long-term competition and investment reallocation

  Strengths

  If Terafab drives domestic US front-end manufacturing, Korean memory companies could increase investment in US DRAM and other facilities.

  Weaknesses

  Existing memory suppliers have limited incentives to license core IP to a potential competitor.

  Comparison

  The report notes that Samsung owns land in Taylor, TX, where additional capacity could be added.

  Risks

  If a new supplier successfully achieves large-scale memory production, it could alter the existing supply landscape; if the IP source cannot be resolved, Terafab's memory plan could be delayed or impeded.

• SpaceX

  Terafab initiator and vertically integrated AI compute entity

  Strengths

  Vertical integration could shorten the design/build/test/redesign cycle and support the vision of terawatt-scale AI compute.

  Weaknesses

  Capital intensity is extremely high, and significant uncertainty exists regarding semiconductor manufacturing experience, IP, supply chains, and mass-production yields.

  Comparison

  UBS compares the strategy with Tesla's construction of the Gigafactory when battery supply was insufficient.

  Risks

  Dependence on Elon Musk's decisions, delays in technical milestones, weaker-than-expected AI commercialization, regulatory and geopolitical pressure, and an unproven space-based compute model.

Key data

• Five-year capital expenditure forecast for SpaceX's AI business~$1.1TForecast by UBS's John Hodulik; covers only the AI business and excludes Connectivity and Launch segments.
• Five-year Terafab capital expenditure estimate~$225BApproximately 20% of SpaceX's AI business capital expenditure.
• Five-year Terafab WFE spending estimate~$135BBased on a typical 60% allocation of capital expenditure to WFE; comparable in scale to this year's global WFE TAM.
• 2027 pilot line scale~$5BUBS believes, based on equipment supplier feedback, that orders for a pilot line already exist, with a scale of approximately $5 billion.
• 2028 Terafab WFE scale~$10BIncluded in UBS's approximately $250 billion WFE assumption for 2028.
• Potential annualized WFE in 2030/2031>$50B/yrUBS says this would be similar in scale to TSMC's WFE spending in the coming years.
• Disclosed initial SpaceX semiconductor fab investment$55B initial; up to $119B if all phases completedLocation: Grimes County, TX.
• Potential capacity structure~80k wsm memory + two ~20k wsm foundry/logic fabsCould also include mask shop and back-end testing and assembly capabilities.

Impact & implications

If Terafab progresses as UBS assumes, the medium-term demand ceiling for the semiconductor equipment industry would rise materially, with SPE suppliers benefiting most directly. Intel could also gain opportunities in technology licensing, foundry cooperation, or a JV, while memory suppliers would face potential pressure from a new entrant and a reallocation of US front-end manufacturing investment. The project could also lead the market to focus more closely over the next several earnings seasons on SpaceX orders, toolmaker slots, construction progress in Texas, and details of cooperation with Intel.

Risks

• A macroeconomic downturn, international trade disruptions, technological disruption, and business-model innovation could alter semiconductor industry volume, ASP, and revenue trajectories.
• Terafab is extremely capital intensive, and the project could face construction delays, cost overruns, supply-chain execution failures, or failure to achieve technical milestones.
• The source of memory IP is unclear, and existing memory suppliers have limited incentives to license core IP to a potential competitor.
• The cooperation structure between Intel and SpaceX has not been confirmed; technology licensing, a JV, a dedicated site, or an equity investment remain scenario possibilities.
• SpaceX's AI business and space-based compute business model have not been sufficiently validated, and future capital expenditure and demand could fall below UBS's assumptions.
• Regulatory, geopolitical, and competitive pressures, as well as dependence on Elon Musk as a key decision-maker, could affect project execution.

What to watch

• Public filings, construction permits, hearings, and changes in investment phases related to Grimes County, TX.
• Hiring of SpaceX Terafab executives, the number of process engineering positions, and changes in specific responsibilities.
• Equipment supplier orders, toolmaker slot reservations, and the delivery schedule for the pilot line.
• Whether Intel and SpaceX disclose details of technology licensing, PDKs, process recipes, the Ohio One site, or JV cooperation.
• The source of memory IP, DRAM or other memory process roadmaps, and US front-end manufacturing investment plans by suppliers including Samsung, SK Hynix, and MU.
• Comments from semiconductor equipment companies over the next several earnings seasons regarding Terafab orders and WFE demand visibility.