Is TSMC beating Intel for Musk’s chip plant?

Is TSMC taking Intel's place in Elon Musk's Terafab project?

TSMC is now in talks to join Elon Musk's Terafab chip project in Texas, but Intel still holds the clearer position as its announced technology partner.

Intel joined Terafab on April 7, 2026, offering expertise in chip design, manufacturing and advanced packaging. Later that month, Musk confirmed that the project planned to use Intel's next-generation 14A manufacturing process.

The situation changed when Musk acknowledged discussions with TSMC. On October 7, he said his companies would build and operate Terafab themselves, suggesting TSMC could potentially lease part of the site. Then, on October 15, TSMC CEO C.C. Wei confirmed that his company had sent teams to Texas and was exploring cooperation with the project.

Wei insisted that any factory using TSMC's manufacturing processes would have to be run by TSMC.

That's where the disagreement stands. Intel has an announced partnership, TSMC wants a possible manufacturing role, and Musk wants control of the main factory. No TSMC agreement has been announced, and Intel has not been removed from the project.

Why is Musk talking to TSMC when he already has Intel?

Musk is talking to TSMC because Terafab needs enormous chipmaking capacity, and relying entirely on Intel's still-developing 14A process would expose the project to considerable manufacturing risk.

Terafab was announced in March 2026 to supply chips for Tesla's autonomous vehicles and Optimus robots, as well as computing hardware for SpaceX and xAI. Musk argued that existing semiconductor suppliers would struggle to keep up with his companies' future requirements.

Intel became the first publicly announced manufacturing technology partner in April. Around the same period, Bloomberg reported that Musk's team had contacted major semiconductor equipment suppliers, including Applied Materials, Lam Research and Tokyo Electron.

TSMC entered the picture later, as the Taiwanese chipmaker was considering additional manufacturing capacity in the United States.

Intel brings a manufacturing process that Musk wants to use in his own facilities. TSMC could potentially provide separately operated production capacity using its established manufacturing technologies.

Those arrangements could coexist. We have no confirmed evidence that Musk approached TSMC because Intel had missed a technical milestone or failed to meet a contractual requirement.

Are Musk and TSMC fighting over who controls Terafab?

Musk and TSMC currently disagree over who would operate any TSMC-related manufacturing facility at Terafab, and neither side has publicly backed down.

On October 7, Musk said his companies would build and run the factory. He suggested TSMC might sublease part of the site, but ruled out a broader operational role.

TSMC CEO C.C. Wei offered a different position during the company's third-quarter earnings conference on October 15. He confirmed that discussions were continuing and explained that any fab using TSMC's manufacturing processes would need to remain under TSMC's operational control.

There is a practical reason for TSMC's position. Advanced chip manufacturing depends on tightly controlled production recipes, equipment settings, process integration and quality systems developed over many years. Handing these responsibilities to a new operator would be a major departure from how TSMC runs its business.

The disagreement could be resolved by dividing the complex into separately operated facilities. Musk's companies could control their own production lines while TSMC manages a different manufacturing area.

That arrangement remains hypothetical, though. Neither company has announced an operating agreement, ownership structure or production allocation.

If you want more recent data on this point, please see our latest semiconductor industry report.

Has Intel already won the most important part of Terafab?

Intel has secured the strongest publicly announced technology role in Terafab, with Musk explicitly choosing Intel 14A as the intended manufacturing process. Whether that becomes a profitable production business remains open.

Intel 14A is the company's next-generation leading-edge process, following Intel 18A and 18A-P. During Tesla's April 2026 earnings discussion, Musk said 14A could be sufficiently mature by the time Terafab reached substantial production.

That was a major endorsement for Intel. The company has spent years trying to build a foundry business capable of attracting large outside customers, a market where TSMC dominates.

Intel's second-quarter regulatory filing showed continued development of 14A and progress toward design and performance milestones being evaluated by potential major customers.

However, the public announcements do not specify guaranteed Terafab wafer orders, minimum purchase volumes or contract revenue.

Intel has persuaded Musk to back its technology, which is already a meaningful achievement. The next challenge is proving that 14A can produce the required chips at competitive yields and costs.

Is TSMC still much better than Intel at making advanced chips?

TSMC remains far ahead of Intel in commercial manufacturing for outside customers, even as Intel makes progress with its latest production technologies.

TSMC reported $46.72 billion in third-quarter 2026 revenue, up 41.2% year over year in US dollars. Its gross margin reached 66.4%, while its operating margin stood at 58.7%.

More revealing is the company's advanced-chip revenue mix. Its 2-nanometer process accounted for 8% of wafer revenue, 3-nanometer contributed 33%, and 5-nanometer another 28%. Including 7-nanometer technology, advanced processes generated 79% of wafer revenue.

Intel's second-quarter 2026 regulatory filing tells a very different story. Intel Foundry recorded $5.765 billion in revenue, but approximately $5.5 billion came from transactions with other Intel businesses. External customers contributed $293 million.

In other words, approximately 95% of Intel Foundry's reported segment revenue still came from Intel itself.

The foundry segment also posted a $2.089 billion operating loss, compared with a $3.168 billion loss a year earlier. Intel is improving, but it has yet to demonstrate TSMC's combination of external-customer scale and profitability.

The figures cover different reporting periods, and Intel's segment revenue includes internal transactions while TSMC's revenue is consolidated. They still reveal a huge difference in the maturity of the two foundry businesses.

Manufacturing indicator	TSMC	Intel
Reporting period	Q3 2026	Q2 2026
Reported revenue	$46.72B company-wide	$5.765B foundry segment
External foundry revenue	Core business	$293M
Operating result	58.7% margin	$2.089B segment loss
Leading-edge process	N2 producing commercial revenue	18A ramping; 14A in development
Advanced-process revenue	79% from 7nm or better	No equivalent published mix
Customer mix	Major outside chip designers	Mostly Intel's own businesses
Terafab involvement	Confirmed discussions	Announced technology partner

If you want more recent data on this point, please see our latest semiconductor industry report.

Could Intel 14A beat TSMC's next-generation chips?

Intel 14A could become a serious competitor to TSMC's advanced manufacturing processes, but we still don't have production data proving that Intel can make better or cheaper chips.

Intel is building on technologies introduced with 18A, including RibbonFET transistors and PowerVia backside power delivery. The company also plans to use high-numerical-aperture extreme ultraviolet lithography in its advanced manufacturing roadmap.

TSMC is following a different path. Its N2 process is already generating revenue, while A16 and A14 represent further generations of manufacturing improvements.

The confusing part is that Intel 14A and TSMC A14 have similar names. Those names don't tell us which process has better transistor density, manufacturing yields, power consumption or cost per usable chip.

TSMC also appears less eager to introduce high-NA EUV immediately. During its October 2026 earnings conference, the company said it expected to begin using the technology for mass production around 2030, with equipment maturity and manufacturing economics guiding its decisions.

Intel could gain an advantage by introducing certain technologies earlier. But advanced manufacturing rewards reliable execution as much as technical ambition.

We would need comparable chips manufactured on both processes, along with credible yield, performance and cost figures, to establish a winner. Intel 14A has not reached that stage.

Is TSMC moving to Texas because of Musk's Terafab?

TSMC is seriously considering additional manufacturing facilities in Texas, but its expansion plans are much broader than Musk's Terafab project.

During TSMC's October 15 earnings conference, CEO C.C. Wei confirmed that company teams had visited Texas to evaluate possible locations.

He explained that major American customers were asking TSMC to increase domestic manufacturing capacity. The company is already building a substantial manufacturing presence in Arizona, with announced US investment commitments totaling $265 billion.

Texas would provide another American location, potentially closer to customers and semiconductor manufacturing partners already operating there.

TSMC also expects enormous investment requirements across its existing business. The company maintained a 2026 capital expenditure budget of $60 billion to $64 billion and anticipated greater spending over the following two years.

Terafab's first announced phase involves more than $16.8 billion in planned investment, so a potential cooperation agreement could be significant. But the two figures measure different things: TSMC's worldwide annual investment budget and the planned cost of one Terafab development phase.

TSMC's interest in Texas makes commercial sense independently of Musk. That gives the company room to negotiate without depending on Terafab for its American expansion.

Is Musk really building one chip factory or several?

Musk's Terafab project now involves at least two distinct manufacturing locations, making it easier to understand why Intel and TSMC might eventually have different roles.

SpaceX confirmed in August 2026 that construction of a semiconductor research facility had begun at Tesla's Giga Texas site in April.

The larger commercial complex is planned for Grimes County, near the former Gibbons Creek power station.

According to the Texas governor's office, the first Grimes County phase involves more than $16.8 billion in planned investment and approximately 3,000 jobs. The announcement also described an ambitious 100-million-square-foot development.

Terafab aims to bring logic manufacturing, memory production and advanced packaging together within the same industrial complex. The chips would serve different needs, from Tesla's vehicles and humanoid robots to AI computing systems.

Musk added more detail during SpaceX's October 20 earnings discussion. He said the Austin research facility was targeting its first functional test chip around May or June 2027, while the main Gibbons Creek plant was targeting commercial volume production in the third quarter of 2028.

The distinction leaves room for several possible manufacturing arrangements.

Project component	Planned role	Publicly reported status
Giga Texas research fab	Test semiconductor manufacturing	Construction began April 2026
Initial research chip	Validate manufacturing equipment and processes	Targeted May–June 2027
Grimes County Terafab	Main commercial manufacturing complex	First phase announced
Initial investment	Finance major factory development	More than $16.8B
Jobs	Engineering, fabrication and operations	Approximately 3,000 planned
Logic manufacturing	Produce advanced computing chips	Planned
Memory manufacturing	Supply memory for computing systems	Planned
Advanced packaging	Integrate and connect chip components	Planned
Commercial chip production	Manufacture at industrial scale	Targeted Q3 2028
Potential TSMC facility	Additional independently operated capacity	Discussions only

Could Terafab become a major business for TSMC?

Terafab could eventually become a valuable customer or manufacturing partnership for TSMC, but the project has not disclosed enough production commitments to measure the opportunity.

The planned first phase exceeds $16.8 billion in investment. That's substantial, even for a capital-intensive company such as TSMC.

However, the investment would cover much more than semiconductor manufacturing services. Land, buildings, infrastructure, cleanrooms, production equipment and supporting systems would all consume capital.

TSMC would only capture revenue from the activities actually assigned to it under a future agreement.

The company already manufactures chips for major customers including Apple and Nvidia, and its latest results show demand remains particularly strong for advanced processes.

TSMC's quarterly revenue rose 16.2% sequentially to $46.72 billion, while the company raised its annual growth expectations to approximately 44% in US dollar terms.

That gives TSMC a strong negotiating position. Musk would have to offer sufficient demand, suitable economics and a workable operating structure to justify committing additional manufacturing resources.

Without agreed wafer volumes, chip prices or manufacturing capacity, putting a dollar value on a potential Terafab contract would be guesswork.

Is Musk's 1-terawatt chip target realistic?

Musk's goal of producing enough chips for more than 1 terawatt of computing capacity annually is exceptionally ambitious, and no public production plan yet shows how Terafab will reach that level.

The target is intended to cover computing requirements across Tesla, SpaceX and related AI activities.

But translating 1 terawatt into semiconductor production is difficult because watts measure power, while chip factories produce wafers containing a variable number of usable processors.

Consider the scale. One terawatt equals one trillion watts. If each hypothetical computing system consumed 1,000 watts, that would represent one billion systems. At 10,000 watts per system, it would represent 100 million.

These examples illustrate the size of the ambition rather than Terafab's expected output.

The actual number of wafers needed would depend on processor design, chip size, power efficiency, manufacturing yields, memory requirements and packaging.

Musk has repeatedly argued that his companies' future computing needs could exceed what existing suppliers are prepared to provide. Yet no independently verified production forecast has established the demand implied by Terafab's target.

The strategic objective is clear, but the proposed scale deserves considerable skepticism. Neither Intel nor TSMC has publicly committed enough manufacturing capacity to make the terawatt goal an established production plan.

If you want more recent data on this point, please see our latest semiconductor industry report.

Has Tesla already chosen TSMC over Intel for its AI chips?

Tesla already uses TSMC and Samsung for important AI-chip programs, but Intel's Terafab partnership concerns a different manufacturing strategy.

In July 2025, Musk said TSMC would manufacture Tesla's AI5 chips, initially in Taiwan and later in Arizona.

He also confirmed that Samsung would produce Tesla's AI6 chips at its Texas facility under a semiconductor supply agreement valued at approximately $16.5 billion.

Samsung was already manufacturing Tesla's AI4 hardware.

That gives us three chip generations involving established manufacturing partners. Tesla has repeatedly favored access to different semiconductor suppliers rather than relying exclusively on one manufacturer.

Intel's opportunity comes from Musk's plan to build production facilities under his companies' own control, using technology supplied by an outside partner.

Samsung deserves attention here, too. Its multibillion-dollar agreement gives it a confirmed commercial relationship with Tesla beyond the exploratory Terafab discussions currently involving TSMC.

We haven't seen evidence that Tesla plans to cancel its TSMC or Samsung production arrangements because of Intel. Nor has Samsung publicly displaced Intel from Terafab's technology development.

Musk is pursuing several ways to secure advanced chips simultaneously, which makes the idea of an exclusive winner increasingly difficult to defend.

Could Intel lose Terafab's chip orders without losing its partnership?

Intel could keep its Terafab technology partnership while TSMC takes some of the future chip production, and that would still represent a meaningful commercial loss if the volumes were substantial.

Intel's announced role includes semiconductor design, manufacturing technology and advanced packaging expertise.

TSMC is discussing a possible manufacturing arrangement that would give it operating control over its own production facilities.

If TSMC eventually builds or operates part of the complex, that capacity would likely use TSMC's own manufacturing technologies. Intel would not automatically earn revenue from those wafers.

This exposes a weakness in Intel's current position. Being selected as a technology partner doesn't guarantee that every chip produced across the broader Terafab development will use Intel's manufacturing process.

Tesla's existing use of Samsung and TSMC also shows that Musk's companies are comfortable dividing chip production among several manufacturers.

We still don't know which Terafab products will use Intel 14A, how many wafers Intel-related production might represent, or whether Intel will receive ongoing manufacturing fees.

For now, Intel could lose future business, but there is no evidence that TSMC has already taken away committed production revenue.

Does Intel need Terafab more than Musk needs Intel?

Intel has more to gain from Terafab as proof that its foundry technology can attract major external customers, while Musk already has relationships with multiple established chip manufacturers.

Intel's second-quarter 2026 filing highlights the issue.

External foundry revenue reached $293 million, up from approximately $22 million a year earlier. At first glance, that's an increase of more than thirteen times.

However, Intel explained that much of the increase came from Altera becoming an external customer after its deconsolidation in 2025.

That makes the growth figure less impressive as evidence of new customers choosing Intel instead of TSMC.

Intel Foundry's operating loss improved by approximately $1.08 billion year over year, but the business was still losing more than $2 billion per quarter.

A major external customer successfully manufacturing chips on Intel 14A would strengthen Intel's credibility considerably. Musk's companies could provide that opportunity.

Musk, meanwhile, has access to TSMC for AI5, Samsung for AI6 and potentially other suppliers for future products.

Intel has the technology relationship today. Musk has more alternatives if the manufacturing plans change.

If you want more recent data on this point, please see our latest semiconductor industry report.

Can Terafab really start producing chips by 2028?

Terafab is moving beyond its initial announcement, but Musk's target of commercial production in 2028 leaves very little room for delays.

The project has several concrete milestones behind it. Intel joined in April 2026, research-fab construction began at Giga Texas, and SpaceX subsequently announced the Grimes County commercial development.

Texas officials also confirmed a $30 million Texas Enterprise Fund grant tied to the larger project.

The latest production timetable came during SpaceX's October 20 earnings discussion.

Musk said the Austin research facility was targeting a first functional test chip in May or June 2027. He suggested the initial chip could be something relatively simple, such as SRAM, to verify that the manufacturing line works.

Commercial volume production at Gibbons Creek is targeted for the third quarter of 2028.

Musk acknowledged that the schedule was extremely aggressive and uncertain.

That caution is justified. Producing a basic memory test chip demonstrates only part of the manufacturing process. Advanced AI processors require more complex designs, reliable yields, packaging integration and extensive testing.

Even the research milestone would arrive roughly a year before the proposed commercial production date.

If the schedule slips, Intel's 14A roadmap and TSMC's expanding American production capacity could look quite different by the time Terafab actually needs high-volume manufacturing.

Would TSMC make Musk's chip factory less risky?

TSMC could make Terafab's manufacturing plans more reliable by providing proven advanced-chip production, particularly while Intel 14A and Musk's own facilities are still being developed.

TSMC already manufactures advanced processors for major outside customers, with its 2-nanometer process generating commercial revenue.

Its financial results also show that advanced technologies account for the overwhelming majority of wafer revenue, giving the company extensive experience operating sophisticated production lines at scale.

Intel is making genuine progress. Its regulatory filings confirm ongoing 18A production development, the introduction of 18A-P risk production in June 2026 and continuing work on 14A.

However, Intel's next-generation process has yet to demonstrate comparable outside-customer production volumes.

A TSMC-operated facility could therefore give Musk another source of chips if the Intel-based manufacturing plan runs into problems.

The trade-off is control. TSMC would insist on operating its own facility, limiting Musk's ability to run every part of the project directly.

There would also be practical integration challenges. Two companies operating advanced semiconductor processes in neighboring facilities cannot simply exchange production equipment, manufacturing recipes or partially completed wafers without extensive planning.

TSMC could reduce dependence on Intel, but bringing it into Terafab would add another major manufacturing relationship to manage.

Could TSMC and Intel both win Terafab?

TSMC and Intel could both benefit from Terafab because Musk's chipmaking plans cover different facilities, products and stages of production. A shared outcome is commercially plausible, even though no joint arrangement has been signed.

Intel is positioned to contribute its 14A process and manufacturing expertise to Musk-operated production.

TSMC could supply additional advanced-chip capacity through a separate facility operated under its own management.

A third possibility is that TSMC expands independently in Texas and supplies Tesla or SpaceX without becoming part of Terafab's ownership or operating structure.

All three would fit parts of the public discussion.

What remains unclear is the division of future production.

If Intel 14A becomes the main process used across Musk-operated facilities while TSMC contributes supplementary capacity, Intel would retain a valuable position.

If TSMC eventually produces most of the high-value processors and Intel contributes primarily technical support, the commercial outcome would look very different.

Both companies could participate in the project, but we would still need wafer volumes and contract values to know which one captured the larger opportunity.

What would prove TSMC has beaten Intel for Terafab?

TSMC would have beaten Intel for Terafab if it secured advanced-chip manufacturing work originally expected to use Intel technology and converted that into meaningful production volume.

A signed TSMC manufacturing agreement would be an important first step, but the terms would matter.

For example, TSMC receiving control over the main Terafab fabrication operation would be a major change from Musk's announced strategy.

Similarly, a decision to replace Intel 14A with a TSMC process for the project's central AI-chip designs would demonstrate that Intel had lost its intended technological position.

By comparison, an independently operated TSMC factory on nearby land would mainly show that Musk had secured another semiconductor supplier.

Intel could strengthen its position by demonstrating working Terafab chips on 14A and receiving substantial long-term production commitments.

The most revealing evidence will eventually be actual manufacturing allocations rather than partnership announcements.

Development	What it would tell us
TSMC signs a Terafab manufacturing contract	Discussions become a commercial agreement
TSMC takes over the main factory	Major win over Musk's self-operated model
Intel 14A is replaced by a TSMC process	Direct technological displacement
TSMC receives orders originally intended for Intel	Intel loses potential manufacturing business
Intel produces working Terafab test chips	Intel's technology gains practical validation
Intel receives guaranteed wafer orders	Partnership gains measurable commercial value
TSMC builds an independent Texas factory	Expansion without necessarily displacing Intel
TSMC subleases and operates separate capacity	Both companies can retain different roles
Terafab reaches qualified high-volume production	Real output can finally be compared

If you want more recent data on this point, please see our latest semiconductor industry report.

Is TSMC beating Intel for Musk's chip plant?

No. Intel is still ahead in the announced partnership for Musk's Terafab project, although TSMC has a clear advantage in proven advanced-chip manufacturing.

Intel secured its role in April 2026, and Musk identified Intel 14A as the intended manufacturing technology. TSMC entered discussions later, with CEO C.C. Wei confirming interest in Texas and insisting on operational control over any TSMC-run facility.

As of now, those talks have produced no publicly announced agreement.

TSMC's commercial strengths are hard to dispute. It already manufactures advanced chips at enormous scale, works with Tesla on AI5 and has the financial resources to consider further American expansion.

Intel's position is more fragile commercially. Its foundry business remains heavily dependent on internal customers, and 14A still needs to demonstrate that it can compete at high volume.

But those differences don't overturn the facts of the Terafab partnership. Intel remains the announced technology partner, Musk still plans to operate the main complex himself, and TSMC has yet to secure a defined production role.

Intel is winning the early Terafab partnership race, while TSMC is better positioned in terms of demonstrated manufacturing capabilities. The biggest unresolved question is how much future production Intel will actually capture once the factories begin operating.

TSMC has made the contest more interesting, and Intel has good reason to take the threat seriously. For now, though, calling TSMC the winner would be premature.