Terafab seeks equipment for million-wafer chip ambition | Magica
Terafab seeks equipment for a million-wafer ambition led by memory
Editorial Team
••📖7 min read
Terafab is discussing equipment supplies for a Texas chip complex targeting roughly 700,000 memory wafers and 300,000 logic wafers monthly. Supplier agreements show progress, but production technology, yields and demand remain decisive uncertainties.
Terafab has reportedly outlined a million-wafer monthly target to suppliers, with about 70% earmarked for memory.
Equipment discussions in Japan and agreements with Korean suppliers give the project a more concrete procurement trail.
The target remains separate from demonstrated manufacturing capability, with process choices, yields and future demand still unresolved.
Tesla and SpaceX's Terafab semiconductor project has reportedly told equipment suppliers it wants capacity for about 1 million silicon wafers a month in the 2030s, as its procurement team seeks equipment in Japan and builds a supply chain in South Korea. The proposed split—roughly 700,000 memory wafers and 300,000 logic wafers monthly—would make memory manufacturing the larger part of an effort extending beyond Tesla's existing role as a chip designer.
The schedule needs qualification. A November 8 account describes the goal as capacity by 2032; another account places it after 2032. Both trace the supplier target to Nikkei, rather than independently confirming it. These are accounts of an informal production ambition, rather than separate confirmations of a completed investment or an operating fab.
The strategic change is the proposed move from designing chips to manufacturing them. Elon Musk, chief executive of electric-car maker Tesla, is pursuing Terafab with SpaceX, the space company he founded and leads. Tesla has already designed autonomous-driving chips while relying on Samsung Electronics and Taiwan Semiconductor Manufacturing Co., or TSMC, for production, according to the reporting on its manufacturing plans. In the Texas government’s project announcement, Musk described SpaceX’s existing Texas operations developing and launching Starship and manufacturing satellite-internet equipment. Neither company's existing manufacturing record establishes an ability to mass-produce semiconductors.
SpaceX’s rendering of the proposed Terafab complex in Grimes County, Texas. Source: Wccftech.
Anthropic raised $81.3 billion in a record IPO and opened 34% above its offer price before retreating. Rapid revenue growth and adjusted operating profits sit alongside vast computing commitments and limited investor control.
Terafab was unveiled in March. It is intended to supply Tesla's autonomous-driving systems and Optimus humanoid robots, alongside SpaceX satellites and proposed space-based AI data centers. Musk argues that existing suppliers' expansion will not meet the companies' future needs; that is his rationale for internal production, rather than an established demand forecast. The plan brings design, manufacturing, packaging and testing together to accelerate development.
Its development is staged: a research line at Tesla's Austin plant precedes a larger production complex in Grimes County, Texas. Musk's stated milestones are a working test chip in May or June 2027 and mass production in the second half of 2028, according to the account of the Japanese procurement visit. That schedule does not establish when the full million-wafer capacity would be available.
That team reportedly met Japanese equipment makers early in November, outlining orders and discussing deliveries. One supplier was asked to deliver within a few months. The companies involved, equipment types and order quantities were not disclosed.
More specific evidence comes from supplier reporting from October. Tesla and SpaceX representatives approached about 10 Korean companies at SEMICON West in San Francisco on October 13–15, although some discussions had begun months earlier.
HPSP had agreed to supply one high-pressure hydrogen annealing system for the R&D line. Cleaning specialist KoMiCo had signed a three-party nondisclosure agreement with Tesla and SpaceX for discussions about the Texas pilot line. Testing-component supplier ISC had progressed to exchanging samples and R&D design information. Those steps indicate different levels of commitment; an NDA or sample exchange is not a volume-production order.
Hanmi Semiconductor was discussing thermocompression bonders after an equipment presentation requested by SpaceX. Materials supplier Fujifilm was holding regular procurement meetings, but its official said the products supplied would depend on which technologies were selected. A broad supplier search can precede a settled production process.
The supplier choices also have a history. Many already serve Tesla or SpaceX, while former Intel staff and procurement personnel from GlobalFoundries have brought existing industry relationships into Terafab. Korean suppliers described pressure for low prices, including one industry official's claim that the project sought prices below Chinese suppliers' offers. An equipment-maker official cited the exclusion of Chinese suppliers and the higher cost of Japanese alternatives as reasons Korea was attractive.
The reported long-term target allocates roughly 700,000 wafers monthly to memory and 300,000 to logic. These are capacity ambitions, not operating output. Source: Wccftech.
Memory capacity is not yet a memory business
The reported allocation puts about seven-tenths of Terafab's eventual wafer capacity into memory, calculated from 700,000 out of 1 million wafers monthly. That establishes the proposed emphasis, but does not reveal the number of saleable chips or how much computing capacity the plant could deliver.
Tesla is recruiting designers for high-bandwidth memory, or HBM, and low-power DRAM, according to the reporting on its plans. The memory target does not specify how capacity would be divided between those products. Nor has the project disclosed whether it would manufacture and stack DRAM itself or package chips obtained elsewhere. Producing HBM would require reliable yields across DRAM fabrication and stacking and packaging, as well as design.
According to the account of the Japan visit, chip manufacturer Intel said in April that it would collaborate on Terafab's chip design, manufacturing and packaging. One November account says Terafab intends to use Intel's 14A process for logic chips; the other reporting describes the technology choice more cautiously. Discussions with TSMC have also been reported, without a settled form of participation. These reported roles do not establish a final ownership or operating arrangement.
There is already a substantial external manufacturing alternative. Samsung has a $16.5 billion chip-supply contract with Tesla, and Musk has said its Taylor, Texas, factory will make Tesla's next-generation AI6 processor, according to the account of Tesla's existing supply arrangements. The report says Samsung plans to begin producing Tesla chips using a 2-nanometer process in the second half of 2027. Successful internal production could give Tesla more choices for later chips and more room to negotiate volumes and prices. Delays or poor yields would extend its reliance on established foundries.
Samsung and SK Hynix, established AI-memory suppliers, also face a conditional change in future orders. The effect depends on Terafab's manufacturing scope and speed, and whether Tesla's vehicles and robots and SpaceX's proposed data centers generate the expected demand. A target for all memory wafers does not establish finished HBM output.
Texas announced first-phase investment of more than $16.8 billion and 3,000 projected new jobs. These figures do not establish completed spending or the full expansion cost. Source: Office of the Texas Governor.
The next proof is working silicon and a viable expansion
The announced first phase already carries a substantial investment target. In an August 6 announcement, Texas put the first phase at more than $16.8 billion in capital investment and projected 3,000 new jobs. It said a $30 million Texas Enterprise Fund grant had been extended to SpaceX and that the facility qualified under the state's Jobs, Energy, Technology, and Innovation program. Those figures describe the announced first phase and public support, not the full cost of achieving a million wafers a month.
The state describes the enterprise fund as performance-based. Its announcement does not establish that the grant has been paid or that the full investment has been spent. It also quotes Iola school-district superintendent Jeff Dyer saying the district agreed to tax incentives with TeraFab AI, LLC. A project entity is therefore named in the public record, although the October supplier report says the ownership structure has not been disclosed.
Scale must also be separated by phase. The October supplier report cites industry expectations for an initial production facility capable of 100,000 wafers monthly before 2030—one-tenth of the eventual million-wafer ambition. It does not provide a full expansion budget.
Equipment arrivals will be an intermediate test. Advanced fabs typically take at least three years from planning to mass production, according to the account of the Japan visit, and installed tools still need process validation and defect reduction. Musk himself described his schedule as very aggressive. Tesla and SpaceX enter this effort without a record of mass-producing semiconductors, even as they recruit experienced suppliers and personnel.
Demand is a separate constraint. Hisaya Suzuki of market researcher Omdia questioned whether the completed plant could fill its capacity when SpaceX's future semiconductor needs remain unclear. A working research chip would establish progress toward a process; sustained yields, a defined memory-production plan, financing for expansion and credible demand would be needed to judge whether the much larger capacity target can become an economic manufacturing operation.
The Kimi developer reportedly abandoned planned Nvidia B300 capacity for 2027 after new US restrictions. The deal exposes financing risks for overseas AI clouds, but neither a model delay nor a $150 million cash loss is established.
Kong is expanding from API management into hosting AI agents and applications. Volcano bundles backend services with a LavaFlow repair preview, but human review, plan restrictions and deployment controls narrow its autonomous-software pitch.
Anthropic's pre-debut price indications reached $105, about 26.5% above its offer price, after an $81.3 billion IPO. Strong demand for the Claude maker confronts enormous computing commitments and a governance structure that limits ordinary shareholders' influence.
Google has booked New Glenn for a dedicated Project Suncatcher launch in early 2029. Laboratory tests support parts of its orbital AI design, but deployment, launch reliability and commercial economics remain unproven.
Canva is rolling out free controls for Free and Pro accounts to disable supported generative AI tools and reduce AI-made assets in search. The settings extend individual choice while leaving AI enabled by default and preserving the company's investment in generation.
Alibaba targets 30%–45% cloud revenue growth next fiscal year and may reduce its long-term capacity plan. Domestic chips underpin its unchanged 2030 revenue ambition, but production, funding and customer allocation remain critical tests.
Singapore’s central bank plans a common protocol for financial institutions to assess customer AI agents, alongside a sandbox for treasury and payments. Its digital settlement network has completed a S$100 million bill issuance, but wider adoption, cost savings and agent accountability remain unresolved.
Jeff Bezos says Prometheus tools are in private trials, including two Blue Origin projects. Its claims of faster design now face the harder test of manufacturing gains, as rivals automate existing engineering workflows.
The IEA projects data-center electricity demand reaching 1,000 terawatt-hours annually by 2030, but connection queues, equipment shortages and local opposition could limit how much of the investment boom gets built.
European individuals secured about $1.1 billion of Anthropic shares, roughly 1.4% of the public offering’s value. The UK offer combines broad participation with rationing, while the AI lab’s governance and compute commitments define what investors are buying.
Federal prosecutors accuse Igor D’iu of acting for Russia to surveil employees of a Silicon Valley defense technology company identified in reports as Palantir. An FBI affidavit describes screen photographs, employee dossiers and $45,000 in reported payments, but leaves the handlers’ identities and the success of attempted electronic collection unresolved.
Alibaba’s September-quarter profit more than tripled as investment gains and a commerce recovery amplified stronger cloud earnings. Cloud revenue rose 53%, but AI application losses widened and infrastructure spending left free cash flow negative. A large share placement bolstered liquidity.
President Donald Trump rejected international restrictions on advanced AI at APEC. His administration has already outlined voluntary cybersecurity cooperation, alongside deregulation and support for American AI exports and infrastructure.
London-based LATO has raised $4.7 million to expand AI agents that interview customers and experts for investment research. Its early traction claims come as rivals offer similar research workflows and Salesforce absorbs Listen Labs; comparative cost and quality remain unproven.
Snap’s free Easy Lens CLI connects AI coding assistants to Snapchat’s AR creation and publishing workflow. It builds on earlier prompt-based tools as TikTok offers a competing approach.
Nvidia’s first earnings report since Jensen Huang’s departure showed revenue more than doubling, but shares fell 2.9%. Interim CEO Colette Kress faces a permanent leadership search, lower forecast margins and expanding infrastructure commitments, including conditional guarantees.
Needham’s $150 target stands about 81% above Anthropic’s $83 IPO price, versus Morningstar’s $92 fair-value estimate. Rapid enterprise growth and fresh financing support the business, but pricing power, compute commitments and shareholder control complicate the investment case.
A cyberattack disrupted access to European Commission payslips, leave management and medical reimbursements. The HR platform is back online, while the health insurance system remains unavailable and the extent of any staff data theft is unknown.
Waymo suspended Bay Area service during the November 18 storm after reports of robotaxis obstructing flooded streets, raising questions about how it withdraws its fleet when road conditions deteriorate.