Tesla Becomes Intel Foundry's First Major Customer for 14A as Musk Commits Terafab to Unproven Process
Elon Musk picked Intel's 14A process for Tesla's Terafab AI chip complex, handing the struggling Intel Foundry its first marquee external win for a node that is still years from volume production.
Overview
Tesla has committed its planned Terafab semiconductor complex in Austin to Intel’s forthcoming 14A process node, making the electric-vehicle and robotics maker the first publicly identified marquee customer for Intel Foundry’s most aggressive technology bet. CEO Elon Musk disclosed the decision on Tesla’s April 22 earnings call, and Intel CEO Lip-Bu Tan confirmed the relationship a day later when Intel reported first-quarter results, saying he could “think of no better partner than Elon Musk” for the work ahead.
The partnership matters less for what it ships in 2026 than for what it signals: Intel Foundry, which has spent the past two years searching for a high-profile external anchor tenant, finally has one — even if 14A silicon is still years away from volume production.
What We Know
Musk announced Tesla’s choice during the Q1 2026 earnings call on April 22, framing it as a forward-looking bet rather than an immediate manufacturing commitment. “Given that by the time Terafab scales up, 14A will be probably fairly mature or ready for prime time, 14A seems like the right move, and we have a great relationship with Intel,” he said, as reported by PC Gamer.
Intel’s CEO returned the endorsement on the company’s own April 23 earnings call. “Elon and I share a strong conviction that global semiconductor supply is not keeping pace with a rapid acceleration in demand,” Tan said, according to Tom’s Hardware. He described the work as “a broad relationship in which the two companies will learn a lot together” but declined to detail specific volumes or financial terms, according to Fortune.
Intel reported Q1 2026 revenue of $13.6 billion — 7 percent year-on-year growth and well above analyst expectations of a 2 percent decline — and guided Q2 to between $13.8 billion and $14.8 billion, Fortune reported. Shares rose more than 22 percent in after-hours trading on the print and the customer announcement combined.
On the same call, Tan addressed 14A’s manufacturing status directly: “We’re making great progress in terms of yield and cycle time,” he said, per Fortune, while indicating Intel was engaging with multiple potential 14A customers but would announce them only with the customer’s consent.
The 14A process is the successor to Intel’s 18A node and will be the company’s first to use High-NA EUV lithography in volume. Intel’s previous-generation 18A process entered high-volume manufacturing in late 2025 and now powers the Panther Lake Core Ultra 300 series, but yields remain below the level required for healthy margins. According to Tom’s Hardware, 18A is making “steady progress” but is not expected to reach industry-standard yield levels until 2027 — a backdrop that makes Tesla’s bet on the next node, 14A, materially more speculative.
Terafab’s Scale and Tesla’s Pivot
Terafab is the chip-manufacturing complex Tesla and SpaceX unveiled earlier this year, as previously reported by The Machine Herald. The Austin project is designed to combine logic, memory, and advanced packaging on a single site, with a stated ambition of producing up to one terawatt of annual AI compute. Reuters, citing Bernstein, reported that “building enough chip capacity to power one terawatt of annual compute would cost between $5 trillion and $13 trillion in capital expenditure,” in its Factbox on the Terafab plan republished via Yahoo Finance — a figure that underscores how speculative the full vision remains.
Tesla currently sources its AI silicon from TSMC and Samsung. The company recently taped out its AI5 processor using TSMC’s process technology — a fact Musk emphasized when he confirmed during the April earnings call that 14A is the destination for future Terafab production rather than current chips. The decision to route forthcoming generations through Intel rather than build proprietary process technology from scratch is the most concrete narrowing of Terafab’s manufacturing question to date.
Musk has said Terafab “will technically be two fabs, each making only one chip design,” Fox Business reported when the project was first unveiled in March: one fab tuned for chips that go into Tesla vehicles and Optimus humanoid robots, the other for chips destined for SpaceX’s space-based AI data centers. Reuters confirmed in April that the two-fab plan persists under the new Intel partnership: “SpaceX and Tesla would build two advanced chip factories at this sprawling facility, one to power cars and humanoid robots, and another designed for the space data centers.” Both, under the new arrangement, would lean on Intel’s 14A node.
What Intel Gets — and What Tan Hinted At
For Intel Foundry, the value of the announcement is mostly strategic. Tan, who took over as CEO in March 2025, has publicly tied Intel Foundry’s survival to the company’s ability to win at least one major external customer for 14A; the company spent much of 2025 conducting customer evaluations on its 18A process kit without converting them into binding manufacturing commitments. Tesla becomes the first publicly named flagship customer.
The broader context for Intel’s foundry pivot is well established: Intel reclaimed full ownership of its Ireland Fab 34 from an Apollo-led consortium for $14.2 billion on April 1, the company announced, unwinding a 2024 minority-stake arrangement and signaling renewed confidence in its manufacturing roadmap. The Tesla deal, three weeks later, fits into that narrative shift from defensive financial engineering toward offensive customer wins.
Tan also flagged the partnership’s research dimension. “We are excited to explore innovative ways to refactor silicon process technology, looking for unconventional ways to improve manufacturing efficiency that will eventually lead to a dynamic improvement in the economics of semiconductor manufacturing,” he told analysts, as Tom’s Hardware reported. The phrasing leaves room for joint work on equipment, process flow, or fab automation rather than a straightforward foundry contract — consistent with Musk’s stated ambition to dramatically lower per-wafer costs at Terafab.
What We Don’t Know
Neither company has disclosed wafer volumes, pricing, or the specific Tesla products that will run on 14A. Tan said only that customer announcements would come at customers’ discretion, and Musk did not commit to a date for first 14A silicon at Terafab.
The production schedule for 14A itself remains a moving target. Intel has not committed publicly to firm risk-production or volume-ramp dates for 14A on its recent earnings calls; Tan said only that yield and cycle-time progress was being made. Process design kit releases continue, but final yields, defect density, and per-wafer cost at 14A are unknown.
It is also unclear how exclusive the relationship is. Tan’s reference to broader 14A engagements suggests Intel is courting multiple customers in parallel, and Tesla has not committed to making Intel its only foundry partner. Tesla’s existing TSMC and Samsung relationships continue, and the AI5 processor will tape out and ship from those vendors regardless of how the Intel partnership progresses.
Analysis
The announcement is a study in mutual dependency. Intel needs a flagship 14A customer to validate its foundry pivot to Wall Street and to justify the capital outlays already committed for High-NA EUV equipment. Tesla needs a domestic manufacturing partner that can credibly support the volumes Terafab envisions, and TSMC’s Arizona ramp is already heavily booked by Apple, Nvidia, and AMD.
Neither party is locked in. Musk’s framing — that 14A will “probably” be ready when Terafab scales — is conditional, and Tan’s refusal to discuss volumes preserves Intel’s flexibility to court other anchor tenants. What both companies have produced is an announcement that benefits each at low immediate cost: Intel gets a marquee logo for its foundry roadshow, and Tesla gets a domestic-manufacturing narrative that supports its Terafab capital plans without committing dollars to 2027-vintage silicon.
Whether the partnership delivers actual wafers will depend on milestones that neither company has yet hit. Intel must demonstrate that 14A reaches yield maturity faster than 18A did, and Tesla must show that Terafab can break ground, hire, and progress through the multi-year fab construction cycle. Until then, the most concrete artifact of the deal remains the words on two consecutive earnings calls — and the 22 percent overnight move in Intel’s share price.