Elon Musk has decided that building a semiconductor fabrication plant cannot be particularly difficult and reckons his Tesla and SpaceX operation will be churning out advanced chips by 2028.
The billionaire has set the third or fourth quarter of 2028 as the target for volume production at Terafab, his proposed semiconductor manufacturing complex in Grimes County, Texas.
According to eeNews Europe, Musk told investors during SpaceX’s October earnings call that Dutch lithography outfit ASML and other equipment suppliers were accelerating deliveries to support the project.
The first milestone is a functioning test chip at a research facility in Austin, expected around May or June 2027. This will probably use relatively simple static random-access memory (SRAM) structures to determine whether the manufacturing process works.
Musk eventually wants Terafab to produce what he describes as a terawatt of artificial intelligence computing capacity annually, presumably after the engineers have worked out how to manufacture the first functioning chips. Minor obstacles remain, including building a semiconductor fabrication plant, installing extraordinarily complicated manufacturing equipment, developing reliable processes, and achieving commercially acceptable yields.
Established chipmakers spend years mastering these things, but Musk appears convinced that the industry has been waiting for someone sufficiently clever to explain how it should be done. His optimism would be more reassuring if his previous technological predictions had been more closely tied to the calendar.
In 2016, Musk predicted that Tesla would demonstrate a fully autonomous journey from Los Angeles to New York by the end of 2017. The promised demonstration failed to materialise, although the deadline subsequently enjoyed several opportunities for resurrection. By 2019, he was promising more than a million autonomous Tesla robotaxis operating by 2020. Six years later, that particular prediction remains an impressive illustration of the difference between announcing a technological revolution and delivering one.
Then there is the second-generation Tesla Roadster, unveiled in 2017 with production promised for 2020. Customers were invited to put down substantial deposits, including $50,000 reservations. The car subsequently missed deadline after deadline and, as of November 2026, has yet to reach customers in production form. Musk has spent almost nine years demonstrating that announcing a sports car is considerably easier than manufacturing it.
His ambitions extended to turning ordinary houses into solar power stations. Tesla unveiled its Solar Roof in 2016 and later said it would install 1,000 roofs every week. Industry research found that actual installations were a fraction of that target. Tesla stopped offering the premium solar roofing product through its website in August 2026, nearly a decade after its grand introduction.
The company’s Optimus humanoid robots have followed a familiar trajectory. Musk predicted Tesla would manufacture roughly 10,000 robots in 2025, but missed the target. In January 2026, he acknowledged that no Optimus robots were performing useful work in Tesla’s factories.
Of course, Musk’s greatest talent for ambitious scheduling has traditionally involved getting people to Mars. In 2016, he envisaged uncrewed missions reaching Mars in 2022, followed by the first human missions in 2024. Neither happened. His subsequent ambition to launch an uncrewed Mars mission in 2026 was overtaken by another change of direction, as SpaceX shifted its immediate attention toward establishing a settlement on the Moon. Musk had previously dismissed the Moon as a distraction from Mars, showing that even destinations in the solar system are subject to management revisions.
Building advanced chips involves significantly more than buying expensive lithography machines and telling the engineers to work faster.
ASML’s extreme ultraviolet lithography systems are among the most complicated pieces of manufacturing equipment ever developed. They require extensive supporting infrastructure and represent one component of an enormously demanding production process.
A modern semiconductor fabrication plant needs highly controlled environments, specialist materials, sophisticated process engineering and thousands of precisely coordinated manufacturing steps.
Even companies with decades of semiconductor experience struggle to bring new manufacturing processes into volume production while maintaining acceptable yields. Terafab will need to develop that expertise while constructing and equipping its facilities.
The planned 2027 SRAM test chip would establish only that some elements of the manufacturing process were working. It would say little about the project’s ability to manufacture competitive advanced processors in commercial quantities.
Musk acknowledged during the October conference call that his company was still learning semiconductor fabrication. That admission may be the most technically credible thing he has said about the entire project.
To be fair, SpaceX has achieved substantial engineering successes, particularly with reusable rockets, and Tesla established itself as a major electric vehicle manufacturer. Musk’s companies have demonstrated that ambitious engineering projects can succeed. Unfortunately, their successes have frequently arrived long after their founder’s original predictions.
Terafab presents another opportunity to discover whether Musk’s timetable reflects genuine manufacturing progress or the familiar assumption that complicated technology will eventually surrender to sufficient executive enthusiasm. For now, the project has a proposed production deadline, ambitious capacity targets and assurances that equipment suppliers are cooperating.
What it lacks is a proven semiconductor manufacturing process capable of delivering the advanced chips Musk expects to produce in 2028.







