Elon Musk, president of electric car maker Tesla, said on Wednesday (23) that the company plans to use Intel’s 14A manufacturing process to produce chips for the Terafab project, an advanced artificial intelligence semiconductor complex that will be installed in the state of Texas, in the United States.
What is the Terafab project?
Rocket company SpaceX, also owned by Musk, the AI unit xAI and Tesla itself will build two advanced chip factories in a large complex in Austin: one aimed at use in Tesla vehicles and Optimus humanoid robots, and another dedicated to data centers in Earth orbit.
“Either we build Terafab or we don’t have the chips,” Musk said at a presentation in Austin in March, adding that current global production of microprocessors meets only a fraction of his companies’ future needs.
Continues after advertising
Musk said he was grateful to current suppliers — citing Samsung, TSMC and Micron — but said he believed that the combined demand from his companies would end up exceeding the total production capacity of these manufacturers. The billionaire, however, did not give a timeline for the project and has a history of highly ambitious announcements, many of them with delays.
Intel announced this month that it will join the project, bringing its consolidated experience in chip manufacturing to the venture.
Location and scale
Musk said Terafab will handle all stages of chip production, including design.
On Tesla’s earnings call on Wednesday, he stated that the details of the Terafab rollout are still being worked out. In the short term, Tesla will build a research factory at its Giga Texas campus in the Austin region. The initiative is expected to cost around US$3 billion and “be capable of producing perhaps a few thousand wafers of chips per month, but the intention is to really test ideas,” Musk said.
“What we’ve discovered so far is that Tesla will do the research factory, SpaceX will do the initial part of the large-scale Terafab. And then we have to figure out the rest,” he said.
According to Musk, Terafab is expected to produce the equivalent of 1 terawatt of computing capacity per year, compared to about half a terawatt currently produced in the United States. Building enough capacity for 1 terawatt annually will cost between $5 trillion and $13 trillion, according to Bernstein estimates.
Continues after advertising
Technology
Tesla plans to use Intel’s 14A manufacturing process to produce chips in the Terafab project. The contract represents Intel’s first major customer for this technology, a breakthrough for the North American manufacturer, which has been struggling to consolidate its foundry business (contract manufacturing) and compete with the Taiwanese TSMC.
Musk said he believes that by the time Terafab is scaling up, Intel’s 14A process will “probably be quite mature or ready for prime time” and “seems like the right decision.”
According to BloombergMusk’s team contacted several suppliers in the semiconductor chain, including Applied Materials, Tokyo Electron, Lam Research and Samsung, for the Terafab project. The group sought price quotes and delivery times for a wide range of manufacturing equipment, and in recent weeks contacted manufacturers of photomasks, substrates, deposition, cleaning, testing equipment and other tools.
Continues after advertising
A Reuters reported that SpaceX plans to manufacture its own graphics chips (GPUs), core components for training AI models.
The “unknown”
Although Musk has said that Terafab will target chips for cars, humanoid robots and space data centers, several points remain open, such as:
- Who will finance the very expensive chip manufacturing equipment?
- Who will be responsible for operating the factories?
- How soon will the complex actually start operating?
