TECH NEWS – Elon Musk’s company is reportedly moving into GPU design, but that does not mean desktop graphics cards with a SpaceX logo are on the way.
According to recent reports, SpaceX, the private aerospace and artificial intelligence company founded by Elon Musk, is planning to manufacture its own graphics processors in the near future. The company is expected to go public this summer, with the IPO reportedly valued at around $1.75 trillion. As part of that process, it must file an S-1 registration with the U.S. Securities and Exchange Commission, outlining its finances and risks ahead of the public offering. Reuters reviewed part of that filing and noted that SpaceX lists the manufacturing of its own GPUs among its major capital expenditures. Those chips are expected to be tailored for AI-related tasks, closer in spirit to something like Google’s Tensor Processing Unit approach than to the idea of a consumer graphics card.
It is still unclear how much money SpaceX is prepared to invest in this hardware effort, but the move is not especially shocking when viewed alongside Musk’s recent work with Intel. Under that broader collaboration, Intel is expected to design, manufacture, and package ultra-high-performance chips at scale in order to accelerate the Terafab initiative, which aims to produce 1 TW of compute power per year. The Terafab project is an advanced AI chip manufacturing complex planned for Austin, Texas. At the moment, the massive effort is focused on chip fabrication, packaging, and testing. It is also a major collaboration involving SpaceX’s xAI unit and Tesla, although it remains unclear exactly what kinds of chips the facility will end up producing.
Musk also said during an earnings call that Terafab will rely on Intel’s 14A manufacturing process, which is among the company’s most advanced technologies, but not yet fully mature. By the time Terafab is supposed to ramp up, though, 14A is expected to be far more ready for prime time, so the logic behind that decision is easy enough to understand. As for SpaceX’s own GPU ambitions, nobody currently knows whether Intel will be the company that manufactures them or whether some other partner will have to step in. It is also an awkward time to announce a hardware push like this, given that Nvidia’s main manufacturing partner, TSMC, is already operating under heavy demand, and plenty of other production lines are similarly packed.
That probably explains why SpaceX also acknowledged in its S-1 filing that it does not have long-term agreements with many of its direct chip suppliers. The company expects to continue sourcing a significant share of its compute hardware from third parties, and there is no guarantee that it will achieve its Terafab goals within the desired timeframe, or at all.
That is unlikely to sound ideal to investors, but time will tell whether SpaceX’s GPU ambitions are enough to convince the market that the gamble is worth it.
