Tesla's new D3 chip for space operations and AI challenges

Tesla unveiled its new specialized D3 (Dojo 3) chip at the TERAFAB exhibition in Austin, designed for space operations and tackling major computational challenges in artificial intelligence and orbital computing. During a presentation on Tesla's chip strategy, Elon Musk emphasized that this processor will be a crucial component in the company's future plans, alongside projects with xAI and SpaceX.

Unlike previous Dojo series chips—D1 and D2, which are geared toward ground-based supercomputers for training autonomous systems and data processing—the D3 is built to operate in the vacuum of space and isn't constrained by Earth's power grid or cooling limitations. The idea is that placing computing power in orbit will overcome barriers to scaling traditional data centers and support AI development at the terawatt level and beyond.

The D3 chip is optimized for the extreme conditions of orbit: it can handle higher power and temperatures, and it offers enhanced radiation resistance, which is essential without Earth's protective magnetic field. These features make it suitable for use in orbital server systems that run on constant solar radiation, reducing the need for bulky batteries and complex cooling setups.

Musk also stated that in the coming years, the cost of launching computing chips into space could be lower than building and operating conventional ground-based data centers. The plan is to integrate D3 processors into large orbital server racks with capacities up to 100 kilowatts, launched via the Starship rocket, to provide round-the-clock AI services in space.

Unlike other projects, such as the AI5 and AI6 chips focused on autonomous vehicles and robotics, the D3 will serve as the hidden backbone for a massive computational infrastructure. This setup could handle AI data processing, support interplanetary networks, and even help build a "space internet" infrastructure.