Tesla aims to revive its Dojo project, intending to dedicate the in-house AI chip to space-based artificial intelligence computing, according to Elon Musk. This decision marks a shift from the company's previous strategy, announced five months prior, to halt the Dojo supercomputer development.

Musk stated on X that the decision to restart Dojo3 was influenced by the progress of Tesla's AI5 chip design. The AI5 chip, manufactured by Samsung, was originally intended for training self-driving models on Earth. The new direction suggests a potential application of AI in space exploration or satellite technology.

The initial decision to suspend the Dojo project followed the departure of key personnel, including Dojo lead Peter Bannon. Approximately 20 Dojo team members subsequently joined DensityAI, a startup focused on AI infrastructure founded by former Dojo head Ganesh Venkataramanan and ex-Tesla employees Bill Chang and Ben Floering. At the time, reports indicated that Tesla planned to increase its reliance on Nvidia, AMD, and Samsung for compute and chip manufacturing.

The Dojo supercomputer was designed to process vast amounts of data collected from Tesla vehicles to train its autonomous driving system. By developing its own silicon, Tesla aimed to reduce its dependence on external suppliers and optimize performance for its specific AI workloads. The company has not yet detailed the specific applications of its space-based AI compute plans.

The revival of Dojo indicates a renewed commitment to in-house chip development, despite the earlier move towards outsourcing. The long-term implications of this strategic shift for Tesla's AI capabilities and its partnerships with other technology companies remain to be seen.