Tesla's Dojo3 Chip Revived for Space AI

Tesla is reviving a previously shelved project with a cosmic twist. The company aims to restart work on Dojo3, its third-generation AI training chip, but with a mission that has fundamentally changed.

Originally conceived for training self-driving models on Earth, the chip's new purpose is now aimed at the stars. According to the company's leadership, Dojo3 will be dedicated to space-based AI compute, marking a significant pivot in Tesla's hardware strategy.

This development suggests a broader ambition for Tesla's custom silicon, extending beyond terrestrial vehicles and into the realm of space exploration and satellite-based artificial intelligence.

The Dojo3 project was initially part of Tesla's ambitious plan to build the world's most powerful AI training supercomputer. Its primary goal was to accelerate the development of Full Self-Driving (FSD) technology by processing vast amounts of real-world driving data.

However, the project was previously abandoned, leaving the future of Tesla's custom AI chips in question. The recent announcement to restart development comes with a complete reimagining of the chip's application.

The shift in focus is dramatic:

• From terrestrial automotive AI training
• To space-based computational workloads
• Supporting future space missions and satellite networks
• Enabling AI processing outside Earth's atmosphere

This pivot aligns with Elon Musk's long-term vision for humanity as a multi-planetary species, where advanced AI will be essential for operating in space environments.

The decision to repurpose Dojo3 for space-based compute highlights Tesla's evolving identity. While the company remains an automotive leader, its technological ambitions now clearly extend into aerospace and deep space exploration.

This move could provide Tesla with a unique competitive advantage. By developing AI chips capable of operating in the harsh conditions of space, the company could secure a dominant position in the emerging market for space-based AI services.

Dojo3 will be for 'space-based AI compute'.

The statement from leadership is concise but carries significant weight. It suggests that Tesla's custom silicon division is no longer solely focused on solving Earth-bound problems like autonomous driving, but is instead looking to the future of computing beyond our planet.

While technical specifications for the revived Dojo3 remain under wraps, the shift to space-based compute likely requires significant architectural changes. Space environments present unique challenges, including extreme radiation, temperature fluctuations, and the need for ultra-reliable, low-power operation.

Tesla's expertise in designing high-performance, energy-efficient AI chips for its vehicles could be adapted to meet these stringent requirements. The company's experience with its Dojo supercomputer architecture, which uses custom training tiles, provides a strong foundation for building specialized processors for space applications.

This development may also signal Tesla's intent to compete with other tech giants exploring space-based AI, potentially creating a new ecosystem for satellite and orbital computing services.

The revival of Dojo3 with a space-focused mission represents a bold new chapter for Tesla's AI ambitions. It demonstrates the company's willingness to pivot its most advanced technologies toward emerging opportunities.

As Tesla continues to develop its space-based AI capabilities, the industry will be watching closely. The success of this project could redefine what's possible for AI processing in space and accelerate humanity's expansion into the cosmos.

For now, the focus remains on bringing Dojo3 back to life, with its new purpose firmly set on the stars.