Elon Musk has unveiled Tesla’s most ambitious AI roadmap to date, expanding beyond self-driving cars and humanoid robots to eventually include space-based applications.
In a recent announcement on X, Musk detailed how Tesla’s upcoming generations of in-house AI chips will be utilized across the company’s growing ecosystem. Musk stated that AI4, the current hardware in Tesla’s EVs, will achieve self-driving safety levels well beyond human capability. He also outlined plans for AI5 to enhance vehicle performance significantly, AI6 for Optimus robots and data centers, and AI7/Dojo3 for space-based AI compute.
Musk’s updates build on previous statements about Tesla’s custom silicon endeavors. He recently mentioned that the next-generation AI5 chip, also known as Hardware 5 (HW5), is nearing completion, with work on AI6 already in progress. AI5 is expected to deliver a substantial 50x improvement over the current AI4 hardware, serving as the foundation for Tesla’s future Full Self-Driving system, Robotaxi fleet, and Optimus humanoid robots.
While AI4 currently powers Tesla’s FSD system, Musk has emphasized that AI5 will elevate vehicle safety to levels surpassing human capability, a critical step towards achieving widespread autonomy. However, production timelines are long-term, with Musk aiming for volume production of AI5 by 2027 and integration into production vehicles by mid-year. AI5 will be manufactured in the U.S. by both TSMC and Samsung.
Looking ahead to AI6 in 2028, Tesla has secured a significant $16.5 billion deal with Samsung for domestic production of the chip, with Musk touting it as potentially the best AI chip available. Unlike AI5, which is primarily geared towards vehicles, AI6 is designed with Optimus robots and large-scale data center compute in mind.
Musk’s vision for AI7, or Dojo 3, is particularly noteworthy. This generation is slated to be space-based AI compute, hinting at orbital infrastructure that could support global autonomy networks and satellite-based processing. Tesla is aiming for a rapid nine-month design cycle for its custom silicon, signaling a swift release schedule for future AI chips.
As Tesla expands its footprint in Robotaxis, humanoid robots, and AI training systems, the company is considering the construction of a massive chip fabrication facility. With applications spanning vehicles, robots, data centers, and space-based computing, Tesla’s demand for AI silicon is only set to grow in the coming years.