Article Complete
AIM
Tesla Now Targets 2027 for Mass Production of AI5 Chips
Tesla Now Targets 2027 for Mass Production of AI5 Chips
Chip Dream Deferred

Tesla Now Targets 2027 for Mass Production of AI5 Chips

Initial samples and a small number of production units are expected in 2026, with large-scale rollout set for 2027.

Published Updated

Key takeaways

  • Tesla plans small-scale production of AI5 chips in 2026, with mass production set for 2027.
  • Collaborate with TSMC and Samsung for AI5 chip manufacturing, ensuring consistent AI software performance.
  • Develop the AI6 chip expected to deliver double the performance, targeting mid-2028 for volume production.
  • Differentiate AI7 chip design will require new fabs, with no specific production timeline announced yet.
  • Align production plans with Tesla's long-term strategy to enhance AI and self-driving systems.
5 min

Elon Musk announced that Tesla will start small-scale production of its AI5 chip in 2026, with mass manufacturing planned for 2027. The chips will be made by Taiwan Semiconductor Manufacturing Company (TSMC) and Samsung.

Musk said in a post on X that slightly different versions of the AI5 chip will be produced at each facility. “They translate designs to physical form differently, but the goal is that our AI software works identically,” he explained.

Tesla is also developing the AI6 chip, which will use the same fabs as AI5. Musk said it is expected to deliver “roughly 2X performance” and could reach volume production by mid-2028.

On the contrary, the AI7 chip will require “different fabs” as it involves a new design approach. Musk described it as “more adventurous,” though no specific timeline was provided for production.

The production of AI6 chips was confirmed to take place in Samsung’s upcoming foundry in Taylor, Texas, earlier this year. This will take place post Samsung’s current production of AI4 chips and TSMC’s upcoming AI5 chip.

The announcement aligns with Tesla’s long-term hardware roadmap to strengthen its AI and self-driving systems. The company’s collaboration with both TSMC and Samsung highlights its effort to diversify manufacturing and manage supply chain risks while improving chip performance and scalability.

In a post on September 7, Musk mentioned that AI5 will likely be the best inference chip for models with parameters below ~250 B. He added that it has the lowest cost silicon and best performance per Watt by far.

He predicted, “AI6 will take that much further.”

In another post, he emphasised the architecture differences, saying, “Switching from doing 2 chip architectures to 1 means all our silicon talent is focused on making 1 incredible chip.”

Membership

Stay ahead with AI insights.

Unlimited access to premium articles, exclusive analysis, and the stories shaping India's AI economy.

Subscribe now

AIM Americas

1190 Miraloma Way Suite P, Sunnyvale, CA 94085

AIM India

1st Floor, Sakti Statesman, Bellandur, Bengaluru – 560103

© 2026 AIM. All rights reserved.