Home
Tesla AI5 Chip Tape-Out Complete After Musk's Direct Oversight, Rivaling Blackwell with Dual-Chip Power
Tesla has reached a pivotal milestone in its journey to develop custom silicon.
Elon Musk recently announced on X that the next-generation AI5 chip has completed tape-out. This signifies that the final design has been sent to the semiconductor foundry and has entered the physical manufacturing stage. Mass production is slated for 2027, at which point AI5 will fully succeed the current AI4, serving as the core computing platform for Tesla's Full Self-Driving system and Optimus humanoid robot projects.
The performance specifications are striking. Based on earlier information, a single AI5 chip delivers performance comparable to NVIDIA's Hopper architecture, while a dual-chip configuration approaches the level of Blackwell. Crucially, both cost and power consumption are substantially lower than equivalent NVIDIA offerings. Musk has previously indicated that AI5's key performance metrics represent an approximately 40x improvement over AI4, with a 9x increase in memory and an 8x boost in compute power. Furthermore, AI5 is projected to be the optimal inference chip for models with under 250 billion parameters.

On the manufacturing front, AI5 production is a joint effort between Samsung and TSMC. Fabrication will occur at Samsung's facility in Taylor, Texas, and TSMC's plant in Arizona, utilizing entirely U.S.-based production capacity. Notably, in his announcement post, Musk mistakenly tagged an account belonging to a different semiconductor company with a similar name to TSMC, causing a brief stir on social media.
In his post, Musk unusually highlighted the intense pressure behind this chip: "Solving AI5 is critical for Tesla's survival. We had to dedicate two full teams to it concurrently, and I personally spent several months working on it every Saturday." With the tape-out now successfully completed, he stated he finally has the bandwidth to resume development on the previously paused Dojo3 supercomputer processor project.
Development of the AI6 chip is also progressing according to plan. Tesla's roadmap for its proprietary chips is steadily advancing.
Related article
Stripe to Acquire OpenRouter as AI Model Routing Gains Momentum
Stripe has agreed to acquire OpenRouter, an AI model-routing platform that gives developers access to hundreds of models through a single interface. The deal adds model selection and routing to Stripe’s existing work around AI usage and token-based b
Former OpenAI Researcher Barrett Zoph Returns to Google to Support Gemini Development
Barret Zoph, who previously abruptly left and joined OpenAI due to a disagreement with Thinking Machines Lab CEO Miriam Mularati, has once again changed employers. Alphabet's Google announced that Zoph will return as Vice President of Research, bring
How Recursive Intelligence Raised $335M at $4B Valuation in 4 Months
The co-founders of startup Ricursive Intelligence seemed destined to be co-founders.Anna Goldie, CEO, and Azalia Mirhoseini, CTO, are so well-known in the AI community that they were among those AI engineers who “got those weird emails from Zuckerber
Related Special Topic Recommendations
Comments (0)
0/500
Tesla has reached a pivotal milestone in its journey to develop custom silicon.
Elon Musk recently announced on X that the next-generation AI5 chip has completed tape-out. This signifies that the final design has been sent to the semiconductor foundry and has entered the physical manufacturing stage. Mass production is slated for 2027, at which point AI5 will fully succeed the current AI4, serving as the core computing platform for Tesla's Full Self-Driving system and Optimus humanoid robot projects.
The performance specifications are striking. Based on earlier information, a single AI5 chip delivers performance comparable to NVIDIA's Hopper architecture, while a dual-chip configuration approaches the level of Blackwell. Crucially, both cost and power consumption are substantially lower than equivalent NVIDIA offerings. Musk has previously indicated that AI5's key performance metrics represent an approximately 40x improvement over AI4, with a 9x increase in memory and an 8x boost in compute power. Furthermore, AI5 is projected to be the optimal inference chip for models with under 250 billion parameters.

On the manufacturing front, AI5 production is a joint effort between Samsung and TSMC. Fabrication will occur at Samsung's facility in Taylor, Texas, and TSMC's plant in Arizona, utilizing entirely U.S.-based production capacity. Notably, in his announcement post, Musk mistakenly tagged an account belonging to a different semiconductor company with a similar name to TSMC, causing a brief stir on social media.
In his post, Musk unusually highlighted the intense pressure behind this chip: "Solving AI5 is critical for Tesla's survival. We had to dedicate two full teams to it concurrently, and I personally spent several months working on it every Saturday." With the tape-out now successfully completed, he stated he finally has the bandwidth to resume development on the previously paused Dojo3 supercomputer processor project.
Development of the AI6 chip is also progressing according to plan. Tesla's roadmap for its proprietary chips is steadily advancing.
Stripe to Acquire OpenRouter as AI Model Routing Gains Momentum
Stripe has agreed to acquire OpenRouter, an AI model-routing platform that gives developers access to hundreds of models through a single interface. The deal adds model selection and routing to Stripe’s existing work around AI usage and token-based b
Former OpenAI Researcher Barrett Zoph Returns to Google to Support Gemini Development
Barret Zoph, who previously abruptly left and joined OpenAI due to a disagreement with Thinking Machines Lab CEO Miriam Mularati, has once again changed employers. Alphabet's Google announced that Zoph will return as Vice President of Research, bring
How Recursive Intelligence Raised $335M at $4B Valuation in 4 Months
The co-founders of startup Ricursive Intelligence seemed destined to be co-founders.Anna Goldie, CEO, and Azalia Mirhoseini, CTO, are so well-known in the AI community that they were among those AI engineers who “got those weird emails from Zuckerber











