① Elon Musk stated that the Tesla AI5 chip design team has completed an outstanding design review, and the AI5 may be the best inference chip for models with fewer than 250 billion parameters; ② Previously, Tesla halted its chip design project Dojo; ③ The AI5 and AI6 chips, which Tesla is vigorously developing, are expected to support the training of its artificial intelligence and autonomous driving.
Tesla has revealed that it is internally assessing the integration of thousands of AI5 chips for the training of next-generation artificial intelligence models. This assessment appears to have yielded positive results.
On September 7 local time, Tesla CEO Elon Musk posted on social media that he just completed an outstanding design review with the Tesla AI5 chip design team, stating that this chip will become an 'epic' product, and the subsequent AI6 chip is expected to be the most outstanding AI chip to date. Musk added that he believes the AI5 may be the best inference chip for models with fewer than 250 billion parameters, as 'it has the lowest silicon cost and the highest performance-to-power ratio; AI6 will go even further.'
Previously in August, Tesla halted its chip design project Dojo, and project leader Peter Bannon will also leave Tesla. Musk revealed that Tesla decided to shut down the Dojo project because it was not reasonable to spread resources across two different AI chip designs. Tesla had previously adopted a 'training + inference' dual-track parallel chip development strategy, where Dojo was used for training and the FSD chip was used for in-car inference.
Musk's latest remarks again emphasize that shifting from developing two chip architectures to focusing on a single architecture means that all of Tesla's chip talent will concentrate on creating this exceptional chip. 'It seems that this is undoubtedly an obvious and correct decision. If you want to be involved in the development of a 'life-saving chip,' we welcome you to join the Tesla chip team. It's a matter of life and death, and every second counts.'
The AI5 and AI6 chips that Tesla is vigorously developing are expected to support the training of its artificial intelligence and autonomous driving. It is reported that the AI5 is a transitional or main chip for specific scenarios, produced by Taiwan Semiconductor; the AI6 will serve as the 'unified heart' of Tesla's future AI ecosystem and will be manufactured by Samsung Electronics.
Based on Tesla's previous statements, the AI5 will focus on training computational clusters for vehicle inference (e.g., autonomous driving) and is expected to enter mass production by the end of 2026; the AI6 chip will first be applied in Tesla's robotic taxi, Cybercab, and humanoid robot, Optimus, and may later expand to AI data centers, challenging the industry leadership position of companies like NVIDIA's H200 GPU. The first samples of the AI6 are expected to begin production at Samsung Electronics' foundry and packaging plant in South Korea, followed by mass production at Samsung's foundry in Texas, USA, which is anticipated to commence operations in 2025.
On September 2, Tesla officially released the fourth chapter of its 'Master Plan' (Master Plan Part 4). The self-developed chips by Tesla are a key step in its 'Master Plan,' which is expected to bring significant performance improvements and cost optimization to Tesla's future core products, as well as providing a greater advantage in the integration of hardware and software, reducing reliance on external suppliers and providing a computational power foundation for the rapid iteration of its autonomous driving technology and robotic products.
Edited by Lambor
