Elon Musk lays out the plans for Tesla’s future AI ecosystem, utilizing both Samsung & TSMC 2nm tech for AI6 & AI6.5 chips.
Tesla’s 2nm Duo Confirmed: AI6 Goes Samsung With LPDDR6 & AI6.5 Goes TSMC With Improved Performance
A few days ago, Tesla announced the successful tape out of its AI5 chip, made at Samsung. The chip is just one of the many custom silicon designs that Elon Musk and his enterprises are making to fulfill their in-house AI demand.
Musk also talked about the next-generation chips, such as AI6 and the Dojo3 supercomputer project. The company is aiming to move the production of its custom AI chips to Terafab once the project is completed, but will remain a key customer for TSMC and other semiconductor giants.
Best was working with such a great team of AI hardware & software engineers! It was more fun than going to parties on Saturdays by far.
Least awesome was that we had to make several design concessions to move fast, but were able to finish tapeout 45 days ahead of schedule.
AI6…
— Elon Musk (@elonmusk) April 15, 2026
While Terafab is still years away from completion, Tesla’s next-gen chips, namely AI6 and AI6.5, will be made at Samsung and TSMC, respectively. Confirming in a post on X, Musk shared some more details on the AI chip roadmap. While he is relieved that the AI5 design is complete and fabbed out, he did state that several design concessions have to be made for faster delivery of the chip, which is why they were able to achieve tape out 45 days ahead of schedule.
And while some of the original design plans have been missed out, the upcoming Tesla AI6 chip will address these. AI6 will be manufactured on Samsung’s 2nm process node technology in Texas, & will deliver double the performance over AI5. While AI5 leverages LPDDR5X memory, AI6 will be using the newer LPDDR6 memory standard, which addresses those design misses mentioned above.
After AI6, Tesla plans to make a further optimized version of the chip called AI6.5, which will further improve the performance, and this chip will be made using TSMC’s 2nm technology in Arizona.
One of the coolest features coming to AI6 and AI6.5 is the halving of the TRIP AI computation accelerators dedicated to SRAM. This will result in effective memory bandwidth to be an “order of magnitude” greater than the DRAM bandwidth for any calculations made within the SRAM cache. Tesla’s AI HW and Inference software lead also stated how he felt satisfied in deleting legacy blocks from previous Tesla IPs, leading to a more efficient and optimized chip for Tesla, SpaceX, and xAI’s future AI endeavors.
While AI5 is being made at both Samsung & TSMC with volume production slated between 2026 and 2027, the AI6 and AI6.5 chips should target a 2027-2029 timeframe.
About the author: A Software Engineer by training and a PC enthusiast by passion, Hassan Mujtaba serves as Wccftech’s Senior Editor for hardware section. With years of experience in the industry, he specializes in deep-dive technical analysis of next-generation CPU and GPU architectures, motherboards, and cooling solutions. His work involves not only breaking news on upcoming technologies but also extensive hands-on reviews and benchmarking.
Follow Wccftech on Google to get more of our news coverage in your feeds.
