
According to Tesla CEO Elon Musk, the company has completed the tape-out of its next-generation artificial intelligence chip, AI5, marking a key milestone in its in-house semiconductor development program. Musk publicly congratulated Tesla’s chip design team on the achievement and expressed appreciation to Samsung Electronics and TSMC for their manufacturing support.
Tape-out refers to the finalization of a chip design and its release for fabrication. Musk also confirmed that Tesla is actively developing additional high-performance chips, including AI6 and Dojo 3, underscoring the company’s long-term roadmap in AI computing hardware. He noted that these chips will be deployed across Tesla’s autonomous driving systems and humanoid robotics platforms.
Industry sources indicate that AI5 is expected to enter mass production across multiple advanced semiconductor manufacturing sites, including TSMC facilities in Taiwan and Arizona, as well as Samsung Electronics’ foundry plant in Taylor, Texas. Full-scale production is projected around 2027, approximately one year after tape-out. Musk further suggested that AI5 could become one of the highest-volume AI chips ever produced.
In a notable strategic shift, AI5 is expected to be jointly manufactured by TSMC and Samsung Electronics, while AI6 is planned to be produced exclusively by Samsung Electronics. Musk previously indicated that AI6 tape-out could occur as early as December 2026, highlighting an accelerated development cadence for Tesla’s AI silicon roadmap.
A detail from Musk’s announcement photo drew industry attention, showing the marking “KR2613,” which is interpreted as indicating prototype production in the 13th week of 2026 at Samsung Electronics’ foundry operations in South Korea. This suggests that Samsung’s domestic fabrication line has played an important role in early AI chip prototyping.
The broader semiconductor supply chain also reflects strong collaboration across Korean memory and logic providers. Prototype systems reportedly incorporate high-bandwidth memory (HBM) solutions from SK Hynix, while Samsung’s own memory products are also part of the AI5 ecosystem. These components are critical to AI performance, particularly in high-compute workloads such as autonomous driving and robotics.
AI5 is built using Samsung Electronics’ advanced foundry process technology SF2T, a customized 2nm-class node designed for high-performance AI workloads. At its SAFE 2025 foundry forum, Samsung introduced its SF2P+ roadmap and highlighted SF2T as a tailored process targeting next-generation AI chips, including Tesla’s future designs.
Analysts note that the earlier-than-expected adoption of SF2T for AI5 suggests a deepening foundry partnership between Tesla and Samsung Electronics, strengthening their collaboration in advanced node development.
During the announcement, a minor incident also drew attention when Musk mistakenly tagged an unrelated account while referencing TSMC on social media platform X. TSMC does not currently operate an official standalone account on the platform.
With AI5 now completing tape-out, Tesla’s vertically integrated AI semiconductor strategy is expected to accelerate further. Alongside AI5, the company continues to advance its AI6 and Dojo 3 processors, signaling its broader effort to reduce reliance on external AI chip providers such as NVIDIA.