Samsung Foundry has completed the tape-out for Tesla's next-gen autonomous driving AI5 chip, manufactured on its 2nm process, with mass production planned at the Taylor, Texas facility starting in 2027, according to industry sources. A now-deleted LinkedIn post by Samsung principal engineer James Kim confirmed the milestone, which sparked broad market attention.

Tape-out marks the conclusion of front-end design and the handoff to manufacturing. However, large-scale production still requires mask fabrication, engineering sample validation, and product qualification over the coming months.

In April, Elon Musk disclosed that Tesla had submitted the AI5 design to both Samsung and TSMC for evaluation. Due to process differences, chips from each foundry will vary in detail. Samsung has now completed process customization for Tesla's architecture, enabling engineering sample runs.

The use of 2nm for AI5 – previously expected for AI6 – suggests Samsung's 2nm yield has exceeded 60%, aligning with recent industry chatter. Samsung has also been rumored to be in talks with Anthropic for AI chip production. The company declined to comment on customer-specific matters.

On deployment, Musk confirmed in the Q1 2026 earnings call that AI5 will first power Optimus humanoid robots and Tesla's AI supercomputers – not passenger vehicles. Tesla will introduce an upgraded AI4.1 with double memory bandwidth and ~10% more compute for vehicles next year. AI5 will only transition to cars after AI4 supply winds down.

From ICgoodFind: 2nm tape-out is real, and Samsung just passed the gate. The AI5 win matters less for cars in 2026 and more for what it signals: Samsung is now a legitimate 2nm foundry player – and that changes the bidding dynamic for every hyperscaler chip.