Tesla’s next-generation AI5 chip, the successor to the AI4 hardware running current Autopilot and FSD computers, has started trial production — but it’s not going into cars first.
Where the chip is actually being made
Samsung’s Taylor, Texas fab began trial production of the AI5 on its 2nm-class process on September 15, 2026, according to Electrek’s reporting on the Samsung partnership, moving up roughly two months from the plant’s previously stated November ramp. Tesla is running a dual-foundry approach: Samsung is producing on its 2nm-class node in Texas, while TSMC is separately producing a version on its 3nm process, a split Tom’s Hardware reported is meant to diversify Tesla’s supply chain rather than rely on a single manufacturer.
Why your car isn’t getting one yet
Trial production is not the same as chips shipping in vehicles. Elon Musk has said Tesla needs “several hundred thousand completed AI5 boards line side” before it can switch vehicle production over to the new hardware, and that volume isn’t expected until mid-2027. In the meantime, AI5 output is going toward other priorities first, including Tesla’s data centers and the Optimus robot program, before it reaches the assembly line for cars.
That means Model 3, Model Y, and Cybertruck buyers ordering today are still getting AI4-based Autopilot/FSD computers, and that isn’t expected to change for at least several more quarters.
One channel’s take on what it means
The Electric Viking, a high-frequency EV news channel, covered the production start in a video posted September 19, 2026.
Treat the video as one channel’s read on the timeline. For the underlying facts — what’s in production, where, and when cars might actually get it — the reporting linked above is the source to check.
Photo by Andrey Matveev.