— Sawyer Merritt (@SawyerMerritt) October 22, 2025 Musk's decision, confirmed during a recent Tesla earnings call, focused on the AI5 chip, which powers Tesla's latest Autopilot hardware and processes real-time self-driving signals. The plan is to have the accelerator manufactured at Samsung's massive new plant in Taylor, Texas, and TSMC's facility in Arizona. At the very least, this announcement serves as a major vote of confidence for Samsung’s lagging foundry business. Musk had previously indicated that TSMC would be the sole producer of the AI5, with Samsung—which already builds the preceding AI4 chip—only slated to handle the subsequent AI6 processor under a massive $16.5 billion contract. The Tesla CEO went further, offering a surprising endorsement of the South Korean firm's manufacturing edge. Musk claimed that Samsung's new Texas fab has slightly more advanced equipment than TSMC's Arizona site. Regardless of what you think of the man and his recent escapades, this praise is a timely boost for Samsung, which has struggled to close the gap with TSMC, the market leader in this arena. Tesla's rationale for the split is twofold: capacity and optimization. By using two top-tier U.S. facilities, Tesla aims to build an "oversupply" of AI5 chips to support both its vehicle production and its expanding data center needs. Tesla's AI5 (Credit: Elon Musk via X) Because Tesla is the chip's only customer, its engineers were able to simplify the design, removing legacy parts like the image-signal processor and GPU blocks. Musk stated that this allows the AI5 to be half the size of full-reticle designs from rivals like NVIDIA and AMD, as well as give the AI5 the best performance per dollar for AI, maybe by a factor of 10, and up to 40 times greater performance than the older AI4 (a.k.a. HW4) by certain metrics. Ultimately, the shared production deal is a win for both foundries and strengthens the U.S. semiconductor supply chain. For Samsung, the AI5 contract, coupled with the $16.5 billion AI6 deal, provides crucial stability and validation. For TSMC, it secures its position in a high-profile AI program, for now anyway.