In July, Musk wrote on his social media platform X, formerly known as Twitter, “Samsung’s massive new fab in Texas will focus on producing Tesla’s next-generation AI6 chip,” revealing that he had placed an order with Samsung for AI6. At the time, he also stated, “Samsung Electronics is currently making AI4,” and added, “The recently completed AI5 design will be produced first by TSMC in Taiwan, followed by production in Arizona.” His latest comments thus clarify earlier reports suggesting TSMC had secured the entire AI5 order.

The fifth-generation autopilot chipset AI5 is expected to enter mass production at the end of 2026, targeting a performance of up to 2,500 tera operations per second (TOPS). The sixth-generation AI6, scheduled for release between 2027 and 2028, aims for 5,000 to 6,000 TOPS. Musk reaffirmed his plan to deploy Tesla’s proprietary AI chipsets not only in its vehicles but also across the broader Tesla ecosystem, including its humanoid robot “Optimus.”

“Our clear goal is to ensure an oversupply of AI5 chips,” Musk said. “If we end up with too many AI chipsets for cars and robots, we can simply use them in data centers.” He added, “We’re not trying to replace Nvidia. Nvidia has to supply chips to many customers, but Tesla only needs to meet its own demand.”

Source: BusinessKorea