Samsung Electronics will hold an equipment installation ceremony at its Taylor, Texas foundry on April 24, 2026, kicking off the final run-up to mass production of Tesla’s AI5 and AI6 autonomous-driving chips on a 2-nanometer process. Han Jin-man, president of Samsung’s Foundry Business Division, will lead the ceremony alongside executives and representatives from materials, component, and equipment partners, according to a report today in South Korea’s Chosun Daily and corroborated by The Korea Herald.

The milestone lands three years and six months after Samsung broke ground on the site in November 2022. Taylor Fab 1 was originally scheduled to begin operations in October 2024, but Samsung delayed the ramp because it could not lock in enough orders to justify moving in expensive wafer fab equipment. The $16.5 billion Tesla deal signed in July 2025 ended that stalemate. Samsung has since pushed 2nm yields from roughly the 20% range in the second half of last year into the high 50% range, with 60% seen as the threshold for stable mass production.

If you drive a current Tesla, the AI4 chip inside it was made by Samsung. AI5 and AI6 are the silicon Tesla is betting its Full Self-Driving software, Optimus humanoid program, and robotaxi plans on, and Taylor is where the U.S.-made volume will come from.

Samsung Bet A $44 Billion Campus On Tesla’s Chip Roadmap

The Taylor site is a $44 billion campus built around a first fab module designed for 2nm production using Gate-All-Around transistor architecture. Samsung scrapped its original 4nm plan for Taylor once Tesla signed on. The July 2025 contract, worth approximately 23 trillion Korean won, runs through December 31, 2033. Elon Musk has publicly called the $16.5 billion figure a floor, not a ceiling, saying actual output is “likely to be several times higher.”

AI5 production is split between Samsung in Taylor and TSMC in Arizona. AI6 goes entirely to Samsung at Taylor. Musk also floated an AI6.5 variant on a 2nm TSMC Arizona node, adding a redundancy layer on top of the existing Samsung-TSMC split. Tesla has been clear about why: one foundry relationship is a single point of failure for a company whose roadmap now depends on custom silicon scaling faster than any supplier can deliver on its own.

Musk Called The AI5 Design Complete One Day Before The Ceremony

The timing is deliberate. On April 15, Musk posted on X that Tesla had “taped out” the AI5 chip, the point at which a design is finalized and handed to the foundries for production, and thanked both Samsung and TSMC. He called AI5 “one of the most produced AI chips ever.” One day later, Samsung confirmed it would begin moving 2nm tools into Taylor. That is not a coincidence, it is a coordinated signal to investors and customers that a chip the design team finished on Wednesday has a real production line waiting on Friday.

Musk has also said on a November 2025 earnings call that Samsung’s Taylor fab is “slightly more advanced” than TSMC’s Arizona site. The numbers back the claim at the node level. TSMC Arizona runs a 4nm process today. Samsung Taylor is launching on 2nm. Whether yield, throughput, and cost per wafer also back the claim is the question the next 12 months will answer.

Samsung Foundry Has Been Losing Money For Years

Samsung’s foundry division has posted multi-trillion-won annual losses and is betting the Taylor ramp is the inflection point. Industry estimates put third-quarter 2025 foundry losses below 1 trillion won ($700 million), down from much heavier losses earlier in the year. The stated goal is profitability by 2027. That timeline only works if Taylor moves from equipment installation to qualified 2nm volume output inside 18 months, which is aggressive for a site that has never produced a single commercial wafer.

The April 24 ceremony also signals a U.S. supply-chain hedge. Samsung’s Korean supplier cluster, including Dongjin Semichem, Soulbrain, FST, and Hanyang ENG, has spent 2025 setting up production and office sites in Taylor. That is what changes this from a single-plant story into a long-term U.S. foundry play.

Current Tesla Owners Face A Hardware Floor, Not An Immediate Change

For current Tesla owners on HW3 and HW4, nothing changes tomorrow. AI5 mass production is widely forecast for late 2026 or 2027, with AI6 following in 2027 to 2028. That means the first vehicles with AI5 silicon are likely a model-year 2027 story at the earliest. Tesla’s own Spring 2026 software update, which added an FSD subscription tap, 24-hour dashcam, and “Hey Grok” voice control, already gated new features to AI4 and newer hardware. That hardware floor is now effectively permanent. HW3 owners who want AI5-class FSD will need a hardware swap or a new vehicle.

EVXL’s Take

The industry delta here is obvious. Eighteen months ago, the Taylor fab was the textbook example of CHIPS Act over-promising, a $44 billion campus with empty cleanrooms and no anchor customer. Today it is the North American home of the chip Tesla is staking robotaxi, Optimus, and FSD on. That is a full reversal. And it happened because one customer, at one earnings call, decided that TSMC concentration risk was bigger than any pricing advantage of staying with the incumbent.

I have been writing about the Taylor fab since the 2024 delays, and I’ll say plainly: I did not expect Samsung to get to a 2nm move-in ceremony this fast. When Tesla announced Terafab in March and Musk warned that existing foundry capacity would not be enough in three to four years, I read it as Tesla hedging against Samsung slippage. This ceremony is Samsung’s answer to that hedge. The devil’s advocate case is real: a move-in ceremony is not qualified volume, Samsung’s 2nm yield is still below the 60% bar that serious customers demand, and history says Samsung Foundry slips. But every week that Taylor stays on schedule is a week Tesla’s chip supply story gets less fragile.

Samsung Taylor will begin qualified 2nm risk production of Tesla AI5 wafers by the end of Q2 2027, and Tesla’s first AI5-equipped production vehicles will land in customer driveways in the first half of 2028. If either slips more than two quarters, Terafab stops looking like a hedge and starts looking like a replacement.

FAQ

When does Samsung begin making Tesla chips in Texas? Equipment installation begins April 24, 2026. Industry reporting points to qualified 2nm volume production starting in the second half of 2027, with Tesla AI5 and AI6 chips as the first products.

What process node is Samsung Taylor using? A 2-nanometer process built on Gate-All-Around transistor architecture. Samsung’s comparable U.S. rival, TSMC Arizona, is running a 4nm process today.

How big is the Tesla deal? Approximately $16.5 billion, or 23 trillion Korean won, signed in July 2025 and running through December 31, 2033. Musk has said the figure is a floor, not a ceiling.

Will AI5 or AI6 ship in current Teslas? No. AI5 is expected in Tesla vehicles starting in late 2026 or 2027 at the earliest, with AI6 arriving later. Current production cars use AI4, which Samsung manufactures in Korea.

EVXL uses automated tools to support research and source retrieval. All reporting and editorial perspectives are by Haye Kesteloo.