Samsung Completes Tesla AI5 Chip Tape-Out, Taylor Fab Set for Mass Production Next Year
Taylor Wilson
Tesla's next-generation AI5 chip has completed tape-out at Samsung and will be manufactured on a 2 nm process at the Taylor, Texas fab — Samsung's first mega-scale commercial order on 2 nm, and a direct test of whether its foundry unit can return to profit.
What does completing tape-out actually mean?
Samsung chief engineer Kim Jung-kun confirmed on LinkedIn: the Tesla-Samsung AI5 chip has completed tape-out and is expected to ship in Tesla products soon.
Tape-out — the milestone where a chip's final design is handed off to the factory — means AI5 has cleared the design phase and entered mass-production prep.
This means → the chip itself checks out; the next test is entirely on the fab: can yields hold up at scale?
When does the Taylor fab actually start running?
Samsung's new fab in Taylor, Texas is expected to begin initial operations by year-end, with full production for major clients starting next year.
AI5 is one of the first major volume projects the Taylor fab will handle — and Samsung's first time applying its 2 nm process to a mega-scale commercial order.
In plain terms = whether the Taylor fab gets a strong launch basically comes down to this single order.
How critical is this order for Samsung's finances?
Samsung's non-memory division is projected to lose roughly ₩600 billion in Q2; the memory division contributes about ₩8.4 trillion in operating profit, putting overall operating profit at an estimated ₩8.94 trillion.
This means → foundry is currently the only major segment still losing money inside Samsung's semiconductor business. Once Tesla shipments begin next year, the foundry unit has a path to breakeven.
The core variable has not changed: 2 nm mass-production yields must stabilize — if they do not, bigger orders only deepen losses.
How is Tesla structuring its chip supply chain?
The current AI4 and its upgraded variant are produced at Samsung's Pyeongtaek line in South Korea; AI5 volume will be split between Samsung and TSMC.
Further down the roadmap: AI6 is expected to go to Samsung; AI6.5 will be handled exclusively by TSMC.
This reflects Tesla deliberately building a dual-supplier system — Samsung carries continuity from current to next-gen products, while TSMC locks in the most advanced nodes.
Where will AI5 ultimately be deployed?
Tesla's AI chip family is designed for robots, autonomous vehicles, and data centers — all high-volume applications.
AI5 is the first chip on Tesla's roadmap to use a 2 nm process. The node jump delivers a major increase in compute power but also a corresponding rise in manufacturing difficulty.
Put simply = Tesla is handing its hardest-to-make chip to Samsung's newest factory — both sides are betting they can deliver.
Content is for reference only, not financial advice.