Tesla's next-generation AI chip, the AI5, has officially completed tape-out, according to reports from Inside.com.tw (E1) and CNYES (E2). Both outlets describe tape-out as the point at which the final chip design is handed off to the fab, marking the transition into large-scale production preparation. Neither report specifies an exact tape-out date, but both confirm the milestone has been reached.
The news was disclosed by Kim Jung-kyun, chief engineer at Samsung Foundry, via social media, according to both Inside.com.tw (E3) and CNYES (E4). He confirmed that the AI5 chip will be manufactured using Samsung's 2nm process technology — the first time this node has been applied to a hyperscale commercial order, per the same reporting.
Production will be based at Samsung's fab in Taylor, Texas, according to Inside.com.tw (E5) and CNYES (E6). Both reports state the Taylor facility is expected to begin initial operations by the end of this year and move into full production starting next year. AI5 is described as the fab's first major volume-production project once operational.
According to capacity planning cited by both Inside.com.tw (E7) and CNYES (E8), production of the AI5 chip will be shared jointly between Samsung and TSMC. The subsequent AI6 chip is expected to be handled by Samsung, while AI6.5 is planned to be produced exclusively by TSMC.
Samsung Electronics' non-memory division faced a loss of roughly KRW 600 billion in the second quarter of this year, as reported by both Inside.com.tw (E9) and CNYES (E10). Both sources attribute this pressure largely to prolonged strain in the foundry business.
Industry sources cited by Inside.com.tw (E11) and CNYES (E12) note that yield control on the 2nm process will be the key variable determining whether the foundry business can achieve stable profitability going forward.
| Metric | Value | Source |
|---|---|---|
| Process node for AI5 | 2nm | E3/E4 |
| Taylor fab initial operation | End of this year | E5/E6 |
| Taylor fab full production | Next year | E5/E6 |
| Samsung non-memory Q2 loss | ~KRW 600 billion | E9/E10 |
| AI5 foundry split | Samsung + TSMC | E7/E8 |
| AI6 foundry assignment | Samsung | E7/E8 |
| AI6.5 foundry assignment | TSMC | E7/E8 |
The reporting from Inside.com.tw and CNYES lays out a chain of connected facts: Samsung's 2nm process is being deployed for the first time on a hyperscale order (AI5) at a Taylor fab that is only just coming online, at the same time the non-memory division has posted a roughly KRW 600 billion quarterly loss. Industry sources quoted in both reports frame 2nm yield as the deciding factor for profitability — meaning the commercial outcome of the AI5 tape-out and the Taylor fab's 2026 ramp-up are tied directly to whether that yield issue is resolved. The split of future chips — AI5 shared, AI6 to Samsung, AI6.5 to TSMC — also shows Tesla diversifying its foundry dependence across the two companies rather than committing exclusively to one.
According to a report by TechNews (2026-07-14), Intel announced a €5 billion (about US$5.7 billion) investment to modernize its Leixlip, Ireland fab for Intel 3-based Xeon 6 and next-generation Xeon processors. The announcement follows a 2024 global 15% headcount cut that included a voluntary severance program at the same site, according to BusinessNext (bnext.com.tw).
According to TheLec and Tom's Hardware, Samsung Electronics' foundry division has completed tape-out of Tesla's AI5 autonomous-driving chip on a 2nm-class process at its Taylor, Texas fab, with mass production planned for next year — several months after the TSMC version reportedly taped out first.
According to a Bloomberg report carried by TechNews, NVIDIA has cut its list of approved Asian AI chip buyers by more than half under a new whitelist system. The Financial Times, cited by Liberty Times, adds that NVIDIA now conducts on-site audits in Singapore, Malaysia and Japan, with the U.S. Commerce Department involved in oversight.