Taiwan Semiconductor Manufacturing Company (TSMC) confirmed it is in discussions with Elon Musk's planned US chip factory, Terafab, but the world's largest contract chipmaker is drawing a firm line: if there's a partnership, TSMC must run the plant.
On TSMC's third-quarter 2026 earnings call, CEO C. C. Wei said the company has sent teams to Texas to evaluate Terafab and is exploring whether it could use some of that capacity. He described Terafab as “one of our customers who try to build a fab.” But Wei repeated TSMC's non-negotiable condition: in any alliance, TSMC must operate the fabrication plant.
Musk, meanwhile, posted on X that his companies would “build and run the fab,” with TSMC possibly subleasing a section “but nothing more.” That public back-and-forth highlights a fundamental disagreement over who would control the facility.
Why control matters so much
At the cutting edge of chipmaking, a fab is far more than a building full of expensive equipment. The real value lies in the operator's manufacturing playbook — the tightly controlled processes that determine yield (how many usable chips come from each silicon wafer) and consistency. TSMC's decades of refinement in these steps are a big reason its advanced chips are in such high demand.
If a partner provides the building and tools but TSMC doesn't run the line, it can't enforce the process controls that underpin its yields. And customers — from smartphone makers to AI companies — typically require months of validation before they treat new output as dependable supply. In other words, even if Terafab were to start producing chips, those wafers wouldn't automatically count toward anyone's supply plans until they've been thoroughly qualified.
This is why the control question is more than a corporate power struggle. It determines whether Terafab's output can be treated as real, bankable supply for the advanced chip market.
TSMC's own roadmap is the anchor
For now, TSMC's own expansion plans remain the clearest signal for investors. Wei pointed to volume production at a new 3-nanometer plant in Tainan in the first half of 2027, another 3-nanometer plant there in the first half of 2028, and volume production at its Arizona site in the third quarter of 2027.
These timelines are what markets are likely to focus on, because they represent capacity that TSMC fully controls and can guarantee to customers. The company's Arizona project has been a key part of its US strategy, and the 2027 target for volume production there is a major milestone.
Terafab, by contrast, looks more like a longer-term possibility. Until the control issue is resolved, it's hard for investors to model Terafab as a meaningful source of advanced chip supply in the next couple of years.
What it means for investors
For anyone following the chip sector, the takeaway is that “outside capacity” isn't plug-and-play when it comes to advanced semiconductors. If TSMC isn't running the production line, it can't enforce the process controls that underpin its yields, and customers won't count those wafers toward their plans until they're qualified — a process that tends to take quarters.
That's why markets are likely to keep their near-term assumptions tied to TSMC's 2027-2028 ramp in Taiwan and Arizona. Any Terafab collaboration may matter eventually, but until the control question is settled, it looks more like longer-dated optionality than extra 3-nanometer output investors can confidently model.
The broader context: demand for advanced chips, especially for AI applications, remains intense. Companies like Musk's ventures are pushing into chipmaking to secure supply, but the technical and operational barriers are steep. TSMC's insistence on control reflects its position as the dominant player in advanced manufacturing — and a reminder that in this industry, the playbook is as valuable as the plant.
For everyday investors, the key is to avoid assuming that every announced fab project will quickly add to global chip supply. The gap between announcement and qualified production is often measured in years, not months. TSMC's own roadmap, with its specific dates, is a more reliable guide to when new advanced capacity actually comes online.
As the talks with Terafab continue, the market will be watching for any sign of a compromise. But for now, the message from TSMC is clear: if it's not running the fab, it's not really their fab — and that means it's not yet part of the supply picture that matters.
