When Anthropic CEO Dario Amodei, OpenAI's Sam Altman and Elon Musk all began publicly calling for a slower pace of AI capability development this year, Asian chip stocks briefly flinched. Japan's SoftBank fell 13.2%, South Korea's SK Hynix dropped 6.3%, and Samsung Electronics closed 4.05% lower on the news. Taiwan's TSMC, the company whose fabs make the advanced chips nearly all of that AI infrastructure runs on, moved barely at all — down just 1.2%.
The muted reaction lines up with what TSMC has actually been doing with its money. The company raised its 2026 capital expenditure forecast to a range of $60 billion to $64 billion, up from $52 billion to $56 billion, after posting second-quarter revenue of $40.2 billion on strong AI-related demand and lifting its full-year growth forecast to above 40%. Taiwan is currently home to 13 advanced fabs under construction, with TSMC planning to bring newer High-NA EUV lithography systems into production starting in 2030 — a timeline that assumes AI chip demand keeps climbing for years, not months.
That gap between market jitters and capital-spending reality reflects a structural feature of the chip business that AI capability debates don't easily move: fabs take five or more years to plan, permit and build. A decision to slow model development this year would not meaningfully change how much capacity TSMC needs online in 2029 or 2030, because that capacity was already committed based on demand assumptions locked in years ago. Taiwan's government has leaned into that momentum, raising its 2026 economic growth forecast to 7.7% on the strength of AI-related demand, a bet that treats the "should AI slow down" debate as noise sitting on top of a chip investment cycle that isn't going to reverse on a few weeks of headlines.
