TLDRocket
Sign in

Chip Expert Tzi-Dar Chiueh: “China Wanted Taiwan Long Before Our Chip Industry Succeeded”

Trending Topics Jakob Steinschaden

Taiwan makes about 70% of the world’s contract chips, but its AI models still come from the US or China. That gap is the whole story: hardware power, software dependence.

Based on reporting by Trending Topics, Jakob Steinschaden — read the original for the full story.

Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error

Taiwan’s chip industry looks like a paradox from the outside. The island makes roughly 70% of the world’s contract chips through TSMC, and those fabs produce the advanced AI chips used by Nvidia, Apple, Broadcom, Google and Amazon. Yet the AI models running on that silicon mostly come from somewhere else. Taiwan supplies the engine room. Others write the dashboard software.

That split is not accidental, according to Tzi-Dar Chiueh of National Taiwan University. He says Taiwan built its chip edge by pulling top students into electrical engineering, sending many of them to the United States for PhDs, and then combining that talent with government support and a growing industrial base. TSMC itself was spun off by the Taiwanese government in the 1980s, and the company’s rise was not quick. Chiueh says it was lagging well behind Intel, Samsung and AMD about 20 years ago.

The company’s reputation, he argues, rests as much on culture as on process. Taiwanese engineers are known for rushing back to fabs after earthquakes, even in the middle of the night, to check equipment and keep production moving. In the US, Chiueh says, that kind of reflex does not travel easily. Morris Chang, TSMC’s founder, has been openly pessimistic about how far that model can be copied abroad. Even with fabs in the US and Europe, the gap in cost and efficiency remains. Chiueh puts US production at at least 30 to 40 percent more expensive than Taiwan, and says some people claim it is two to three times higher.

Taiwan is trying to answer that pressure with a sharper industrial strategy. Defense spending is set to top 3% of GDP for the first time in 2027, which works out to about 30 billion euros, including about 1.1 billion euros for drones. In tech, though, the country is not trying to outspend the big players on giant AI models. The play is narrower: use open models from the US or China, train them on Taiwanese data, and build around them with local hardware and know-how. That is a pretty sober answer to a very loud industry.

My take — AI-written commentary, not fact-checked reporting

This is what real tech sovereignty looks like: not a flag on a slide deck, but the boring, expensive stuff that actually ships. Europe keeps talking about “our own” AI while happily renting the brain from elsewhere, which is a fine strategy if everyone enjoys being a customer. Open models plus local data is the practical route; pretending every region will train its own frontier model is just ceremonial software nationalism.

Read more about this at: Trending Topics

Related stories

The daily briefing

Every AI story that matters, in your inbox by 8am.

TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.