TSMC has posted a record profit, and a decades-long chip-industry veteran says the AI build-out has "broken the old demand cycle" — a claim that, if it holds, changes how buyers, suppliers and investors should read every quarter from here on.

The remark, made as the world's largest contract chipmaker reported its strongest-ever profit, lands at a moment when practitioners are already questioning whether the semiconductor cycle they've planned around for decades still exists in its old form.

What does 'broken the old demand cycle' actually mean?

For most of the industry's history, chip demand moved in familiar waves. Inventory builds, shortages, overcorrection, glut, repeat. Procurement teams timed purchases around it; finance teams budgeted around it.

The veteran's framing suggests AI infrastructure spending doesn't fit that pattern. Hyperscalers and AI labs are building capacity for compute they expect to need years out, not for units they can sell this quarter. That decouples orders from end-market sell-through in a way the old models don't capture.

If the observation is right, two practical consequences follow:

• Classic cycle-timing plays — waiting out a downturn to buy cheap capacity — may misfire if AI demand keeps absorbing leading-edge supply.
• Capacity planning shifts from quarterly forecast horizons to multi-year commitments, which changes contract structures and capex risk.

How strong was TSMC's result?

Record profit at TSMC is the concrete signal here. The company sits at the choke point of the AI supply chain: Nvidia's accelerators, Apple's silicon, and the bulk of advanced logic all run through its fabs.

When the dominant foundry posts its best-ever profit, it means the customers at the top of the stack are still paying up for advanced nodes — despite a year of public hand-wringing about whether AI spending has outrun returns.

One caveat for readers: a record at TSMC confirms demand from those who can afford leading-edge capacity. It says less about the broader chip market, where mature-node segments have spent much of the period in a separate, weaker cycle of their own.

Who has to reprice this?

For equipment makers and materials suppliers, a broken cycle cuts both ways. Sustained AI demand supports longer order books, but it also means the violent inventory corrections that used to reset prices and clear excess capacity may not arrive on schedule.

For investors, the comment is a nudge that the sector's historical multiple logic — cheap at the trough, expensive at the peak — may need recalibration if the troughs stop arriving when the spreadsheets say they should.

And for anyone building AI products, TSMC's record is a quiet reminder that the scarce resource remains advanced compute. The constraint hasn't moved, even as the debate over AI's return on investment has gotten louder.

The bottom line

A record profit at TSMC plus a veteran's verdict that the old demand cycle is broken: one is a data point, the other is a thesis. Together they argue that the industry's operating assumptions deserve a fresh look — and that the next downturn, if it comes, may not look like the last one.