Taiwan's semiconductor engineers pull 30% raises while U.S. tech cuts headcount

Robert Walters Taiwan released its fifteenth annual talent report last week, and one number sits at the center: engineers with AI and semiconductor experience are commanding pay increases up to 30 percent. Not equity packages, not retention bonuses—base salary, in a market where annual wage growth had settled into low single digits for the previous decade.

The same week, Wix confirmed it is cutting 20 percent of its workforce. The CEO pointed to currency swings and AI-driven restructuring. The layoffs followed an earnings miss and came with the usual language about efficiency and focus. But the timing matters. This is not a 2022 overhiring correction. This is a company deciding which roles can be delegated to tooling and which geographies no longer pencil.

The contrast is not subtle. Taiwan is absorbing talent at wage premiums that would have been unthinkable three years ago, while U.S. tech platforms are shedding roles that were considered core as recently as last quarter. The AI labor story is not about broad displacement or broad opportunity. It is about a handful of skill sets in a handful of places becoming structurally scarce, and everything else repricing downward.

We have seen wage premiums cluster before—data science in 2015, cloud infrastructure in 2018. But those were hiring waves. Companies built new teams. This cycle is different. The raises in Taiwan are going to people who already hold the jobs. The cuts at Wix are coming from people who were hired to do work that the company now believes can be automated or reorganized around smaller teams. The skill premium is hardening at the top, and the middle is narrowing.

What makes this durable is that the premium is not speculative. TSMC, MediaTek, and the rest of Taiwan's semiconductor ecosystem are not paying 30 percent more because they think AI will matter someday. They are paying it because the work is live, the timelines are tight, and the talent pool is small. Meanwhile, U.S. platforms are learning that the marginal productivity of a customer success associate or a junior content strategist can be replaced by a language model and a tighter workflow. The fantasies about full automation may be overblown, as the Mint piece suggests, but the reshuffling is real and it is moving faster than the aggregates show.

The labor market is not collapsing. It is segmenting. If you are an engineer in Hsinchu with five years of experience in neural network hardware, you have pricing power. If you are a generalist in Tel Aviv working in SaaS operations, your role is being re-scoped or eliminated. The distance between those two outcomes is widening, and geography is deciding more than it has in a decade.