For most of my career, if you were a good software engineer and wanted to be paid well, the advice was simple: go and work for a tech company. Banks, hospitals and insurers were where you went for stability (read “boring”) and a decent pension, not for the top salaries. So a CNBC piece from 29 October caught my attention, because according to Randstad, that’s no longer true.

Randstad’s Tech Talent Report 2026 says healthcare and life sciences companies now pay software engineers up to 17% more than tech companies do, and banks pay a 13% premium. It’s based on a survey of 1,100 business and technology decision-makers in 11 countries, plus around 20 million job postings and 100 million candidate profiles, so it’s not a back-of-the-envelope number. Job postings that require AI skills grew fivefold in the past year, and 55% of the non-tech organisations surveyed said they were recruiting engineers coming out of tech layoffs.

And there are plenty of those: Challenger, Gray & Christmas counted 165,925 job cuts announced by US tech companies in the first nine months of 2026, up 54% on the same period last year. So the talent is leaving tech companies, and non-tech companies are standing at the door with bigger cheques. What gives?

Why the Money Moved

I think the explanation is fairly simple, and Randstad’s CEO more or less said it: the first wave of AI money went to the companies building AI, and the second wave is going to the companies applying it. A bank has decades of messy processes, regulated data and very expensive manual work. An engineer who can point an agent at that and take out a few weeks of paperwork is worth a lot of money to them, arguably more than to a SaaS company that’s already automated most of what it does.

There’s another reason, and it’s one I feel quite strongly about. When a company is paying eye-watering per-seat prices for its CRM, its ticketing system and half a dozen other SaaS tools, there comes a point where building its own starts to look cheap. With agents writing a good chunk of the code, that point arrives a lot sooner than it used to. I believe that’s where a lot of expensive SaaS is heading: replaced by in-house tools written for exactly what the business needs, with nobody paying for a thousand features they never use. And who builds those tools? Software engineers, hired by the banks and hospitals that used to buy the licences.

What It Means If You Hire Engineers

If you’re a CTO at a tech company, your competition for talent is no longer only the startup down the road; it’s also a bank with a bigger budget and a story about “real-world impact”. Telling candidates “we’re a tech company, it’s cool here” doesn’t work when the hospital pays 17% more. What you can offer instead is ownership, speed and the chance to work on the hard problems, but only if those are actually true in your organisation. If a decision takes three committees and a fortnight, the bank is just as slow and pays better 🙂

If you’re a CTO at a non-tech company, this is your moment, so don’t waste it. I see a lot of these organisations hire great engineers and then put them in a structure designed for buying software rather than building it: long procurement cycles, IT tickets for everything, a change board for a one-line fix. Engineers coming from tech companies will put up with a lot for the right salary, but not forever. Give them small teams, real ownership and permission to ship, or they’ll be back on the market in 18 months.

And If You’re the Engineer

Here’s the thing I’d tell any engineer reading this report: domain knowledge has just become a lot more valuable. Writing code is increasingly the agent’s job; knowing which problem is worth solving, how a claims process actually works, or why a clinical trial’s data has to be handled a certain way is not something an agent picks up from a prompt. The engineers who combine a solid grasp of AI tools with a deep understanding of one industry will, quite likely, have the best few years of their careers ahead of them.

Sure, salary surveys come and go, and I wouldn’t change jobs on the strength of one report. But the direction here matches what I keep seeing with the companies I advise: the interesting AI work is moving from the companies building the models to the companies using them.

Bottom line: the best-paid engineering job of the next few years might well be at a company that doesn’t call itself a tech company. Go figure.