Thomson Reuters cut 500 engineers to hire 250 AI-native ones. The talent pool has 47 people.

Thomson Reuters just cut 500 engineering roles while growing revenue 10%.

Not a distress layoff. A calculated swap.

The company wants 250+ senior "AI-native" engineers over two years — people who have shipped production LLM features, not just used Copilot.

Here's the problem: the pool of senior LLM-skilled engineers at legal, tax, and regulatory companies across the US, UK, India, and Canada is 47 people.

That's a 5.3x gap between demand and supply.

And Bloomberg Law, LexisNexis, and Wolters Kluwer are about to run the same playbook. They'll be fishing in the same 47-person pond.

The QA and test automation teams absorbed the heaviest cuts — 135 SDETs and test automation engineers across India, UK, and US. These aren't obsolete workers. Many already integrated Copilot and LLM tooling into their workflows. They were cut because their seats made the spreadsheet work.

Thomson Reuters protected Canada while cutting India, UK, and US. Toronto has 1 senior LLM engineer in the talent index. Bengaluru has 5. That's not strategy — it's political arithmetic.

Watch the salary bands on the first 20 AI-native requisitions. If they match Big Tech comp, Thomson Reuters is serious. If they don't, expect the definition of "AI-native" to quietly shrink to "has shipped a Copilot integration" by month six.

Audit your own engineering workforce today. Count how many people have shipped production LLM features — not used them, shipped them. That number is your actual AI capacity. Everything else is a slide deck.