You updated your résumé last month. React, Figma, project management — the same skills you listed in 2024. And the recruiter messages went quiet. Not a trickle. Quiet.

The job listings that do appear now ask for "agent orchestration," "MCP integration," and "AI ops experience." Agent orchestration — coordinating autonomous AI programs the way a conductor manages an orchestra of bots. MCP — Model Context Protocol, a universal plug that lets AI tools talk to your company's data. These terms barely existed eighteen months ago. Now they sit between you and a paycheck.

The numbers landed

In the first two weeks of April 2026, three major data drops painted the same picture from different angles. On April 8, CNN Business cited Citadel Securities data showing software engineer listings actually grew 11% year-over-year. On April 9, Tom's Hardware reported that 78,557 tech workers lost their jobs in Q1 2026 — the highest quarterly total since early 2024. Analysts attributed roughly 48% of those cuts to AI automation, though only about 20% of companies explicitly named AI as the reason. Then on April 15, TechCrunch reported that LinkedIn's own data shows overall hiring down 20% since 2022.

Contradiction? No. Bifurcation.

Same headcount, different people

Dig into the numbers and the pattern clarifies. Harvard Business School research published in HBR on March 4, 2026 found that AI cut 17% of job postings in automation-heavy roles while increasing demand by 22% in positions built around human-AI collaboration. The Dallas Federal Reserve confirmed the mechanism on February 24: in computer systems design, wages climbed 16.7% since fall 2022 while employment dropped 5%. Fewer people, paid more, doing more — with AI.

Companies didn't fire workers and then hire AI. They fired workers without AI skills and hired workers with them. Total tech headcount across surveyed firms stayed roughly flat. The composition shifted.

The roles disappearing fastest — QA testing, Tier 1 support, content moderation, junior frontend — skew young. The roles growing fastest — AI engineering, agent operations, AI governance — require months of hands-on experience with tools that change every quarter. Anthropic's own Economic Index from March 25 measured the gap: power users with six-plus months of AI experience succeed 3–5 percentage points more often per task than newcomers. Modest. But compounding.

The retraining lag

Here's the uncomfortable part. Retraining programs run 6–18 months behind the tool cycle. Most coding bootcamps still teach pre-agent workflows — writing code line by line, not orchestrating AI agents that write it for you. The workers most displaced (junior roles, non-technical operators) are the ones least able to afford the monthly AI tool subscriptions that build the fluency employers now demand. BCG's March 2026 analysis estimates AI will reshape 50–55% of US jobs within two to three years but eliminate only 10–15% outright. Reshaping means your job title stays; the job inside it changes completely.

What to do with this

Three things that actually matter right now: push your employer for an AI training budget (it's cheaper than replacing you), spend thirty minutes a day using AI agents on real work tasks (not tutorials — real deliverables), and stop optimizing your résumé keywords and start building a portfolio of AI-augmented output.

The tech labor market forked in Q1 2026. AI-fluent and AI-illiterate workers now compete in different job markets with different salary curves. The fork is widening. Your six-month head start — or six-month delay — is hardening into structure. ⚙️