The week's career story is a pincer movement. On one side, the biggest employers in tech are rewriting headcount math around AI capex — buyouts, layoffs, and re-orgs landing simultaneously. On the other, the same companies are reaching past output and into process: recording the keystrokes, screen captures, and click paths of the very employees they still employ, so their next agents can do that work without them. The middle layer of white-collar work is being squeezed from above by capital reallocation and from below by behavioral data collection. For anyone planning the next twelve months, the question is no longer whether to adopt AI tools — it's whether your role produces a work artifact that a model can be trained to replicate.
The Big Story
Snap cuts 1,000 jobs (16%) after CEO says 65% of new code is AI-generated · April 22 · [TechCrunch]
→ Snap shed 16% of its workforce (~1,000 employees), expecting $500M annual savings. Stock rose 7% on the news. The structural read: when AI handles the majority of new code, headcount math shifts from "engineering capacity" to "review capacity" — and the discount factor on senior reviewers goes up.
Also This Week
Meta to cut 8,000 jobs starting May 20 in first wave of AI restructuring · April 22, 2026 · [Quartz]
→ Roughly 10% of staff, concentrated in middle management and non-engineering, funding a $115-135B AI capex year under new CAIO Alexandr Wang.
Microsoft offers first-ever voluntary buyout to ~8,750 US employees · April 23, 2026 · [CNBC]
→ About 7% of the US workforce at senior director level and below, landing the same week as Meta's cuts — a coordinated reset of the senior-IC and middle-management layer across hyperscalers.
Google: 75% of all new code is AI-generated and engineer-approved · April 22, 2026 · [blog.google]
→ Sundar Pichai's Cloud Next disclosure: up from 50% last fall and ~30% in April 2025, with one complex migration finishing 6x faster than a year ago. The reviewer is now the role.
UKG lays off 950 in "AI-first" restructuring · April 21, 2026 · [HR Executive]
→ 6% of the workforce at the HR-software firm itself — when the people who sell workforce tools cut their own workforce, take the signal seriously.
Jensen Huang: "Most people will lose their job to somebody who uses AI" — not to AI itself · April 22, 2026 · [Fortune]
→ Speaking at Stanford GSB, Huang split with Anthropic's Dario Amodei, who recently predicted half of entry-level white-collar roles will vanish — same data, opposite advice.
From the Lab
BCG: AI services drove 25% of $14.4B 2025 revenue, growing 25% YoY · [Bloomberg]
→ AI plus tech services now exceed 40% of BCG's total revenue. For consultants, ML engineers thinking about going independent, and anyone selling AI implementation work: the enterprise wallet is real, the buyer is the C-suite, and the engagement size is partner-tier — not platform-tier. If you can package "we deliver an outcome with AI" rather than "we sell hours," the margin is there.
EU AI Act hiring-bias audit rules locked in — 105 days to enforcement · [asanify.com]
→ August 2 enforcement, €15M or 3% of global turnover for non-compliance, certified auditors already booking out. A new specialty is forming in real time: AI hiring-system auditor. If you sit anywhere near HR tech, ML fairness, or compliance, this is the next 12-month niche with measurable demand and almost no supply.
Worth Reading
- [Economists might be wrong about AI and jobs] — UChicago's Alex Imas argues the speed of AI rollout breaks the standard "labor smoothly reallocates" model. Q1 2026 logged nearly 80,000 tech layoffs, ~48% AI-attributed — the transition isn't theoretical anymore.
- ["Do I belong in tech anymore?" — design engineer's AI burnout post] — Front-page HN thread on unconsented meeting recordings, AI-written code reviews, and docs produced faster than humans can read them. Useful pulse on the cultural fault line you're hiring into or out of.
- [College instructor turns to typewriters to fight AI-written work] — 258 points, 241 comments. The education side of the same problem: when verification is harder than generation, institutions reach for analog. Worth thinking about if you're designing assessments, interviews, or onboarding.
Build the skill that doesn't fit on a screen recording.
— Alexis