Lenny's tech job report: PM demand hits 4-year high as engineering stalls

When generative AI first exploded, everyone assumed software engineers would be the first ones out of a job. Two years later, reality delivered a wild plot twist: code has never been cheaper, but engineering job growth has completely stalled, designers are fighting for scraps, and product managers—once mocked as meeting-happy buzzword dispensers—are winning the AI era.

Tech veteran Lenny Rachitsky just published his late-2026 tech job report. Sourcing global openings from TrueUp and verified U.S. hiring records from Live Data Technologies, the biannual study paints a brutally clear picture of how AI is rewriting the org chart.

PMs soar while code factories stall

The most dramatic split is happening right down the hall between product and engineering.

Global tech companies currently have 7,303 open PM roles. That is a four-year high and a massive 75% rebound from the market trough in early 2023. Among the five core software disciplines—product, engineering, data, design, and user research—PM is the only function whose hiring share expanded, rising from 16% in 2021 to 21% today.

Software engineering, meanwhile, hit an unprecedented wall. For the first time in the report's history, open engineering headcount stopped growing, inching up just 2.5% over the past year. Openings now sit at 65,351, which is actually lower than this past March. Actual hiring tells a grimmer story: U.S. companies hired roughly 113,000 engineers over the past twelve months, less than half of the 239,000 hired at the 2022 peak.

Inside engineering, specialization is getting crushed. Front-end openings cratered by 51%, Android collapsed 47%, and iOS dropped 46%. But forward-deployed engineers (FDEs)—the technical operators embedded on-site to force-feed products into customer infrastructure—rocketed 4.6x. Palantir, Google, and Accenture are snapping them up in droves. With AI copilots handling the interface plumbing, dedicated front-end developers are vanishing into generalists, while customer-facing deployers are golden.

Design hits rock bottom as AI fluency takes over

If engineers are catching a cold, designers are in full hypothermia.

Global tech firms have only 1,924 product design roles open right now. That is an all-time record low, down 28% year-over-year. For every single design opening today, there are 3.8 PM roles waiting to be filled. User research fared even worse: openings shriveled to a miserable 108 globally, down 66% from 2022 levels. With AI speeding up prototype cycles, teams would rather ship and iterate in prod than wait weeks for a user research deck.

Meanwhile, job descriptions across every vertical have been colonized by AI buzzwords. Two years ago, PM listings asked for SQL and Jira. Today, "LLM" is the number one required skill, appearing in 11% of PM postings, followed by Claude, Cursor, OpenAI, and MCP. In engineering, LLM demand just leapfrogged Java. In design postings, mentions of Claude and Cursor have both surpassed plain CSS.

And if you were hoping to code from a beach, bad news: remote work is at a historic low of 14.4%. Meanwhile, nearly 15% of all tech openings worldwide are now packed into the San Francisco Bay Area, completely flipping the script on pandemic-era distributed work.

Why it matters

This data cuts through the speculative noise about AI replacing humans and reveals the actual corporate anatomy of 2026.

For tech workers, pure execution roles are being compressed out of existence. If your daily value is converting a Figma mockup into React components, or summarizing interview transcripts into user personas, AI tools have essentially commoditized you. The market is aggressively rewarding two distinct profiles: big-picture product architects who can orchestrate AI tools to direct business outcomes, and high-agency deployment engineers who can step into messy client rooms and wire complex models into legacy enterprise software.

For the industry, tech's stalled engineering headcount confirms that AI-driven productivity gains are real, but they look like leaner payrolls rather than massive layoffs. Small teams wielding AI agents can maintain codebases that once required dozens of junior developers.

We also think the sharp rebound of Bay Area concentration is telling. When product cycles move at breakneck speed, asynchronous Slack messages become a bottleneck. High-trust, in-person density is winning out. AI did not kill software, but it permanently dismantled the assembly line that built it.