Q1 2026 final numbers came in this week.
78,557 tech workers laid off. Of those, 47.9% — roughly 37,000 people — were cut because AI and automation replaced their roles. Not restructuring. Not market slowdown. The explicit reason was AI.
That's the flood.
Here's the famine: 72% of enterprises running AI agents in production say they have a critical shortage of people who can actually architect autonomous workflows. The demand for what's being called "agentic engineers" — people who can design, build, and manage autonomous AI systems — is accelerating faster than the talent pool can form.
You have a flooded labor market and a famine happening at the same time.
Why the Flood Doesn't Fix the Famine
The 37,000 displaced workers aren't interchangeable with the engineers companies can't find. The profiles don't overlap.
The people being cut: junior developers who primarily wrote boilerplate code. Marketing and ops specialists whose work got automated. Project coordinators whose workflow management tools now run without human input. These are real skills, and real people, and their situation is brutal. But their skill profiles don't translate directly to building and running the autonomous systems that replaced them.
The people companies can't find: engineers who understand how to chain AI models into autonomous workflows, design evaluation frameworks that catch agent drift, build observability into AI systems before problems surface, and manage the operational discipline of running multiple agents in production without oversight failing.
That second profile is built from years of systems thinking, AI/ML exposure, and production engineering experience. You can't reskill for it in a bootcamp quarter. The gap between "worked in tech and got displaced" and "can architect an agentic system" is measured in years, not months.
What the Famine Actually Looks Like
If you've tried to hire for AI-native engineering roles in the last six months, you already know this.
The candidate pool looks deep on paper. A lot of people have "AI" somewhere in their resume. Most of them have prompt engineering experience, or used Claude/GPT to accelerate their work, or ran an internal pilot. That's real experience — but it's not the same as architecting systems that run autonomously at scale.
The interviews reveal the gap fast. They can talk about tools. They struggle when you push on operational design: how do you handle model drift in a deployed agent? What does your evaluation framework look like? How do you structure logging so a non-technical ops team can catch failures? What's your rollback plan when an agent produces bad output in a live workflow?
Those questions filter the flood down to a trickle. The genuine agentic engineers — people who've actually built and run these systems in production — are not on the market in significant numbers. They're employed, usually at the companies that built the capability early, and they're not looking.
The Displacement Timeline Is Running Against You
Here's the thing about the 37,000 people who got displaced by AI: some of them are going to reskill into the profiles companies need. The best of them will. The people who got cut and responded by going deep on AI systems, building agents on their own time, and coming back to the market in 12-18 months with real production experience — those people are going to be valuable.
That pipeline isn't here yet. It's forming.
The companies that will have access to it are the ones building relationships with those candidates now, before they complete the transition. Not when the reskilling is done and 30 other companies are also making offers.
The Actual Hiring Move
Two things happen at the same time in a market like this.
First: you source differently for agentic roles. The best candidates are probably not active. They're in a role, they're heads-down on something that interests them, and they're not responding to generic job postings. If you're posting for a "Senior AI Engineer" and waiting for applications, you're competing with every other company doing the same thing for the same trickle of candidates. You need someone in the market building the pipeline before you need it.
Second: you think harder about what the role actually requires. "Can you build AI agents" is a low bar. The bar that matters is operational: can you build something that still runs correctly in six months without someone watching it? Can you design the evaluation layer that catches drift before it causes a problem? That specificity in the job requirement filters better and attracts the engineers who've actually done it.
The flood gives you negotiating leverage on a lot of tech roles. For this specific one, you're still in a famine. Acting like it's a buyer's market when you're hiring for agentic architecture is the kind of strategic error that loses months.
VC5 Consulting works in the AI-native talent layer — sourcing and placing the engineers companies can't find through standard channels. If you're trying to build an agentic capability and the standard hiring process isn't producing — let's talk.