Remember my recent post about how AI is turning the IT market pyramid into a rhombus? Juniors are disappearing, businesses only need AI-native tech leads.

Well, fresh research on the Western labor market has come out from Oliver Wyman (surveyed 400+ CEOs of global companies) and analysis from Yale University (an organization undesirable in Russia). The point there is not what is happening, but how it is happening. Here are three key insights that explain why entering IT now feels like storming a fortress:

1️⃣ The Illusion of Stability and "The Big Freeze"
The media expected AI to cause a bloody wave of mass layoffs. That didn't happen (global unemployment is at a minimum), so many relaxed: "Aha, AI won't replace anyone!".

Yale economists explain why this is survivor bias. We have entered the phase of "The Big Freeze". Companies are not firing people in droves, they simply stopped hiring new ones. As current employees (armed with AI) start covering more tasks, natural attrition is not compensated by new hiring. The headcount is not cut with an axe, it is carefully "dried out". For those already in the industry — everything is fine. For those trying to enter — a concrete wall.

2️⃣ A Computer Science Degree Is Now Worse Than a Humanities Degree
This is not a joke. According to the New York Fed, unemployment among recent Computer Science graduates has risen to 7%, surpassing even anthropologists and art historians.

The CEO survey puts a point on this issue: 43% of bosses (twice as many as a year ago) officially stated that they are purposefully reducing hiring for junior positions in the next two years. They are shifting focus to mid-levels. That same "diamond" instead of "pyramid" is now not just a theory, but a corporate strategy.

3️⃣ From Chatbots to Agentic AI
If a year ago we were talking about generating code snippets and autocomplete, now business is focused on agentic systems. The difference is fundamental: it's no longer automation of tasks, but automation of workflows. The agent itself breaks a feature into subtasks, calls the necessary APIs, writes code, catches errors and fixes them. A junior coder is no longer needed even as a layer between Jira and IDE. What is needed is an experienced engineer capable of supervising, reviewing, and architecturally controlling an army of synthetic idiots.

What to do about it?

If you are trying to break into development right now, forget the strategy "I'll learn Python syntax, Django and do CRUDs for food". This niche has been eaten.

Your task is to mimic a mid-level from the start.
▫️ Study system design.
▫️ Understand the domain (know how your code brings or saves money for the business).
▫️ Learn to orchestrate AI agents, not compete with them in typing speed on the keyboard.

#market_decides