First of all, in the discussion around the demand for "juniors", it is striking that the numbers do not add up. On the one hand, Minister Maksut Shadaev stated in May at the TAdviser Summit: there is no total hiring freeze, IT companies are not firing developers because AI now writes code, the hype is exaggerated, and the demand for specialists remains. On the other hand, graduates of specialized faculties increasingly complain that they send hundreds of applications for vacancies and receive no response.
There is, of course, no contradiction in this situation: the shortage is real, but it is all at the upper levels. By the beginning of the year, only about 10% of vacancies remained that accepted candidates without experience: entry into the profession for beginners is gradually closing.
The minister himself explains this as follows: the hype around the industry led many to enter IT after online courses. As a result, there was an excess of "juniors" who "say they can do everything" — but companies need strong developers; the quality of training does not keep pace with the increasingly complex developments that businesses undertake. This explanation is fair, but incomplete: it does not cover a graduate of a specialized faculty who has four years of training, a diploma — and the same situation with rejections for vacancies. Therefore, the issue here is not only about the quality of personnel training.
I would not draw an analogy with lawyers and managers in terms of personnel retraining; it seems far-fetched. Lawyers were indeed produced in greater numbers than the economy needed, but this analogy is not entirely applicable to "IT specialists." In five years, the industry has grown by one and a half times and exceeded a million people (according to the minister's estimate), but the market still finds this insufficient. The fact is that businesses demand employees who are ready to deliver results immediately, not in a year. And it is fairer to speak of overproduction not of "IT specialists" in general, but of "entry-level IT specialists." These are different diagnoses, and they are treated differently.
Is AI to blame for everything?
Further discussion about personnel problems usually concerns AI. This argument is convenient: technology is to blame, no more questions. But real market observations show that "juniors" have become more expensive even without any AI: insurance premiums for IT companies have increased, and beginners along with junior "middles" have become more expensive for businesses. Hence, hiring filters are stricter, and the probationary period is extended to six months (offers like "3 months as an intern without salary, and then we'll see" are no longer uncommon). Add to this situation the general market contraction: there are about a third fewer vacancies than a year ago. Add Outlines Tech's estimate that AI cuts the budget by 30–50% for typical positions.
The role of neural networks is different: they took over precisely those routine tasks on which a novice used to gain experience, and made the savings too obvious to refuse.
It's dangerous to forget that "middles" grow out of "juniors"
IT directors of large companies understand that this situation is dangerous due to a decrease in the number of "middles" years later — but they say they need to save now, not in a year.
A familiar IT director of a large company formulated it simply: "We hire juniors, but only enough to replenish the natural attrition of middles. Someone grew up, someone left." Based on the current market situation, he is one hundred percent right. But together, all such companies will find a market without a middle tier in three to five years.
According to Ichak Adizes' management methodology, this is a classic case of short-term results dominating development: we saved the quarter and the year, but undermined the ability to exist in two or three years. Moreover, it will most likely be noticed not by the one who made the decision, but by their successor.
Why some companies are returning to hiring "juniors"
And this is where something that is almost never written about in headlines begins. Some companies that replaced juniors with AI have already returned to hiring. The reason is simple. What the model generated, someone has to check — and for this work, a junior turns out to be cheaper than a middle. In the US, after a year of freeze, large businesses are hiring people again; relying solely on AI did not pay off.
For the owner, it all comes down to mathematics. If you hire a junior now — in a year you will have a middle (or middle-minus), grown internally, in fierce competition, with adequate salary expectations and loyalty that cannot be bought with an offer. If you don't — in a year you will go to the market for the same person, where a middle-minus is reliably employed by the one who nurtured them, and a middle-plus will dictate their own terms and not react to anything but money.
"Junior" is still the cheapest way to get a "middle." The only difference is that the first path requires patience and a mentor, while the second requires a budget that may not be available at the right time.
We also need to talk about AI and the speed of growth — but carefully, because it's easy to make mistakes here. In my experience, with a neural network, a novice grows faster, but only if they are taught to check what is generated. For those who are not taught, AI turns them into a chat operator: the code exists, but there is no understanding of it. In a year, such a person looks like a "middle," but falls apart at the first non-standard task.
What young specialists should do
Hence the advice to graduates: simply writing code can no longer be sold; a model does it, and it does it faster than you. You need to sell something else: the ability to take on a task entirely, critically check what AI has produced, and bring it to a working result. At the same time, clearly understand what this result is needed for by the business. Juniors who are responsible for the result, not for a "closed ticket," are hired even in today's market.
Does this mean that Soviet-era distribution should be brought back?
It is also important to remember the role of the state in these processes. The idea of bringing back distribution for those who studied on a budget comes up regularly — and does not find support even from the Ministry of Digital Development. The reason is simple: you can force a company to hire a person, but you cannot force it to develop them. The bottleneck here is not the number of graduates, but who trains them further. Therefore, it is more logical to pay not for a place in a university, but for mentorship within companies. An internship with a mentor and an employment obligation will be cheaper than buying out the shortage of middles at market price in three years.
Conclusion
Finally, the main point. From the outside, the question "Do we hire juniors or not?" looks like a question of HR budget. In essence, it is a decision about who the company will consist of in a few years — and such a decision cannot be made at the bottom of the business. Either the top person (owner or CEO) makes this choice consciously, or in two or three years they will explain to shareholders why development costs twice as much as a competitor who, in a difficult year, did not get rid of "juniors" like we did.