Summary of "Кто останется в ИТ к 2030 году?"

Summary of the Video’s Main Points (IT Employment to 2030)

The speaker argues that the shrinking availability of IT jobs—and the fear that “AI will replace everyone”—are driven by a mix of economic pressure and rapid technology change. By 2030, the IT workforce will be “thinned out,” with demand moving toward higher-level engineering work rather than basic execution.

1) Why the market tightened (the “perfect storm”)

Overheating of the IT job market (roughly 10–15 years ago) IT was marketed as a quick path to good income (e.g., “digital nomad” prospects), with the promise that short training could lead to employment. Many people entered IT through courses.
COVID-era hiring surge Companies rushed to hire anyone with keyword skills to build digital services and support remote work—often with the plan to “figure out what to do later.” This led to many hires that didn’t necessarily create proportional value.
Post-COVID economic impacts + wartime/armed conflict pressures Prolonged conflicts drained resources, reduced overall spending, and triggered cost cutting, including layoffs and removing staff who didn’t directly bring profit.
Cheap, capable AI tools (e.g., GPT and similar models) AI reduces the time and effort needed for tasks that previously required beginners or junior specialists, meaning fewer entry-level roles are needed.

2) What matters now in IT: value shifts from coding to engineering thinking

Writing scripts/code alone is no longer a differentiator.
Because AI can generate “basic” scripts quickly, competitive advantage shifts to:
- understanding the business problem
- architectural thinking
- experience from real projects
- planning and running migrations, implementations, and changes
- assessing risks, verifying results, and safely bringing solutions into production

The speaker summarizes the change as: engineers will be valued for understanding what/why/how to implement—not for “wow, I can code.”

3) AI as a productivity multiplier—unless companies can’t trust it fully

AI is described as a “multiplexer of your knowledge.”
With the “right patterns” and proper use, tasks can drop from days to hours (sometimes by an order of magnitude).
However, AI can hallucinate, and its output may change after updates—so organizations cannot fully trust AI. This increases the importance of human verification.
Conclusion: engineers who can work effectively with LLMs will be much more productive, while those who refuse AI will fall behind.

4) “Sovereign” / controlled-cloud trend changes engineering requirements

The talk highlights a shift from “Cloud First” toward “Control First.”
Companies want local control of data and services (or at least copies/keys in their own environments) to improve resilience and meet compliance needs—especially for critical systems.
Examples referenced include Microsoft-related concepts such as:
- Azure Arc/Local
- Microsoft 365.local
- sovereign private cloud approaches

For engineers, this means designs must account for:

local data copies
shutdown/offline scenarios
compliance, backups, and resilience

5) IT support will be restructured: more automation, less “tier-1 routing”

Language is becoming less of a barrier due to fast translation layers for support and communication.
With LLMs, support teams can:
- analyze tickets, logs, and errors more effectively
- automate repetitive L1/L2 template issues

The speaker predicts:

simpler queries will be resolved automatically and stay resolved
support value will shift toward diagnostics, which requires real skill
“forwarding tickets” roles are at risk because AI handles the low-level work

6) Who is “at risk” vs. who will “remain” in IT by 2030

At risk

People who merely follow instructions (click-and-go work), especially as AI agents become widely available.
L1 support roles that are largely “try again / reboot” style.
Administrators who operate only from ready-made guides and don’t understand architecture.
Coders writing standard code to assignments without understanding the surrounding system.
Engineers who refuse to use AI (they become slower and less competitive).

More likely to remain (or grow)

Architects and engineers who design complex systems and have proven experience from real projects.
Serious developers who understand architecture, product, and production rollout.
Security/identity specialists, because security problems continue and require human oversight.
Engineers responsible for production operations and hybrid/sovereign cloud environments.
In support: those who evolve into an engineering-like diagnostic role.

A key requirement for everyone who survives: the ability to use AI and verify AI results—solving real problems rather than just following instructions.