Industry

Beyond the Pink Slip: The Doors AI Disruption Is Opening in Tech

Last month I wrote about what leading a team through the AI shift actually looks like day to day — hiring differently, measuring differently, protecting the space for someone to say "I don't fully understand this AI-generated suggestion" out loud. That post was written from inside one team, at one company. This month I want to zoom out, because the conversation across India's tech industry has been dominated by a much harder story: layoffs. TCS cutting roughly 12,200 roles. Industry estimates of the broader IT services sector shedding anywhere from 18,000 jobs this year to, some analysts warn, half a million over the next two to three years as AI eats into the delivery model that built the sector. Headlines calling it the age of the pink slip, warning that even people who still have jobs should be nervous. I don't think that framing is wrong, exactly. I think it's incomplete, and incomplete in a way that matters if you're trying to decide what to do next.

The Cuts Are Real. So Is What's Growing Underneath Them

Here's the part that rarely makes the same headline: AI hiring in India is projected to grow toward roughly 380,000 positions in 2026, against an estimated 53% skill deficit in the workforce available to fill them. TCS is cutting workforce concentrated in middle management and bench capacity, and in the same year training 100,000 of its own people in AI orchestration. Those aren't contradictory facts sitting in tension with each other. They're the same repricing of labor, viewed from two different sides of it. A market doesn't usually announce "we now pay for judgment, not hours" in a press release — it announces it by cutting the roles priced on hours and bidding up the roles priced on judgment, at the same time, inside the same companies.

What's Actually Being Cut

It's worth being precise about this rather than gesturing at "AI is taking jobs," because the precision changes what you should do about it. What's contracting fastest is headcount tied to a specific delivery model: large benches of engineers billed by the hour to execute well-specified, largely repetitive work — the model that built the Indian IT services industry over three decades and employed millions of people extraordinarily well. AI-assisted delivery pipelines compress exactly that category of work fastest, because it's the category with the clearest specification and the most historical examples to learn from. That doesn't mean everyone inside that model is replaceable as an individual. It means the economics of billing for volume, on its own, without judgment attached, are breaking down in real time. That's a narrower and more useful claim than "AI is taking jobs," and it points toward what to do differently rather than just what to be afraid of.

The industry isn't running out of work. It's running out of patience for work that doesn't require thinking.

The Judgment Premium, at Industry Scale

I wrote last month about shifting what I measure on my own team — away from PR counts and lines of code, toward how well someone can explain the reasoning behind a change, and whether they're building the kind of system understanding that lets them debug confidently under pressure. What's happening across the industry right now is the same shift, at a much larger scale. The roles growing fastest — AI/ML architect, cloud architect, data scientist, AI governance and safety specialist — aren't roles that execute a specification faster. They're roles that decide what the specification should be, and they're commanding salaries in the ₹15–40 lakh-plus range in India, well above the historical services-model band. That premium isn't arbitrary. It's the market pricing exactly the skill that AI tooling doesn't compress: knowing what to build, why, and what happens when it's wrong.

New Doors, Not Just Fewer Old Ones

The most useful reframe I've found, running technology for a company mid-flight on an AI-driven platform, is that this disruption isn't just closing doors on one style of career, it's opening a genuinely new set of them. Every industry vertical — logistics, healthcare, finance, retail — is now building or buying AI-native products the way we're building VisionWare+ for warehouse operations, and that creates real, durable demand for people who can sit at the boundary between a business domain and a system architecture, not just between a ticket and a pull request. Platform and AI-infrastructure engineering is its own growing discipline now, distinct from application development. Context and evaluation engineering — the work of making AI systems reliable inside a specific business process rather than generically capable — is becoming a legitimate specialization rather than a side skill. And as AI systems make more consequential decisions, governance, safety, and audit roles are moving from "nice to have" to something regulators and boards are starting to require. None of that existed as a career path five years ago. All of it is hiring now, and hiring against a documented skill shortage, not a surplus.

What I Actually Tell People Who Ask Me About This

I get asked some version of "should I be worried" often enough now that I've stopped giving a vague reassuring answer and started giving a specific one. Use the reskilling paths that exist and are genuinely well-funded right now — large employers are running serious AI orchestration and GenAI training programs internally, and industry-backed initiatives are training well beyond their own headcount. But treat that training as a floor, not a ceiling: fluency with a tool is table stakes now, not a differentiator, in the same way that knowing an IDE was never a differentiator. What differentiates is the same thing it's always been under different tooling — can you own a system end to end, reason about failure modes before they happen, and get uncomfortably close to the business domain you're building for rather than staying purely on the technical side of the fence. That's harder advice to act on than "learn to prompt better," and it's also the advice that actually holds up as the tooling keeps changing under everyone's feet.

What This Adds Up To

I don't think the layoff headlines are wrong to alarm people, and I'd be doing readers a disservice if I waved away the real disruption happening to a delivery model that supported millions of livelihoods. But an industry shedding volume-priced roles while simultaneously running a documented shortage of judgment-priced ones isn't an industry in decline — it's an industry in the middle of repricing itself, loudly and unevenly, with real casualties along the way. The migration from one side of that line to the other is disruptive, and no one owes anybody a soft landing into it. It is, all the same, a migration toward more interesting, better-compensated work for the people willing to make it, not simply toward less work overall. That's the part of the chart the headlines usually leave out.

← Back to all posts