What Cognizant's 15,000 Job Cuts Tell Us About India's Real Problem

2026-07-05 — management consulting India

I was in a client call last month when someone mentioned Cognizant's layoffs. Everyone just nodded and moved on, like we'd all already priced it in.

We haven't.

Cognizant announced Project Leap back in April. Fifteen thousand jobs gone globally. Between twelve and thirteen thousand of those are in India. That's not a rounding error—that's roughly five percent of their entire Indian workforce, which sits at around 250,000 people. The CEO called it building a "broader and shorter pyramid," which is a phrase I've been thinking about a lot because it's doing real work in the room even though it's meaningless.

Fewer junior people. More machines. India absorbs the cost.

The company expects to spend somewhere between $230 million and $320 million on severance and restructuring. They're betting they'll save $200 to $300 million in the same year. Whether that actually happens depends entirely on whether their AI tools do what they say they'll do, which is—and I say this kindly—a question that usually has a disappointing answer when you check back in twelve months.

But here's the thing about the math. Cognizant employed this model for decades because it worked. You'd have a few senior people in expensive cities managing dozens of junior analysts in Bengaluru and Pune and Hyderabad, all billing out to clients at rates that looked reasonable. The margin was in the pyramid. Wide at the bottom, narrow at the top. AI doesn't disrupt the top. It erases the bottom. And nobody built more of a bottom than India did.

Actually, that's not quite right. The bottom wasn't just bigger in India—it was basically the whole country's IT sector. Six million people employed across the industry. Most of them doing exactly the work that AI is now replicating. Customer support. QA testing. Back-office code. The stuff that doesn't require creativity, just consistency.

So how many Indian jobs are we actually talking about?

Cognizant is fifteen percent of the story, more or less. The broader picture is worse. India ranked second globally in AI-related tech layoffs through 2026, behind only the United States. Fifty-six percent of all tech layoffs in 2026 explicitly cite AI as the driver. That's not speculation. That's what happened. And the companies doing the cutting aren't stupid—they're just doing the math first.

McKinsey laid off about 200 people last year. Not engineers. Support staff. Internal operations people. The reason I mention this is that McKinsey isn't cutting customer-facing roles. They're cutting overhead. If a management consulting firm has decided it can run itself with less human support because AI can handle those functions, then every client sitting in that office has just watched a live demonstration of how the economics are supposed to work at scale.

The pattern is everywhere now if you're looking. Tech support roles disappearing into chatbots. QA teams getting smaller while the testing tools get better, or at least get cheaper. Graduate hiring programmes being deferred. The savings are being reinvested in chips and AI infrastructure, which sits in California and Amsterdam. Not Chennai. Not Hyderabad.

I think most consultants—and I'm including myself here—have been talking about this wrong. We've been treating it like an IT services problem specific to India, when what it actually is looks more like a structural shift in how outsourcing models work at all. Clients stopped wanting to pay for junior-heavy teams about five years ago. It just took the tools catching up to make that preference stick.

Cognizant's Project Leap isn't a surprise if you've been listening to their earnings calls. The pressure was building. But the speed is different. This is a compressed response to what should have been gradual, which is why the severance bill is hundreds of millions and why the people absorbing this shock are in Indian tech cities and there's no retraining safety net on the other side.

The structural problem is the speed.