India's GCC Boom Is Repeating a Mistake I Watched a Decade Ago

India will cross 2,100 Global Capability Centers this year — roughly 2.4 million people, close to $100 billion in export value, about 42% of the world's GCC market. Last year GCCs took 38% of all office leasing across the top seven cities. It's the biggest structural shift in Indian white-collar work since the IT-services wave.

I've watched this movie before. And most companies rushing in are pointing the camera at the wrong thing.

The number everyone quotes is the cost saved. That's the number that will betray them.

Twenty years ago I was building “follow-the-sun” delivery teams in India for a US credit-risk software company. The pitch to the board was always cost — 30, 40% cheaper. The teams set up on that logic, and only that logic, stayed brittle: a pool of capable hands executing decisions made somewhere else. The moment the arbitrage narrowed or a cheaper geography appeared, they were exposed — because cost was the only thing they were ever built to provide.

The teams that lasted — the ones where we held on to 85%+ of clients for years — were built on a different idea. They owned something. A product, a domain, a customer relationship, a decision. The cost saving was a by-product, not the point.

Why this matters more in 2026 than it did in 2006

Roughly 80% of new GCCs are being stood up “AI-first.” Good instinct — wrong reason, if the underlying logic is still cost. Because AI eats exactly the work that arbitrage centers were built to do: the rote, the repetitive, the cheap-hands layer. If your capability center exists because Indian execution is cheaper than American execution, generative AI is a cheaper competitor than both — and it's coming for that floor first.

The GCC that survives the AI decade isn't the one that's cheaper. It's the one that owns capability the parent company can't easily replace — judgment, domain depth, customer intimacy, the accumulated context that lives in no document. You can't offshore that, and you can't automate it. That's the whole game.

The question to ask before signing the lease

Before committing to Hyderabad or Visakhapatnam, the question isn't “how much will we save.” It's: what will this center own that nothing else can? If the honest answer is “labour cost,” you're building a 2006 cost center with a 2026 press release — and you'll wonder in five years why it never became strategic and why AI hollowed it out.

India's opportunity in this boom isn't to be the world's back office again, only faster. It's to be where global companies build capability they can't build anywhere else. The centers that understand that will still be here in 2035. The arbitrage plays will have moved on to wherever is cheap next — or been quietly automated.

If you're standing up or running a GCC: what does it own that can't be bought cheaper somewhere else?