In three years, the ‘cost centre’ framing will read like a mistake

Here’s the call. In 3 years, the decks being written today, the ones that justify a GCC on cost arbitrage, headcount at a discount, run-rate savings, will read like a category error, because the parent stopped setting up centres to save money, part of a broader pattern of GCCs decoupling from the parent. The mandate is shifting to owning capability the parent can’t easily build anywhere else: AI and data platforms, product engineering, the work that decides the roadmap rather than executing it.

The shift is underway now, unevenly, centre by centre. And the leaders being hired against the old framing, strong operators optimised to run a tight, cheap, predictable machine, are precisely the wrong profile for the machine the parent is about to ask for. You can be excellent at the last mandate and unfit for the next one.

When a parent stands up a centre, it has to book it somewhere, and ‘cost centre’ is the accounting default: a line to be minimised, measured on how little it consumes. In the arbitrage era that framing did real work. It let a board approve the centre on a number it already understood. The engineer’s objection is simple. You optimise for what you measure, so a centre told its purpose is to be cheap gets very good at being cheap and structurally worse at everything else. The measure shapes the machine.

The parent’s own books have already moved past the label. Zinnov and Nasscom now describe the India base of roughly 2,117 centres, generating about $98.4 billion, as mid-shift from cost to value (Zinnov–Nasscom, India GCC Landscape 2026). EY’s read of life-sciences centres shows the same thing operationally, with parents already running majorities of their finance, HR and supply-chain work out of India (EY India, 2025). The ledger still records a cost. The dependency it records has quietly become a capability the parent can no longer unwind.

The efficiency leader is measured on run-rate, span of control, delivery predictability. They are rewarded for making the known process cheaper and more reliable. The capability leader is measured on what the centre can now do that the parent couldn’t before: new products shipped, decisions owned, roadmap influenced. One protects a cost line. The other builds an asset. These are different people, with different instincts, hired through different searches.

The mistake centres are making right now is assuming the first leader can simply pivot into the second when the mandate changes. Some can. Most can’t, the whole shape of their judgement was formed under a different objective. It isn’t a lack of ability. A leader who has spent a decade defending a cost line reaches for cost answers under pressure, because that is where their instinct was trained. Ask them to bet a budget on a product that might fail, and the reflex fights the mandate. You don’t retrain that on a memo.

The centres that will look prescient in 3 years, the window our three-year outlook for GCC hiring maps out, are the ones hiring the capability leader now, before the parent formally hands over the new mandate, because the good ones are scarce and the search is slow. Waiting until the framing officially flips means competing for that leader at the exact moment everyone else realises they need one too. Foresight here is really just being early to an inevitability, and paying the ordinary price for the search instead of the emergency one.

There’s a sequencing point buried in this. A capability leader hired early builds the bench beneath them, sets the hiring bar for the next 2 layers, and earns the parent’s trust before the mandate is formally on the table. A centre that waits inherits all of that work at once, under scrutiny, with no runway. The savings deck feels safe today. In 3 years it’ll read like the moment the centre bet on the mandate that was leaving instead of the one arriving.

An engineer knows you can’t patch an architecture problem with effort. If you staffed the top layer to run a cheap, predictable machine and the mandate turns to owning product, the gap that opens is a design problem. The leader hits every target on the old scorecard and the centre still can’t ship the thing the parent now needs. Effort has nowhere to grip, because the shortfall sits in the shape of the role. That is a foundation the centre outgrew, and foundations get replaced rather than tuned. The cost of the replacement lands all at once.

So the prediction is finally about compounding. A capability leader hired ahead of the curve appreciates: every quarter the centre owns a little more of the roadmap and becomes harder for the parent to imagine doing without. An efficiency hire kept past the turn quietly caps how large the centre can ever become, because the ceiling was set the day the mandate was written as a savings target. The board will notice which of those it bought. My argument is only that it will notice late, and that the ones reading this now still have the cheaper option open to them.