Abhishek Rungta: The GCC Gold Rush Is Real — But India’s Global Capability Centres Are Getting the AI Layer Dangerously Wrong

I have been building technology businesses in India since 1997. I have watched the outsourcing wave reshape this country’s economy, seen the cloud migration era create an entirely new class of enterprise IT, and witnessed the GCC boom transform India from a delivery destination into a strategic nerve centre for the world’s largest multinationals. Each transition brought genuine opportunity. Each also brought a version of the same mistake, the rush to capture the form of the new thing without doing the hard work of understanding its substance.

We are making that mistake again. And this time, the cost of getting it wrong is significantly higher.

India now hosts more than 1,800 GCCs employing close to two million professionals. These are wholly-owned offshore units where multinationals run strategic business functions globally, finance, technology, customer operations, and increasingly, research and innovation. The numbers are impressive. The strategic ambition is real. 92% of GCC leaders say their centres have evolved beyond simply cutting costs. But when you look at what most GCCs are actually doing with AI, as opposed to what they are saying about it, a troubling gap appears.

According to the EY GCC Pulse Survey 2025, 83% of GCCs are already engaging with GenAI adoption, and 58% are actively developing agentic capabilities. These are strong numbers on the surface. The problem is not the intention. The problem is the architecture underneath the intention.

The AI Layer Is Being Added. The Foundation Was Never Built.

Here is what I see consistently across enterprise engagements: GCCs are layering AI onto processes that were never designed to be AI-ready. They are deploying GenAI copilots onto workflows with fragmented data, inconsistent tagging, and no governance model. They are announcing agentic AI strategies while running on batch-processing infrastructure that cannot support real-time agent operations. They are measuring AI success by the number of use cases launched, not the percentage that made it to production.

GenAI adoption across GCCs is concentrated in customer service at 65%, finance at 53%, and operations at 49%. These are the right functions to target. But targeting the right function with the wrong infrastructure does not produce transformation, it produces an expensive pilot that never scales.

Business intelligence adoption has increased to 86% of GCCs, and data strategy formalisation has risen to 67%. Progress, certainly. But 67% also means a third of GCCs are investing in AI agents while operating without a coherent data strategy. You cannot build an autonomous system on a data layer you have not yet organised.

The Displacement Reality Nobody Wants to Name

There is a harder truth beneath the capability gap. Zinnov and Indiaspora’s March 2026 analysis found that 55% of India’s GCC work sits in the bottom two tiers of the value hierarchy, Commodities and Procedures, directly exposed to AI displacement. Google Cloud Next 2026 made this explicit: the shift to agentic AI removes humans as the unit of production. The parent enterprise no longer needs offshore headcount to scale its work. It needs a fleet of agents, a governance layer, and a small number of human supervisors.

This is not a theoretical risk on a five-year horizon. It is a budget decision that is already being made in boardrooms in the US, UK, and Europe. GCCs that cannot demonstrate AI-native value creation, that remain, in practice, sophisticated labour arbitrage, will face rationalisation. The window to move up the value chain is not years. It is 12 to 18 months.

What Getting It Right Actually Requires

I am not arguing against the GCC model. I am arguing for a more rigorous approach to the AI layer within it. Three things must happen before any GCC announces an agentic AI strategy.

First, data architecture must be solved, not partially, not directionally, but completely. Agents are only as intelligent as the data environment they operate in. Second, process re-architecture must precede automation. Automating a broken process at machine speed produces broken outcomes at machine speed. Third, governance must be built in from the start, not retrofitted after the first compliance incident. Reskilling initiatives have grown to 71% of GCCs in 2025, which is encouraging. But reskilling people is the last step, not the first.

At INT, we have spent 28 years building enterprise technology for some of the most complex operating environments in banking, insurance, and life sciences. The lesson that has never changed: the difference between a transformation that delivers and one that disappoints is almost always the quality of the foundation, not the sophistication of the tool placed on top of it.

India’s GCCs have the talent. They have the mandate. What they need now is the discipline to build the AI layer correctly, even when speed is the loudest pressure in the room.

That discipline is what separates the GCCs that will own the next decade from the ones that will spend it explaining what went wrong.

Abhishek Rungta is the Founder & CEO of INT (Indus Net Technologies), a full-stack digital transformation company headquartered in Kolkata, India. With 28 years of enterprise technology delivery across BFSI, Life Sciences, and Retail. INT. helps enterprises build AI-ready infrastructure at scale. Learn more at intglobal.com.