India’s GCCs Have an AI Innovation Window and a Pilot Problem

India’s 1,800-plus Global Capability Centers were supposed to benefit from AI abundance, and they have, though not in the direction the industry predicted. The Averi 2026 AI content benchmarks report found that AI tools now let marketing teams publish 42 percent more content monthly at 42 percent lower production cost. Social media engagement across platforms fell to between 1.4 and 2.8 percent in 2025, per Hootsuite benchmark data, and 56 percent of marketers told HubSpot the internet now feels flooded with AI-generated material. The same tools driving that content glut are the ones threatening to automate the execution-layer work the GCC sector, which employs nearly two million professionals across India, was built to deliver.

Andrew Dimitriou, Global Chief Client & Growth Officer at DEPT, the global digital agency, flagged the underlying tension in a recent interview with Storyboard18: organizations treating AI primarily as a cost-cutter are trading a real efficiency gain for a larger strategic loss, while the window for something more consequential stays open.

When Content Becomes Cheap

The economics of the volume problem are now measurable. The Averi 2026 AI content benchmarks report found that content output rises 77 percent within six months of AI implementation, and that production cost falls by an average of 42 percent across formats. Companies that published 12 articles a month now publish 17, not because the extra five were planned, but because cheap production made marginal content the default.

What follows is predictable. In HubSpot’s 2026 State of Marketing survey of more than 1,500 marketers, 65 percent said consumers are getting better at identifying AI-generated content. Adobe’s 2026 AI and Digital Trends report, drawn from surveys of 3,000 executives and 4,000 customers, found that half of consumers give promotional content two to five seconds to capture attention before disengaging. Branded content expenditure is projected to reach $107 billion globally in 2026, per Statista data, meaning more money is chasing less attention simultaneously.

Dimitriou’s framing cuts through the production argument. “Anyone can generate content today,” he said. “The value is shifting from creating content to creating competitive advantage.” In his analysis, differentiation now comes from strategic judgement, cultural relevance and the capacity to turn thousands of AI-generated interactions into a coherent brand experience. The scarce resource is the decision about which signals are worth acting on and which are noise.

“If brands apply AI indiscriminately,” he said, “they risk looking, sounding and behaving exactly like everyone else.” Average social media engagement fell below 3 percent across platforms in 2025, per Hootsuite’s benchmark report, while branded content budgets kept growing.

India’s $64.6 Billion Foundation

The GCC sector’s current scale makes the strategic stakes large:

• 1,800+ GCCs operating in India as of mid-2026, employing nearly 2 million professionals
• $64.6 billion in annual revenue; NTT DATA projects the ecosystem reaching $110 billion by 2030
• 92% of GCC leaders affirm their centres contribute beyond cost arbitrage, per the EY GCC Pulse Survey 2025
• 52% of GCCs now share accountability for global decision-making

The ecosystem moved through three distinct phases to reach this point. From 2000 to 2015, GCCs optimised for cost, handling back-office support and process standardisation. From 2015 to 2020, they took on advanced analytics, technology delivery and specialised shared services. The current phase, underway since around 2020, involves end-to-end product ownership, global R&D, and GenAI (generative AI, the category of large-language-model tools now embedded in most enterprise workflows) running inside core business functions.

NTT DATA, which launched its GCC Innovation Acceleration Program in India in March 2026, targeting support for more than 50 companies over three years, identified India as a standout market specifically because of talent depth and technical capabilities. The National Policy for GCCs 2025-2030, set out by the Indian government, identifies four development pillars: talent pipeline development, digital and physical infrastructure, urban hubs as global growth engines, and an innovation-led ecosystem built through academia and startup collaboration.

The EY GCC Pulse Survey 2025, which drew responses from more than 65 GCC leaders across industries, found 92 percent affirming their centres now contribute beyond cost arbitrage. Whether the ecosystem converts those affirmations into production-scale AI is what the sector’s pilot numbers make genuinely uncertain.

The Pilot Graveyard

Between 70 and 90 percent of AI pilots across enterprises fail to reach production. The BCG-NASSCOM report from June 2025 confirmed that while top-performing GCCs have moved past experimentation, most centres are still locked in early-stage testing. The EY Pulse Survey counted 83 percent of GCCs engaging with GenAI adoption, and 58 percent actively building agentic AI capabilities, but most of that activity sits in pilot mode.

The Zinnov and Indiaspora GCC AI Opportunity analysis from March 2026 put a precise figure on the exposure: 55 percent of India’s GCC work sits in the bottom two tiers of the value hierarchy, Commodities and Procedures, directly exposed to AI displacement. More than half the work currently inside Global Capability Centers carries a realistic automation timeline.

The failure pattern is organisational, not technical. A YourStory GCC roundtable in March 2026, drawing leaders from logistics, manufacturing, financial services and analytics, identified a consistent dynamic: when engineering teams own the AI model but business leaders own the use case, pilots collapse the moment they encounter real-world conditions. Variable invoice formats, noisy customer data and fluctuating transaction volumes are normal production realities. A pilot built in a clean test environment doesn’t survive them without a business owner accountable for the outcome.

Three structural blockers keep repeating:

• Data architecture built for batch processing, incompatible with the real-time operations agentic AI requires
• AI adoption measured by the number of use-cases launched, not by the percentage reaching production
• Business outcome ownership separated from technical model ownership, so pilots stall before the workflow changes that would make models viable in practice

Less than 3 percent of India’s engineers are currently AI-ready for production-grade work, per BCG-NASSCOM analysis cited by Datacouch, a GCC analytics consultancy. GCCs that have closed the gap did it through structured internal upskilling rather than external hiring. NeoIntelli’s October 2025 research found that purpose-built AI talent programs inside GCCs can double the bench of production-ready engineers within 18 months.

How Agency Networks Are Rebuilding

AI-linked roles accounted for 44 percent of job postings across India’s media and communications GCCs in 2025, up from 21 percent the year before, per jobs platform foundit data cited in an EY report on India’s media GCC sector. Companies across Bengaluru, Gurugram, Hyderabad and Pune are competing for AI campaign optimisation specialists, marketing data scientists and generative AI-focused professionals.

Three years ago, most of these headcounts were categorised as support functions. Annual reports described them as cost centres, not capability investments. AI compressed the content production cycle from days to hours, which created immediate budget pressure to redeploy the time saved into new workflows, and the redeployment landed in AI-linked roles.

Vishal Srivastava, CEO of Omnicom Global Solutions India, described the shift plainly: “The conversation is no longer just about cost or efficiency. The new currency is value.” India-based teams at Omnicom are now building global platforms, workflows and AI-driven systems for clients, work that would have sat at headquarters two years earlier.

DEPT helped some clients reduce content production costs by as much as 75 percent. Those savings aren’t leaving the system. “They’re being reinvested into new channels, new experiences and new growth opportunities,” the DEPT executive told Storyboard18. “Six months ago, brands weren’t building shoppable experiences inside ChatGPT environments. Today, they are.” That pace of change inside the marketing stack is what makes the transition structural.

From Billing Hours to Buying Outcomes

Global agencies are migrating from time-and-materials pricing toward output-based and outcome-based contracts. In the same interview, the DEPT executive was direct: clients want measurable business results, not simply more hours billed. For India’s GCC leaders, who built large operations around headcount and billing rates, that is a genuine structural reorientation.

What makes the transition difficult is the organisational problem he identified: companies that bolt AI onto decades-old workflows without redesigning them don’t see the transformation the investment promises. New companies carry no legacy systems and haven’t built the entrenched incentives that established GCCs run on. Established centres have to reimagine how work gets done while running at scale.

The strongest AI strategies, in his analysis, won’t lock into a single foundation model provider. Companies that retain ownership of their data, workflows and proprietary capabilities while building intelligence layers across multiple models will be harder to replace. For India’s GCCs, which have historically been integrators rather than builders, this means developing a different kind of asset: proprietary data, curated at scale over years, that no outside model can replicate quickly.

India’s linguistic complexity sharpens the specific opportunity. The country operates across 22 official languages, giving GCCs that master AI-driven localisation a test bed no other market offers at equivalent scale. Among marketers using AI for personalisation, HubSpot’s 2026 State of Marketing found 91 percent reporting improved engagement. In a market this diverse, that localisation capability is simultaneously a domestic product and an exportable service.

The Export Market India Hasn’t Captured

Whether India captures the export moment or optimises itself into commoditisation depends on what the sector does next with the AI cost savings and talent depth it has accumulated. Dimitriou said:

The back-office model will continue to exist, but it will increasingly be automated. The bigger opportunity is for India to create and export entirely new AI-enabled services. That’s where long-term value creation lies.

Evidence from India’s AI startup sector moves in the same direction but with caveats. Zinnov’s 2026 analysis of more than 3,100 Indian AI startups found the ecosystem moving from experimentation to production-grade systems, and identified 2026-27 as the highest-leverage window for building durable AI companies from India. But 72 percent of those startups sit at the application layer, building on foundation models they don’t control, which caps the defensibility of what they create. A team that fine-tunes the same foundation model can replicate most of what an application-layer company ships, which means differentiation requires something the model itself can’t provide.

The GCC challenge mirrors this. Centres that layer AI onto existing execution without building proprietary data assets and redesigned workflows are still executing someone else’s strategy, just faster. Getting from service provider to innovation exporter requires owning what the AI is trained on and what it produces.

The 92 percent of GCC leaders who affirmed to EY that their centres contribute beyond cost arbitrage are describing an aspiration. The pilot failure rate describes what happens when that aspiration meets production. The distance between them is where the sector’s next decade is being decided.

The window Zinnov identifies as the highest-leverage period to build durable AI capability from India runs through 2027, and the sector’s pilot culture is still consuming it.