Shift Happens: Rewiring India’s Global Capability Centres (GCCs) For The AI Era – Analysis

February 13, 2026 
Observer Research Foundation
By Arya Roy Bardhan

Global Capability Centres (GCCs), also referred to as captive or global in-house centres, are wholly owned (or tightly controlled) offshore hubs set up by multinationals to run capabilities for the parent firm. These tasks span shared services (finance, HR, procurement), IT and operations, and even product engineering, analytics, cybersecurity, and R&D.

India has become a global leader in this sector as the role is shifting from back-office cost arbitrage to strategic extensions of headquarters. Today, these centres manage end-to-end digital products and platforms, run centres of excellence in AI/data/cloud, and increasingly drive transformation mandates across business functions. India’s convening of the AI Impact Summit from 16-20 February 2026 offers a timely chance to discuss the industry’s future and mobilise resources amid widespread job loss fears.

Figure 1: Maturity of GCCs Across the Top Nine Countries (by number of hubs)
Source: BCG

India is estimated to have over 1,700 GCCs with around 1.9 million direct employees and US$64.6 billion in annual revenue in 2024. Since India’s national accounts do not separately report the GCC sector, the ideal way to express GDP footprint is as a proxy. Thus, the sector accounts for around 1.6 percent of India’s nominal GDP and employs about 0.3 percent of India’s workforce. GCCs now employ roughly 35 percent of India’s tech-services workforce. Over the past two decades, their evolution has been steep: from early-2000s captive information technology (IT), business process outsourcing (BPO), and shared services units; to 2010s multifunction hubs incorporating engineering, research and development (ER&D), and analytics; and, in the 2020s, to high-value product, R&D, and innovation mandates. Another marker of this maturation is their expansion from roughly 1,300 centres, 1.3 million employees, and US$33.8 billion in revenue in 2020to a far greater scale today.



Figure 2: Snapshot of the GCC Ecosystem in India
Source: Zinnov-NASSCOM Report
Automation: Neither a Boon Nor a Bane

There is a core concern with AI. Many GCC functions, including customer support, finance operations, quality assurance (QA) testing, documentation, analytics, and chunks of software delivery, are task-heavy knowledge work. Modern AI, especially generative AI (GenAI), is adept at automating or speeding up the “rules-and-text” parts of those tasks. This is a task reallocation problem, where technologies tend to substitute for workers in codifiable or routine tasks while complementing non-routine problem-solving and complex communication tasks. In the automation-and-new-tasks framework, the net employment effect is theoretically ambiguous — displacement occurs when AI takes over tasks previously performed by labour, while reinstatement occurs when new tasks or activities expand labour demand elsewhere (often in higher-skilled, complementary roles). For GCCs, that maps to concerns about entry-level pipeline roles and full-time-equivalent (FTE)-based delivery models being compressed, even as demand rises for AI product engineering, model risk/governance, security, and domain-led transformation roles.

Empirically, the best early evidence points to augmentation or productivity uplift rather than one-for-one replacement, though with distributional twists. A large-scale field study of a GenAI assistant in customer support found 15 percent higher productivity, with the biggest gains for less-experienced agents, suggesting that AI can lift the floor and change skill gradients. Controlled experiments have also shown large speed or quality gains in professional writing tasks. For software work, randomised controlled trials at three companies report around a 26 percent increase in task completion. Here again, less experienced developers had higher adoption rates and productivity gains. At a macro level, the International Labour Organization argues that GenAI’s dominant impact is likely to be job transformation and augmentation, rather than wholesale job elimination. However, exposure is meaningful in clerical and some professional occupations, which are highly relevant to back-office-heavy segments of the GCC stack.

A seminal early study by the OpenAI team on the effect of Large Language Models (LLMs) on the US labour market found that around 80 percent of workers could have more than 10 percent of their tasks affected, while 19 percent could have more than half of their tasks affected. Adding software and tooling on top of the LLMs yielded further incremental gains across the workforce. In the services sector, LLM access could make a meaningful share of tasks faster, at the same quality. This framework flags tasks with heavy reading and writing, information synthesis, coding, classification, and standardised decision rules as the most exposed, which is exactly the task profile that dominates most modern service workflows. Applying that lens to Indian GCCs — which span technology, engineering, and operations, including shared services and globally owned roles — the likely outcome is partial replacement and broad augmentation rather than blanket job loss.

Figure 3: Tasks with Medium and High GPT-exposure, by Occupational Category
Source: ILO (Extracted using AI)

Under the replacement effects (task displacement), routine throughput layers might get automated or compressed. For instance, L1 support scripts, ticket triage, SOP drafting, basic reconciliations or payroll checks, compliance checklists, test-case generation, boilerplate code, and documentation might require lower labour inputs. GCCs may need fewer junior FTE hours per unit of output, flattening the delivery pyramid. Within augmentation effects, higher-value work will expand because AI raises individual productivity and widens the feasible scope. Developers can ship more, analysts can iterate faster, shared-services teams can move from processing to exception handling and control, and new roles can grow around AI productisation (evaluation, governance, security, domain translation). This is consistent with evidence-based warnings that GenAI’s dominant impact is often job transformation, even when many clerical-style tasks are exposed.

No Progress without Policy

The task-based models imply that the government’s best lever is to tilt AI adoption toward augmentation — raising worker productivity, rather than pure substitution. AI can displace labour by automating existing tasks, but employment can be preserved or expanded if policy helps generate new tasks and raises labour demand through innovation and output expansion. For Indian GCCs, this entails:large-scale, employer-linked skilling allowing workers to operate AI systems and move up the task ladder.
cheap, reliable access to compute, datasets, and model ecosystems enabling firms to build AI-enabled workflows that complement humans. This is also the intent of the IndiaAI Mission, which is based on the pillars of compute marketplace, innovation centre, datasets, skilling, etc.
credible governance (privacy, accountability, auditability) to speed adoption in regulated enterprise workflows where GCCs operate. India’s own Responsible AI work highlights these trust-building principles. The economic literature also stresses that automation has historically not eliminated work because it complements labour and expands demand — policy should therefore focus on human capital and enabling complements, not just on subsidising capital deepening.

To protect and increase employment within the sector while giving GCCs a transformational role in services-led growth, the goal is to push GCCs from low-cost hubs into innovation, IP, and product platforms. This can create complementary roles, even as routine layers compress, aligning with empirical evidence. Thus, pairing augmentation policies with active labour market supports (placement, certification, transition assistance) will be crucial. A services strategy matters because services are already the economy’s backbone (contributing 55 percent of GVA in FY25). Here, GCC-led productivity and export upgrading can be a direct growth channel if India also improves the enablers — digital trade rules and data-policy clarity, R&D incentives, deep-tech clusters around campuses, and structured work-integrated learning. While it is already being addressed through Skilling for AI Readiness (SOAR), the National Education Policy (NEP) should also explicitly include contemporary subjects like AI in curricula.
Financial software


Platforms for Policy, Partnership, and Public Capacity

AI is unquestionably a disruption for GCCs because it targets the codifiable layers that have historically anchored scaled services delivery. However, economic logic is equally clear that the employment outcome is not pre-determined. In task-based models, the net effect depends on whether policy and industry steer innovation toward complements and new tasks that expand labour demand rather than toward automation of existing tasks, which displace labour. If India grounds its response in granular, task-level data — what GCC roles actually do, where AI is deployed, how productivity gains translate into higher-value mandates and exports — then AI can become a transformation engine.

Upgrading GCCs into product, R&D, and enterprise transformation hubs can lift services productivity and exports in an already services-heavy economy. In that context, the upcoming India–AI Impact Summit in New Delhi can be more than a showcase. Rather than being an abstract forum, it has the potential to convene governments, corporations, researchers, and investors around quantifiable outcomes. It should be appropriately utilised to mobilise the resources needed to transform AI adoption in GCCs from headcount reductions into service-led growth.

About the author: Arya Roy Bardhan is a Junior Fellow with the Centre for New Economic Diplomacy at the Observer Research Foundation.

Source: This article was published by the Observer Research Foundation.


Observer Research Foundation

ORF was established on 5 September 1990 as a private, not for profit, ’think tank’ to influence public policy formulation. The Foundation brought together, for the first time, leading Indian economists and policymakers to present An Agenda for Economic Reforms in India. The idea was to help develop a consensus in favour of economic reforms.