What Is a Global Capability Centre (GCC)? A Plain-English Guide

Somewhere in your industry, a competitor’s India centre is filing patents, running AI platforms and closing the global books — while their rivals still buy the same work from vendors, year after year, margin included. Understanding the Global Capability Centre model is no longer optional reading for operating executives. This guide explains it from first principles: what a GCC is, why the model won, what the evidence says about its economics, and the one variable that decides whether yours will work.

The idea in brief. A GCC is a company’s own centre in another country — overwhelmingly India — performing core work: engineering, data, finance, research, operations. Unlike outsourcing, the people, knowledge and intellectual property belong to you, and their value compounds. India hosts more than 1,700 such centres employing close to two million people, per industry body NASSCOM, and the model has shifted decisively from saving cost to acquiring capability. Setting one up is now a well-worn path; making it excellent is a talent problem, not a legal one.

The definition, stripped of jargon

A Global Capability Centre (you will also hear captive centre, global in-house centre or GIC) is a wholly owned unit of a company, located offshore, doing work that used to sit at headquarters. The essential word is owned. The engineers in a GCC carry your badge, commit to your codebase, sit inside your security perimeter and build careers inside your company. Nothing is bought from a vendor; everything produced — code, models, process improvements, institutional knowledge — accumulates on your balance sheet of capability.

That single design choice separates the GCC from its frequently confused cousin, outsourcing, and it drives every downstream difference in economics and behaviour. A vendor’s incentive is to sell you more hours; your own centre’s incentive is to make hours unnecessary. Over years, those two incentive systems produce very different institutions.

GCC versus outsourcing: an honest comparison

Neither model is universally superior; they solve different problems, and mature companies typically run both. The comparison, stated fairly:

The crossover economics deserve their own picture, because they explain the boardroom pattern — vendors win the first year, ownership wins the decade:

The shape of that chart is the whole strategic argument. A vendor relationship is a flat line of cost forever — the margin never goes away, and the learning walks out the door at contract end. A GCC front-loads investment (setup, leadership, slower initial ramp) and then bends the curve as teams season, attrition-adjusted productivity climbs and the vendor margin disappears. Deloitte’s long-running global shared-services surveys have tracked versions of this crossover for two decades; the industry’s revealed preference — captives growing faster than the services industry that birthed the market, per NASSCOM data — says companies believe it.

How a back office became a second headquarters

The model’s evolution explains its present shape, and the four eras are worth knowing because each one’s assumptions still circulate in boardrooms — usually one era out of date.

1. The pioneers (1985–2000). Texas Instruments opened a Bengaluru design centre in 1985 — the industry’s founding act — betting that Indian engineers could do frontier work remotely. Banks and airlines followed with data processing. The thesis was labour arbitrage, and it worked.
2. The scale era (2000–2010). Y2K remediation, the internet build-out and post-merger banking integration drove massive captive growth. Centres proved they could run global processes reliably at a fraction of home-market cost. This is the era that fixed “offshore = cheap back office” in a generation of executives’ minds.
3. The capability turn (2010–2020). A quiet inversion: as Western talent markets tightened, companies discovered their India centres could not merely support products but build them. Engineering headcount overtook operations headcount at leading centres; “cost centre” began sounding like an insult.
4. The ownership era (2020–today). Surveys by EY, Zinnov and NASSCOM now consistently report access to skills — not cost — as the primary driver of new centre announcements, with a large share of new GCCs opening with AI, data or product-ownership mandates on day one. Global role charters increasingly sit in India: the head of a platform, or of a finance tower, simply lives in Bengaluru.

The lesson to extract: every era’s laggards adopted the previous era’s model. Building a 2010-style cost centre in the ownership era is the most common — and most expensive — strategic error in this field.

Why India, specifically — the evidence

The concentration is extraordinary: India hosts more GCCs than the rest of the offshore world combined, and the gap is widening, not closing.

Five structural reasons, each independently verifiable:

• Scale of supply. A technology workforce above five million (NASSCOM’s industry statistics), refreshed by more than a million engineering graduates annually — the All India Survey on Higher Education tracks the pipeline — plus enormous cohorts in finance, science and design. Scale converts hiring from a lottery into a planning exercise.
• Forty years of compounding experience. Each generation of centres trained the managers who lead the next. This operating know-how — how to run a 2,000-person centre that headquarters trusts — exists at depth nowhere else. It is the moat rivals cannot buy.
• Durable economics. Fully loaded costs typically run 40–60% below equivalent Western roles (consistent across compensation studies; see our companion cost analysis for the full model). The gap narrows at the elite top and remains decisive at scale.
• Institutional ecosystem. Specialised real estate, GCC-literate law and accounting firms, state governments competing with incentives — Invest India maintains the national welcome mat — and policy vehicles (SEZs, GIFT City) built for exactly this.
• English and time-zone geometry. The working day overlaps usefully with Europe and brackets the US, enabling follow-the-sun operations for those who design for it deliberately.

What GCCs actually do now

The old picture — support tickets and maintenance — is a decade stale. A modern tour through India’s centres finds: AI and machine-learning teams building core models; product managers holding global roadmaps; chip designers taping out silicon (India hosts roughly a fifth of the world’s chip-design workforce — see our semiconductor analysis); pharmaceutical statisticians preparing regulatory submissions; global finance towers closing multi-entity books; and cybersecurity operations watching the world’s networks around the clock. Industry analyses estimate a substantial share of new centres open with an explicit AI mandate — the strongest single signal of where the model is heading.

The three ways to build one

1. Do-it-yourself. Incorporate, lease, hire. Maximum control and lowest long-run cost; slowest start and highest demand on your own management bandwidth. Right for committed, experienced builders.
2. Employer of Record (EOR). A partner legally employs your first cohort while you direct the work. Live in days; ideal for proving the thesis with 10–50 people. Costs cross over against your own entity somewhere between roughly 30 and 75 heads (the full decision framework, with the crossover chart, is in our EOR-vs-entity analysis).
3. Build-Operate-Transfer (BOT). A partner stands the centre up and runs it to stability, then transfers it to you. De-risks the first two years at the price of partner margin and some culture-transfer friction at handover.

These are stages more than rivals: the modern default is EOR to start fast, your own entity once conviction is proven, with BOT for companies that want speed and eventual ownership without building the machine themselves.

What the sceptics get right — and wrong

Three criticisms deserve engagement rather than dismissal:

• “Captives fail and get sold back to vendors.” This happened — notably in the late 2000s, when several banks divested captives. The post-mortem pattern: centres built as cost plays, with weak leadership and no mandate. The model matured precisely by metabolising those failures; divestments have become rare while new openings accelerate to record rates.
• “Attrition eats the savings.” It can — at badly run centres. The data (reconstructed fully in our attrition analysis) shows well-run GCCs posting voluntary attrition of 10–15%, comparable to tech hubs anywhere, with the 2021–22 spike now clearly visible as a cycle, not a climate. Attrition is a management outcome.
• “AI will hollow out the work.” The strongest current counter-evidence: companies are locating the AI build-out itself in India centres. Task-shaped work is genuinely exposed; capability-shaped centres are absorbing new work faster than old work shrinks. (Our closing article in this series takes the question head-on.)

The variable that actually decides success

Here is the pattern behind a decade of successes and failures, and it is not location, structure or incentives. Registering an entity is a solved problem — competent firms do it weekly. The unsolved problem at every centre is talent: hiring a leader that India’s best will follow; winning engineers in one of the world’s most competitive markets; converting them into a team headquarters trusts; and keeping them long enough for the economics to compound. The selection-science literature — start with Schmidt & Hunter’s landmark meta-analysis — says ability can be measured; the operating record says centres that measure, win. Companies that staff the GCC project with lawyers and real-estate specialists build offices. Companies that staff it with talent strategists build capabilities. The office is the cheap part.

A 90-day agenda for the GCC-curious executive

1. Weeks 1–4: Inventory the work. Which functions are continuing, core and remote-executable? Draft the three-year mandate you would want — not the tasks you would tolerate offshoring.
2. Weeks 5–8: Test the talent thesis. For your specific role mix, map supply, compensation and competition across two or three candidate cities. Visit. Meet potential centre-head profiles before you commit — their reaction to your mandate is market intelligence.
3. Weeks 9–12: Choose the build model against your speed and control needs; pressure-test the business case with realistic attrition, ramp and fully loaded assumptions; and lock the first quarter’s hiring sequence — leader first, always.

A worked example: the 150-person centre, honestly modelled

Abstractions persuade nobody, so consider a composite pattern — assembled from real engagements, identifying details removed — of a mid-size European industrial-software firm planning a 150-person centre. Their first business case, built at headquarters, compared average salaries and promised 55% savings from month one. The rebuilt case, using the methods described across this series, looked different and performed better:

• Year one ran at a modest net saving, not 55% — leadership search, EOR fees, employer-brand investment and sub-scale overheads absorbed most of the arbitrage. The board, warned in advance, did not panic. Boards that expect month-one miracles usually cancel exactly when the curve is about to turn.
• Year two delivered roughly 30% savings as the entity went live, the founding team’s referral network cut agency spend by half, and the first product module shipped from India — which mattered more than the savings, because it unlocked the next mandate.
• Year three reached the promised range — approximately 45–50% below home-market cost per productive employee — while the centre owned two product lines outright. Attrition ran at 11% against a metro market averaging several points higher, purchased by exactly the “soft” investments the original case had deleted as luxuries.

The pattern generalises: the GCC economic curve is a J, shallow in year one and compounding thereafter, and the discipline that pays is honesty about the J’s left side. Centres that promise the right side immediately borrow credibility they must later repay with interest.

Key terms, decoded

The field’s vocabulary confuses newcomers because it grew in layers, era by era. A working decoder:

• GCC / GIC / captive: three names, one thing — your wholly owned offshore centre. “Captive” is the banking-era term, “GIC” (global in-house centre) the consulting-era refinement, “GCC” the current standard. Age of the speaker’s vocabulary usually reveals the era of their assumptions.
• BOT (Build-Operate-Transfer): a partner constructs and runs your centre, then hands over the entity, team and operations at a pre-agreed trigger. The “T” is the clause to negotiate hardest — transfer pricing, retention guarantees, and what happens to the partner’s management layer.
• EOR (Employer of Record): a partner legally employs your people while you direct their work — the speed tool of the first year (fully unpacked in our EOR-vs-entity analysis).
• Value chain / mandate: the industry’s shorthand for what a centre is trusted to own. “Moving up the value chain” means graduating from executing specifications to owning outcomes — and “mandate” is the unit in which that trust is granted.
• SEZ / STPI / GIFT City: India’s location-incentive regimes — export zones, the software-park scheme, and the financial-services enclave respectively (decoded fully in our incentives guide). Useful tiebreakers; dangerous drivers.
• Follow-the-sun: operations designed so work passes between time zones daily — support and security operations’ favourite argument for India’s geography.
• Regretted attrition: departures you actively wanted to prevent — the number that matters, as against headline attrition, which mixes signal with the system working (our attrition analysis separates them).

Five questions boards actually ask

“How fast can we realistically be operational?” Legally hiring: days, via EOR. First pod productive: one to two quarters. Full 150-person centre at steady state: 18–24 months. Any promise materially faster than this is compressing the leadership search or the quality bar — the two places compression is fatal.

“What is the minimum viable size?” Below roughly 25–30 people, overheads and management attention rarely justify a standalone centre; EOR-based teams handle the experiment phase better. The natural first milestone is a 50–100-person centre with one clear mandate.

“Should we worry about IP and data security?” Structurally, a GCC is the strong option: your employees, your systems, your perimeter — which is precisely why regulated industries (banking foremost) run some of India’s largest centres. The genuine work is ordinary enterprise security discipline, not exotic country risk.

“What kills these projects?” In observed order: leadership mis-hires (the dominant cause — see our centre-head analysis), mandate starvation (centres fed only leftover work lose their best people to competitors offering ownership), and business cases built on fantasy attrition and ramp assumptions that discredit the project when reality arrives.

“Build ourselves, or use partners?” The honest answer is staged: almost everyone now starts with partners for speed (EOR, talent, workspace) and internalises functions as scale justifies. The strategic error is not partner choice; it is outsourcing the talent strategy itself — the one capability that must be owned from day one.

Methodology & data notes

Figures in this article are compiled from public industry sources — principally NASSCOM and Zinnov landscape reporting, EY and Deloitte survey series, and government statistics — and are presented as indicative ranges rather than point estimates; where charts say “illustrative,” the curve shape, not the specific values, is the claim. India GCC statistics vary by definition (what counts as a centre, which fiscal year); we cite the commonly used mid-points and encourage readers to consult the primary sources below for current-year figures.

References & further reading

• NASSCOM — GCC landscape reports and India technology-industry statistics
• Zinnov — Zinnov–NASSCOM India GCC Landscape studies
• EY India — GCC Pulse survey series
• Deloitte — Global Shared Services and Outsourcing surveys
• McKinsey Global Institute — global services trade and workforce research
• AISHE — All India Survey on Higher Education (graduate pipeline)
• Schmidt & Hunter (1998) — The Validity and Utility of Selection Methods in Personnel Psychology, Psychological Bulletin
• Invest India — national investment promotion agency

HexGn is the talent partner for companies building GCCs in India — entity, compliance and EOR handled, but always led by the part that decides the outcome: the leaders and teams inside the centre.