GCC vs ODC in India: Ownership, Cost and Tax Filings

GCC vs ODC in India: which one should you actually build? It looks like a cost question. It is an ownership question, and ownership decides what you file with the Indian tax authorities every year.

Choose a captive GCC and your India entity becomes an associated enterprise of your parent. Every invoice between them has to be priced at arm's length, and an annual accountant's report follows. Choose a vendor-run ODC and none of that lands on you.

If the decision to build in India is already made and only the structure is open, six things settle it: cost, speed, who legally employs the team, who owns attrition, what you file, and what it costs to walk away.

India's GCC market now holds 2,117 GCCs across 3,728 units, employing about 2.36 million professionals in FY2026 (Wisemonk India Investment Intelligence 2026). This is a well-worn path with two very different bills attached.

What is the difference between a GCC and an ODC in India?

An ODC is a delivery arrangement. A GCC is a business unit you own. An ODC usually covers engineering and executes your roadmap. A GCC spans engineering, finance, analytics, support and HR, owns capability and outcomes, and is your own registered Indian entity by definition.

That sounds tidy. In practice you have probably read three definitions that contradict each other, because vendors, analysts and in-house teams all use "ODC" for different things while a global capability centre has stayed relatively fixed.

The two models separate on eight things:

One naming note before the arithmetic: "captive" and "GCC" describe the same structure, and captive centres in India is simply the older word for a centre you own outright.

Why does "ODC" mean two different things?

Because the term describes an arrangement, not an owner. One usage means a named team sitting inside a vendor's Indian entity, billed to you monthly. The other means a development centre you own, staffed only by your people.

Both are called an offshore development centre, and both are correct usages. That is why the definitions clash.

The thing that actually varies is whose entity the team sits in: the vendor's, an EOR's, or yours.

Ask that question and every downstream answer falls out of it. Who pays statutory contributions, who carries the employment risk, whose margin sits on the rate, and what happens when you stop.

If you want the full build mechanics rather than the comparison, it is worth reading how an offshore development center in India is built and run before you price either option.

Ownership settled, the next question is the one that actually gets asked in the room: does the ODC really cost less?

Will an ODC actually be cheaper than a GCC in India?

Over one to two years, almost always yes: a vendor ODC carries no setup cost and no entity overhead. Past that the arithmetic inverts, because vendor margin is a recurring 25% to 45% on every engineer while a GCC's $500,000 to $3 million setup is spent once.

The break point is headcount and time horizon, not the rate card.

Here is the fear worth naming out loud, because finance leads raise it every time: that the savings turn out to be fake.

They are not. Offshore engineering in India runs 40% to 60% below US cost in both models, so the saving is real either way. The argument is about who keeps it.

The row that decides most of it is the second one. A vendor quotes you a blended rate, keeps 25% to 45%, and pays the engineer the remainder. You are paying a market rate for the seat and the engineer is not receiving one.

That gap is not a moral point, it is a retention point, and it shows up in your velocity a year later. For the underlying numbers, the full cost of setting up a GCC in India breaks the setup figure into its parts.

For scale on the run rate: fully loaded monthly cost per engineer in India runs from roughly $1,000 for QA to around $7,000 for an engineering lead (Wisemonk, as of July 2026), and those are the numbers a vendor rate is marked up from.

Run both models out far enough and they cross. If you want the modeled version rather than the principle, the five-year cost comparison between a GCC and outsourcing does the arithmetic at a fixed headcount.

Before you model anything, get the per-head number right with an employee cost calculator. Then the other half of the trade-off matters: how long until anyone is actually writing code?

How fast can each model actually be running in India?

An EOR-run ODC can have named engineers working in 2 to 6 weeks and a vendor-run ODC in 4 to 8. A captive GCC takes 3 to 6 months with a partner running registration in parallel, or 9 to 12 months built alone.

Company registration and operational readiness are two different milestones. You have almost certainly seen "12 weeks" and "18 months" quoted for the same thing, and both can be honest. They are measuring different finish lines.

The four realistic timelines to working engineers:

• EOR-run ODC: 2 to 6 weeks, because there is no entity to register at all.
• Vendor-run ODC: 4 to 8 weeks, most of it recruiting and contracting.
• Captive GCC with a partner: 3 to 6 months, with incorporation, registrations and hiring running in parallel.
• Captive GCC built alone: 9 to 12 months, because the registrations run in sequence and nobody is doing them full time.

Company registration is a legal event. Operational readiness is a different thing: registrations, bank accounts, payroll, statutory enrolments, an office or a remote setup, and people with laptops.

The first can finish months before the second, and conflating them is why published timelines disagree so wildly.

Build-operate-transfer sits somewhere between the two, and the numbers quoted for it are especially slippery. Three to six months usually describes the build phase only, not the operate and transfer phases that follow. The steps in a GCC build make the sequencing clear.

In the builds we have supported, the slippage almost never comes from incorporation. It comes from registrations and banking nobody scheduled, which is the pattern in the stall points that catch India capability centre builds.

Timelines settled, the question underneath both of them is who actually employs these people.

Who is the legal employer of your India team in each model?

In a captive GCC, your Indian subsidiary is the legal employer and every statutory obligation is yours. In a vendor-run ODC the vendor's entity employs the team. In an EOR-run ODC the EOR does. India recognizes one legal employer per worker, so this is never shared.

Most buyers treat this as a paperwork detail. It is the point the whole decision turns on, because the legal employer carries every item on the following two lists.

Whoever is the legal employer owns these statutory costs and filings:

• Provident Fund: retirement savings, similar to a 401(k), with matched employer and employee contributions.
• ESI: state-run health cover for lower-paid staff.
• Gratuity: a lump sum payable on exit after a qualifying service period.
• Professional tax: a levy set at state level, so the rate and the due date vary by state.

If you want the mechanics rather than the list, the PF, ESI and gratuity obligations are where most first-time India employers get caught.

The same choice decides who carries the regulatory layer sitting above payroll:

• Labour law: India's four Labour Codes came into force on 21 November 2025 and rationalize 29 labor laws into four, with central rules still being finalized as of September 2026.
• Data protection: the DPDP Act rules were notified on 13 November 2025 with an 18-month phased compliance timeline, and consent managers must be Indian companies.
• Foreign exchange: salary is paid in Indian rupees through a compliant route. A US parent cannot simply wire dollars into an engineer's Indian bank account and call it payroll, and whoever legally employs the team is the party that has to get this right.
• Permanent establishment: PE exposure turns on the nature of the activity, not merely on holding an entity, and a vendor ODC run too tightly is the classic trigger.

One more consequence follows from the same answer. Whoever employs the engineer is the counterparty on the employment contract that assigns the code, which is why protecting IP when hiring in India starts with the employment structure rather than with the NDA.

Who ends up owning the attrition?

Here is the finding buyers find hardest to hear: the attrition a company blames on India is often a function of the model, not the country.

A margin taken off the top lands somewhere, and it lands on the engineer's pay. Underpaid engineers leave, and the team you spent four months building turns over in fourteen months.

The captive numbers back it up. India GCC attrition fell from 13% in 2023 to 11% in 2024 to 9% in 2025, which is well below what outsourced engineering benches run at.

That is not an argument for a GCC on its own. It is an argument for knowing which side of the margin your engineer's salary sits on before you sign.

Everything to this point is a running cost. The next one is a filing, and it only exists on one side.

What tax filings does each model create in India?

A captive GCC is an associated enterprise of its parent, so every cross-border invoice it raises is an international transaction that must be priced at arm's length, and it owes an annual accountant's report. A vendor-run ODC is a third-party service purchase.

No associated enterprise, no international transaction, no report. This is the part that rarely makes it into the build plan, and it is the point where an org-chart decision quietly becomes a filing calendar.

A captive is not an independent counterparty. Indian transfer pricing treats your India subsidiary and your parent company as associated enterprises.

So whatever your subsidiary charges for engineering work has to stand up as an arm's length price under India's transfer-pricing rules (Income-tax Act 2025, as of September 2026).

The report that proves it changed on 1 April 2026. It is now Form No. 48, furnished under section 172 read with section 173 of the Income-tax Act 2025, filed online only and due one month before the income tax return under section 263(1).

Form No. 48 replaces Form 3CEB, the report filed under section 92E of the old Income-tax Act 1961. Behind it sits a benchmarking study justifying the margin you charged.

What each side actually files in India:

• Captive GCC: a corporate tax return, an annual transfer-pricing accountant's report on Form No. 48, and the benchmarking documentation supporting the margin.
• Vendor-run ODC: your vendor invoices, booked as a third-party service purchase, with withholding handled at source under your own jurisdiction's rules.

What is the safe harbour election worth?

Enough that it changes the shape of the argument. A safe harbour election lets a captive declare a margin the tax authority accepts without examining it, which converts an open-ended dispute risk into a fixed number.

Under the Budget 2026-27 measure, software development, IT-enabled services, KPO and contract R&D are clubbed into a single Information Technology Services category at a common 15.5% margin.

• Eligibility: rises from about $31 million to about $208 million of international transactions (Rs 300 crore to Rs 2,000 crore), as of September 2026.
• Approval: automated, with no tax-officer examination.
• Term: the election runs for five years. Unilateral advance pricing agreements carry a two-year target.

The election is filed on Form No. 49, under section 167 of the Income-tax Act 2025 read with rules 86 to 102 of the Income-tax Rules, 2026, merging the former Forms 3CEFA, 3CEFB and 3CEFC into one.

Worth separating from this: the safe harbour covers your pricing, not your presence. Permanent establishment risk in India is a different exposure, and it can bite a company with no Indian entity at all.

So the captive creates a filing obligation the vendor ODC does not, and India has just made that obligation about as inexpensive to discharge as it has ever been.

If you want to know where you stand before choosing, a permanent establishment risk quiz takes a few minutes. Filings are annual, though. The next cost is one-off and much larger.

What does it cost to walk away from each model?

Exiting a vendor ODC is a contract notice period. Winding down an Indian entity can take one to two years of clearing dues and approvals, and it touches company law, FEMA, labor, tax, transfer pricing and GST at the same time.

Only one of the two is reversible in a quarter. Nobody builds a team planning to unwind it, which is exactly why this cost goes unpriced. Set the two side by side:

• Vendor-run or EOR-run ODC: serve the notice period in the contract, transition the work, and you are done. Weeks, not years.
• Captive GCC: settle employee dues, close statutory registrations, obtain clearances across multiple authorities, then deregister the company. One to two years is a realistic window.

Treat that as a decision input, not a warning. If the roadmap is genuinely multi-year, the exit cost is close to irrelevant and the GCC wins on almost everything else. If the roadmap is not, the exit cost is the entire argument.

There is also a middle path that is not a full wind-down: closing an India entity without losing the team moves the same people onto an EOR and keeps the capability. Which leaves the practical question.

When should you choose an ODC, and when should you choose a GCC?

Choose an ODC when the roadmap is defined, the team is engineering-only, and you want people working inside two months. Choose a GCC when you intend to own the capability for several years, want functions beyond engineering, and have the headcount trajectory to absorb the setup cost and the filings.

The honest test is not company size. It is time horizon and what you want to own at the end of it. An ODC fits when:

• The work is engineering and the roadmap is already defined.
• You need named people working inside two months, not two quarters.
• The team is under roughly 25 to 30 people, which is where a per-head fee model still makes sense.
• You want the option to stop on a notice period rather than a wind-down.
• Nobody on your side wants to own an India filing calendar this year.

A GCC fits when:

• The India team is a multi-year capability, not a project.
• You want functions past engineering: finance, analytics, support, HR.
• Headcount is heading past 25 to 30 and the per-head fee is starting to look expensive.
• You want a P&L, a site leader and direct control of hiring and pay bands, and tier-2 cities run 25% to 30% below tier-1 on cost.
• An annual transfer-pricing report and a benchmarking study are acceptable overhead.

These are not permanent choices, and most companies pass through more than one. If that is where you are, how the India operating models sequence into each other sets out the order that tends to work.

To pressure-test the headcount trigger against your own numbers, run it through an EOR vs entity calculator. That leaves one hybrid worth taking seriously.

Can an EOR-run ODC give you GCC-style control without an entity?

Yes, for the control most teams actually mean. An EOR hires the engineers on your behalf under Indian employment contracts, you manage the work, the roadmap and the people decisions, and there is no Indian entity to register or to wind down later.

What you do not get is a P&L, a site leader, or functions beyond the team you hired. When buyers say they want GCC control at ODC speed, this is usually the thing they are describing. It is worth being precise about the trade:

• What you keep: direct selection of every engineer, your own roadmap and priorities, your tooling and process, and pay bands you set rather than a vendor's blended rate.
• What you do not get: a legal entity of your own, an India P&L, a site leader with a local mandate, or scope past the team you hired.
• What you avoid: setup cost, the transfer-pricing report, and a one to two year exit.

The EOR route runs out at the point where India stops being a team and starts being a business unit. Once you want finance, analytics and support under one India leader with a budget, the entity becomes the cheaper answer.

For the direct version of that trade-off, an EOR compared with a GCC in India sets the two against each other properly.

How can Wisemonk help you build an India engineering team?

Wisemonk is an India-native Employer of Record (EOR) that helps global companies hire, pay, and manage talent in India without setting up a local entity.

For an offshore development team, that means named engineers working your roadmap within weeks, on compliant Indian employment contracts, without registering a company in India first.

If you later decide the team should become a capability centre you own, the entity work starts from a team that already exists rather than from a blank page.

We support 300+ global clients and more than 2,000 employees across India, process $20M+ in annual payroll, and hold a 4.8/5 rating on G2. Pricing starts from $99 per employee per month as of September 2026.

Here is how we help:

• GCC and capability centre setup: stand up and staff an India capability centre, from operating model to hiring to run-state, priced on a custom quote.
• Entity setup: incorporating the Indian company and obtaining the tax and employer registrations, priced on a custom quote.
• PEO (HR services): for companies that already hold an Indian entity, we run payroll, PF, ESIC, professional tax and TDS filings, benefits and onboarding under your own registrations, from $49 per employee per month as of September 2026, and we can source and ship laptops to your engineers too.
• Managed payroll: the India pay run and its filings executed while you keep your own entity and HR team, priced on a custom quote.
• Background verification: 19 check types across identity, employment, education, court records, police, credit and moonlighting, with a basic check back in about five minutes and a full report in 7 to 10 days.

From our experience helping companies build engineering teams in India, the ones that scale cleanly are the ones that settle who legally employs the team in month one, rather than discovering it in a tax assessment in year three.