Build-Operate-Transfer vs Direct GCC vs GCC-as-a-Service: which model actually de-risks your first 24 months?

TL;DR

Direct GCC ownership wins on long term cost and IP control but loses on speed to first FTE. BOT wins on de-risking the first 18 months but routinely overpays on the transfer event and loses 20 to 30% of the team in the six months after transfer. GCC-as-a-Service wins on optionality and is the dominant model when you are under 100 FTEs or your three year plan is not yet locked. The framework: choose by your conviction on FTE count and tenure, not by what feels safest in the boardroom.

Every GCC conversation eventually arrives at the same three letter acronyms: DIY, BOT, GaaS. Every setup vendor has a favourite. The favourite is almost always the model that maximises that vendor's revenue, not yours. After advising on or executing 130 plus India GCC builds across these three models since 2019, the pattern is clear: each model wins on a specific axis, each fails on a specific axis, and the cost of choosing the wrong model is between USD 4 and USD 12 million over five years for a 200 FTE centre.

Model 1: Direct GCC (Do It Yourself)

You incorporate an Indian entity (typically a wholly owned subsidiary), lease your own office, hire your own employees, build your own IT stack, and own the GCC outright from day zero. The setup partner you hire is on a fixed fee mandate, not equity or revenue share.

When Direct wins

• You are sure you want 200 plus FTEs within 24 months
• Your work involves IP that cannot live in a third party entity (frontier AI research, semiconductor IP, regulated financial models)
• Your CFO can absorb USD 1.2 to 1.6 million in one time entity, capex and consulting cost
• You have an Indian leader (or are willing to hire one in month one) who can run the centre

When Direct fails

• Speed: typical time from board approval to first FTE is 6 to 9 months
• Leadership risk: if the GCC Head is wrong, you have built an empty entity around the wrong person
• Hiring risk: you carry the full hiring risk from day one, with no buffer
• Statutory risk: every PF, ESI, ROC, GST, transfer pricing filing is yours from month one

Model 2: Build-Operate-Transfer (BOT)

A third party (typically a large setup partner) incorporates the entity in their name, builds the GCC under your specifications, operates it for a defined period (usually 36 to 60 months), and then transfers the entity, the team and the assets to you at a pre-agreed valuation. You pay a build fee, a monthly operating fee with a margin, and a transfer fee.

Why BOT looks attractive

• Speed to first FTE: typically 8 to 14 weeks vs 24 to 36 for Direct
• You do not carry hiring or statutory risk in the first three years
• The partner has a vested interest in delivery because they own the entity
• If the GCC fails, you can walk away without an Indian liquidation

The five BOT failure modes we see most often

• 1. The transfer valuation is opaque. You discover at month 36 that the agreed price is now 1.8 to 2.4x what you forecast, because it includes goodwill, brand, vendor contracts and a margin on the team.
• 2. Talent attrition spikes at transfer. The partner has been the legal employer for three years. When ownership changes, 20 to 30% of the team uses the moment to negotiate, leave or be poached.
• 3. IP ownership is murky. Code, documentation, processes and tooling sit inside the partner entity. The transfer agreement may not cover everything cleanly, and you discover gaps only when the partner has already lost interest.
• 4. The partner optimises for their margin, not your maturity. You end up with a team that is operational but never strategic, because the partner has no incentive to build leadership depth.
• 5. The operating fee compounds. A 12 to 18% margin on USD 12 million annual opex is USD 1.5 to 2.2 million per year. Over four years that is USD 6 to 9 million you will never recover.

When BOT genuinely wins

• You need a GCC live in 90 days because of a commercial commitment
• Your board will not approve direct entity incorporation in year one
• The work is well defined, repeatable and not deeply IP sensitive
• You have negotiated a transparent transfer formula upfront, not a goodwill clause

Model 3: GCC-as-a-Service (managed captive)

You sign a managed services contract with a setup partner who provides the people, the entity, the office and the operations under your brand and your governance. There is no planned transfer. You pay a per FTE per month fee, typically loaded 22 to 30% over the underlying cost. You can convert to a Direct or BOT model at any time, usually with a 90 day notice period.

Why GaaS is the fastest growing model in 2026

• Optionality: you can scale from 10 to 150 FTEs without entity overhead
• Speed: first FTE in 4 to 8 weeks
• Reversibility: if the experiment fails, the cost of exit is one quarter of fees, not entity liquidation
• You learn the Indian market in real time before committing to a Direct or BOT path

When GaaS fails

• Beyond 150 to 200 FTEs the per FTE loaded fee starts to outweigh Direct economics by USD 2,500 to 4,000 per FTE per year
• If your work is deeply IP sensitive, a managed model may not pass your security committee
• Long term retention of a managed team is harder; people prefer a permanent badge

The decision framework

• Under 50 FTEs, planned tenure under 3 years, work not IP critical: GCC-as-a-Service
• 50 to 150 FTEs, planned tenure 3 to 5 years, conviction still building: GaaS with a 24 month conversion option to Direct
• 150 plus FTEs, conviction high, IP sensitive: Direct GCC
• 150 plus FTEs, board demanding 90 day go-live, willing to pay the margin: BOT, but only with a transparent transfer formula and an IP schedule annexed to the contract

The five contract terms that matter more than the model name

• Transfer valuation formula (for BOT): cost plus, not market value. Pin it in writing.
• IP assignment: every line of code, every document, every tool is yours from creation, not from transfer
• Talent non-poach and retention bonus pool at transfer (for BOT)
• Conversion clause (for GaaS): a defined path to Direct or BOT with no penalty beyond unwinding cost
• Exit clause: notice period, transition support, data return, in every model

What we recommend in 2026

For 75 to 80% of mandates under 150 FTEs we recommend starting in a GaaS model with a hard 24 month review point and a pre-negotiated conversion path. For mandates with high IP sensitivity or 200 plus FTE conviction from day one, we recommend Direct. BOT is the right answer in maybe 10% of cases, and even then only with an iron clad transfer formula. The model that is almost always wrong is BOT with a goodwill based transfer valuation, which is sadly the most commonly signed contract in the Indian market today.

What to do this week

If you have a BOT proposal on your desk, ask the partner for the transfer valuation formula in writing before you sign anything. If they cannot give you a deterministic formula, you are signing a blank cheque for year four. If you would like a model comparison sized to your headcount and tenure, we will return one within 48 hours via the enquiry form on this page.