10decodersAsk five advisors what a GCC in India costs and you will get five different numbers, because the honest answer is a range that depends on how many people you hire, which city you pick, and whether you own the entity or have someone run it for you. This is the breakdown behind that range: what you pay once, what you pay every year, and where the AI-first version of the model saves real money.
01Why the number is a range, not a figure
Three variables move the total more than anything else. The first is headcount: cost scales almost linearly with loaded seats, so a 20-person center and a 200-person center live in different budget universes. The second is city tier, where a Bengaluru or Hyderabad address carries a premium over an emerging hub like Coimbatore, Madurai, or Chennai. The third is the operating model, meaning whether you register and run your own legal entity, use a build-operate-transfer arrangement, or work through a delivery partner. Each of those pushes the same team's cost up or down by a wide margin.
02One-time setup costs
These are the costs you pay to exist and open the doors. They land in the first three to four months, before the center produces much of anything.
Ranges reflect published 2026 industry estimates; actuals vary by city, size, and model.
03Recurring run-rate
This is the number that matters after year one, and salaries dominate it. The gross figure is only part of the story: the loaded cost of an engineer includes statutory contributions and overhead that first-time buyers routinely under-budget.
The gap between gross salary and fully loaded cost is roughly 20 to 30 percent once provident fund, gratuity, insurance, office, and IT are layered in. Budget on the loaded number, not the offer letter.
04Loaded cost per engineer
Once everything is layered in, a fully loaded engineer in an Indian GCC typically lands between $25,000 and $80,000 a year depending on seniority, skill, and city. Against a comparable US hire, that is a 40 to 60 percent reduction in total operating cost, which is the arithmetic that has drawn more than 2,100 GCCs to India. The premium roles, AI and machine learning engineers, security specialists, and senior architects, sit at the top of that band and are where the market is tightest.
05City tier changes the math
Bengaluru and Hyderabad remain the deepest talent markets and carry the highest salary and real-estate costs to match. Emerging hubs trade a shallower senior talent pool for materially lower salaries and lease rates, which is why a growing share of new centers are choosing them for roles that do not require a dense specialist market. The right answer is rarely all-in on one city; it is placing specialist roles where the talent is and volume roles where the cost is.
06Own entity, BOT, or partner-led
The operating model is a control-versus-speed trade. Registering your own entity gives you maximum control and the lowest steady-state cost per seat, at the price of a longer setup, carried compliance overhead, and full exposure to hiring risk. Build-operate-transfer shortens the runway by having a partner stand the center up and hand it over later, at a premium. A partner-led model keeps the center off your books entirely while someone else carries delivery. For most mid-market companies the decision comes down to how quickly they need output and how much operational risk they want to hold.
07The micro and nano GCC: the lowest-risk way to start
Every cost above assumes you begin at scale. You do not have to. A micro GCC is a small capability center, usually somewhere between eight and thirty people. A nano GCC is smaller still, a single pod of a few specialists pointed at one workload. Both exist to do one thing: prove the model before you commit to building the rest of it.
This is the lowest-risk way to enter India. The upfront outlay is a fraction of a full build, you can run it without registering and carrying your own legal entity on day one, and you scale only the roles that have already earned their place. If the pod delivers, you grow it. If a workload does not pay off, you have risked a few months, not a few million dollars.
This is where we focus at 10decoders. We build and operate micro and nano GCCs for mid-market companies, staffed lean and built around delivered digital workers rather than rows of seats. A single engineer running a library of agents covers work that used to need a team, so the cost base is measured in outcomes instead of headcount, and you own the IP the pod produces from the first month.
Start with a nano pod, prove one workload, then scale only what works. You put months and thousands on the line before you decide, not years and millions.
08An illustrative starter budget
Here is what a lean starting point looks like: an eight-person micro GCC, built and operated for you, with no own-entity setup to carry in year one. Treat it as a shape, not a quote.
Illustrative only. Against the $1.2M to $2.1M a full 30-person own-entity build runs, a micro start puts a fraction of the capital at risk, and you scale seat by seat as each workload proves out.
Thinking about a micro or nano GCC?
We will scope a lean starter pod around your first workload, built and operated for you, with a clear path to scale only what proves out.
Talk to Edrin Thomas, CTOthomas@10decoders.com · 10decoders.com