GCCs Out-Hired Indian IT Services 200K to 110K. Your Boolean Misses 583 Mid-Market Captives.
India's GCCs added 200,000 net hires in FY26 vs 110,000 for IT services. Here is how to source the mid-career engineers moving into 2,117 captive centers.
If your India sourcing playbook still runs on "TCS" OR "Infosys" OR "Wipro" OR "Cognizant", you are searching a shrinking pond. The Xpheno data published in Business Standard last week (May 2026) made it official: India's Global Capability Centers out-hired the entire IT services sector almost two to one in FY26, and the mid-career engineers you actually want have already moved.
The problem is not that recruiters are lazy. The problem is that the destination cohort is structurally invisible to Boolean.
The FY26 inversion, in numbers
Xpheno's tally for FY26: GCCs added roughly 200,000 net headcount. IT services added about 110,000. That gap was 110K vs 100K in FY25 and 60K vs 50K in FY24. FY26 is the widest divergence on record, and it is not a one-quarter wobble.
The services side is shedding, not stalling. TCS's workforce shrank by about 24,000 in FY26 to 584,519 employees. Fresher offers fell from 44,000 in FY26 to roughly 25,000 for FY27, a near-halving. Sector-wide, IT fresher hiring has collapsed from 600,000+ in FY22 to an estimated 120,000 in FY25, with FY26 flat.
Meanwhile Nasscom's GCC count has crossed 2,117 centers, employing 2.36 million professionals and generating $98.4 billion in revenue, up from roughly $65 billion two years ago. 506 Forbes Global 2000 companies now operate a GCC in India. TeamLease Digital projects GCC recruitment up 50 to 75% in FY26 vs FY25, creating 340,000 to 380,000 new tech jobs over five to six quarters.
The net delta also undersells the gross flow. IT services involuntary attrition in FY25 ran 15.1% at Wipro, 14.1% at Infosys, and 13.3% at TCS. Most of those people did not leave the industry. They walked across the street into a captive.
Why your Boolean misses them
A standard "ex-TCS, 10 to 14 years, Bengaluru" Boolean will surface laterals who landed at JP Morgan, Barclays, Goldman, Wells Fargo, Walmart Global Tech, or AWS India. Those are the easy ones. The hard ones, and the larger pool, sit in the long tail.
Of the 2,117 GCCs in the Nasscom-Zinnov FY26 dataset, 583 are mid-market and 423 generate under $100M in revenue. That is more than a thousand captive employers most sourcers cannot name on demand. Many of them:
- File under opaque India-registered entities. "C3 India Delivery Centre" was Crowe Horwath IT Services. There are dozens of similar shells.
- Sit under parent holding companies on LinkedIn with no India page, so the employer search filter quietly drops them.
- Use generic titles ("Senior Engineering Manager") with no parent brand in the headline, because the engineer themselves is unsure how to position a captive role.
- Live in Tier-2 satellites (Bhubaneswar, Coimbatore, Madurai, Visakhapatnam) that recruiters do not check.
Walmart Global Tech expanded its data science team to Bhubaneswar. Kraft Heinz runs a 600-person GCC in Ahmedabad spanning analytics, finance, and tech. Eisai Pharma is in Visakhapatnam, Flex in Coimbatore, Infinx Healthcare in Madurai (700 jobs created). None of these are brands a recruiter free-associates with "India tech." All of them are hiring mid-career laterals out of services right now.
The destination cohort is structurally invisible to Boolean. That is not a skill issue. It is a graph problem.
This is why we built Refolk. You describe the person in plain English ("senior backend engineers in Pune with 10+ years at TCS or Infosys who moved to a US bank or insurance captive in the last 18 months") and get a ranked shortlist that resolves across LinkedIn, GitHub, and the open web. The long tail of 583 mid-market GCCs stops being a list you have to memorize.
The "missing middle" is the actual prize
Freshers are not the trade here. Mid-career is.
The cohort moving from services into captives has three properties that make them disproportionately valuable:
- Domain depth. Ten to fifteen years inside a TCS BFSI account means real exposure to a specific bank's tech stack, regulatory posture, and incident history. Captives pay a premium for that institutional memory.
- Global delivery experience. They have run distributed teams across time zones, which is precisely what a US- or UK-headquartered captive needs from its India leadership.
- No green-card lock-in. Unlike the H1B cohort in the US, this group can move on a four-week notice. Counter-offers from services firms are thin because the services firms are themselves shedding.
Several GCCs are actively recruiting laid-off mid-level IT employees specifically because they bring the domain knowledge and global experience captives cannot grow in-house fast enough. Kamal Karanth at Xpheno has been the loudest voice on this for two years, and the FY26 data finally caught up to his thesis.
The AI hiring narrative is partly theater
A clean read of the FY26 data: nearly half of GCCs set up since FY21 were designed AI-first, 1,200+ have embedded AI/ML capabilities, 250+ Centers of Excellence, and a stated pool of ~250,000 AI professionals. New GCCs in 2026 are being designed with 15 to 25% of headcount in AI/ML within 18 months of launch.
That is the press release. The job-posting reality is narrower.
A March 2026 snapshot of bellwether GCCs showed very limited deep AI-engineering openings relative to total GCC hiring. Most "AI hiring" is applied AI and AI literacy roles bolted onto existing data, cloud, and platform teams. The actual hiring volume is in:
- Cloud platform engineering (AWS, GCP, Azure)
- Data engineering (Databricks, Snowflake, dbt)
- Site reliability and observability
- Domain backend (payments, claims, supply chain)
- Security engineering (especially for BFSI captives)
If you have been chasing "GenAI engineer" titles in India, you are competing for a small sliver of the market with every other recruiter who read the same Nasscom press release. The volume is in domain plus cloud plus data retreads, and those people will not put "GenAI" in their headline.
Tier-2 is overhyped for senior roles
The hub-and-spoke narrative is real, but it is being misread. Pune grew from 210 GCCs in 2019 to over 360 in 2025. Bengaluru hosts 880+ GCC centers, Hyderabad 355+, with five new Hyderabad GCCs being stood up concurrently as of early 2026. Tier-2 attrition runs 30 to 40% lower than Tier-1's 16 to 18%, which is why CFOs love the spoke story.
But the senior layer, the 12 to 20 year mid-management cohort, is scarce in Tier-2. Most of that talent is embedded in metro GCCs or is not actively looking. Recruiters chasing the Coimbatore or Indore narrative for senior captive roles will hit a wall.
The right sourcing read is hub-and-spoke: Tier-1 leadership, Tier-2 execution. A TCS Pune lead who is 12 years in is more likely heading to a Pune-based captive spoke of a US bank than to a Bengaluru AI role. A Chennai TCS architect is more likely landing in a Coimbatore or Madurai spoke of the same employer than relocating.
A practical playbook for FY26 sourcing
If you are running India req against the FY26 reality, six things to change this quarter:
1. Replace brand-list Boolean with intent search. "TCS OR Infosys" filters for past employer, not destination. The destination is 2,117 captives with no clean tag. Describe the person you want in plain English. This is exactly the friction Refolk was built to remove: ask for "ex-services mid-career engineers now at BFSI captives in Pune" and stop maintaining a Boolean string that was already out of date two quarters ago.
2. Build a captive-shell dictionary. For your top 20 client accounts, map the India entity name, the LinkedIn page (often missing or under the parent), and the actual hiring city. "C3 India Delivery Centre" type shells are everywhere. A 200-line internal dictionary will return more relevant profiles than a 2,000-character Boolean.
3. Treat attrition data as supply forecast. Wipro at 15.1%, Infosys at 14.1%, TCS at 13.3% involuntary attrition is your supply curve. Cross-reference against TeamLease's 340,000 to 380,000 new GCC roles over five to six quarters. The math is unambiguous: most of those captive seats are being filled by services laterals.
4. Stop chasing GenAI titles. Source for cloud, data engineering, and domain backend with AI-adjacent project signal in the work history. The deep ML hires are a separate, much smaller market.
5. Use GitHub for the captive cohort that services hid. Ten years at TCS often means a thin public profile. But the captive move often coincides with the engineer finally being allowed to ship open-source contributions, internal blog posts, or conference talks. Pull GitHub and open-web signal alongside LinkedIn, not after it.
6. Anchor outreach on the move, not the brand. "Saw you moved from Infosys Pune to [captive] in 2024" lands harder than "Excited about your work at [captive]." The lateral is the salient identity event. Acknowledge it.
The 2:1 hiring gap is not a forecast. It is the FY26 result. The recruiters who treat the captive long tail as a first-class search target, not a footnote, will own the mid-career India market through FY27. The ones still running brand-list Boolean will keep wondering why response rates dropped.
FAQ
How is a GCC different from a traditional IT services employer in India?
A Global Capability Center is the in-house captive arm of a multinational, headquartered abroad but operating in India as cost center, R&D hub, or product line. The engineer is a direct employee of (for example) JP Morgan, Walmart, or Deutsche Bank, not of a third-party services vendor like TCS or Infosys billing them out. The economics, attrition, comp ladder, and reporting line are all different, which is why the mid-career flow from services to GCCs has accelerated.
Why does brand-name Boolean miss so many captives?
Of 2,117 Nasscom-counted GCCs, 583 are mid-market and 423 generate under $100M in revenue. Many file under opaque India entity names ("X India Delivery Centre," "Y Technology Services LLP") that do not match the parent brand. LinkedIn's employer filter often resolves to the parent, dropping the India arm, and engineers themselves often list the local entity in their profile. Boolean keyed to brand names systematically loses this long tail.
Is Tier-2 sourcing actually worth the effort for senior roles?
For execution layers (3 to 8 years experience), yes, Tier-2 has lower attrition (30 to 40% lower than Tier-1's 16 to 18%) and reasonable supply in Coimbatore, Ahmedabad, Kochi, and Bhubaneswar. For senior leadership (12 to 20 years), no. That talent is concentrated in metro GCCs in Bengaluru, Pune, Hyderabad, and Delhi-NCR, and is largely passive. Source the spoke for ICs, the hub for leadership.
What changed in FY26 that did not exist in FY25?
Three things compounded. TCS, Infosys, and Wipro moved from headcount stall to outright shrink (TCS lost about 24,000 employees and cut FY27 fresher offers to 25,000). The GCC count crossed 2,000 with $98.4B in revenue. And TeamLease projected 340,000 to 380,000 net new GCC roles over five to six quarters. The supply of laterals and the demand from captives both stepped up at the same time, which is why FY26 is the widest divergence on record rather than a smooth continuation of the FY24 to FY25 trend.