5 ways offshore engineering teams in India fail in 2026, and the structure that prevents it

Summary. India hosts 2,117 Global Capability Centres across 3,728 units as of March 2026, generating $98.4bn in revenue with 2.36 million people, according to the Zinnov-Nasscom GCC Landscape in India 2026 report. That is 32% growth since FY2021, and 506 Forbes Global 2000 companies now run a centre in India. The maturity data is the part worth reading twice: 13% of centres sit at Outpost, 43% at Satellite, 39% at Portfolio Hub and only 5% at Transformation Hub. In other words, the single largest cohort has capability at scale and still does not own outcomes. That is the failure mode this article is about, and it applies to a five-person partner team exactly as it applies to a five-hundred-person captive centre. Nasscom put the wider Indian tech industry at $282.6bn in revenue with 5.8 million employees for FY2025, which works out to roughly $48,700 of revenue per employee, a more useful planning anchor than any hourly rate card.

The teams that fail rarely fail on skill. They fail on how the work was divided, who was allowed to decide, and what the contract measured. Five patterns cause most of it.

This is written from the perspective of a firm that builds these teams, so read the closing section as what it is. Everything before it is the advice we would give a buyer who ended up choosing someone else.

What the market actually looks like in 2026

The Zinnov-Nasscom GCC Landscape in India 2026 report is the most complete public dataset on the ecosystem, drawn from 200 primary interviews with GCC heads, CTOs and CHROs, analysis of more than 1 million GCC job postings, and real estate data across India's top 8 GCC cities.

Its headline numbers as of March 2026: 2,117 GCCs, 3,728 GCC units, $98.4bn in revenue, 2.36 million people, 583 mid-market GCCs and 504 private-equity-backed centres. India is described in the same report as the number one AI hiring market globally.

The maturity framework matters more than the totals.

Maturity stage	Share of India GCCs	What the centre actually does
Outpost	13%	Cost arbitrage; delivery excellence is the ceiling
Satellite	43%	Capability at scale, without end-to-end ownership
Portfolio Hub	39%	End-to-end ownership of product, platform and IP
Transformation Hub	5%	AI-led operations with CXO mandates from India
Combined Portfolio and Transformation	44%	Own outcomes rather than execute specifications

Two readings of that table are both correct. The optimistic one: 44% of centres now own outcomes end to end, which was rare five years ago, and the report notes that 27% reach Portfolio Hub within five years. The uncomfortable one: 56% still do not, and Satellite is the largest single group. Capability was never the bottleneck.

Zooming out, Nasscom's Strategic Review put the Indian tech industry at $282.6bn in revenue for FY2025 with 5.8 million employees, growing 5.1% and adding 126,000 net new roles, split between $224.4bn of exports and $58.2bn of domestic revenue.

Sindhu Gangadharan, Chairperson of Nasscom, framed the shift this way: "Enhanced AI implementation, rise of Agentic AI, and the growing maturity of GCCs as value hubs are reshaping industry dynamics. As the Indian tech industry approaches the $300Bn revenue milestone in FY 2026, the intersection of technology, geopolitics, and trade demands a bold response."

Stop pricing this in dollars per hour

Published offshore rate cards are close to useless for planning. Search for a senior India developer rate in 2026 and the vendor rate cards will disagree with each other several times over for the same job title. The spread is not measurement error, it is different definitions of "senior" being sold to different buyers.

Two derived benchmarks from authoritative data travel much better.

Nasscom's FY2025 figures give roughly $48,700 of revenue per employee across the Indian tech industry ($282.6bn across 5.8 million people). The Zinnov-Nasscom GCC data gives roughly $41,700 per person ($98.4bn across 2.36 million). Both are averages across juniors and principals, services and product work, so treat them as the gravitational centre rather than a quote. What they tell you is the order of magnitude a sustainable India engineering seat carries at industry level, which is the number to sanity-check a proposal against.

Then add what the rate card never includes:

Cost line	Typically in the rate	Who actually absorbs it
Engineer time	Yes	Vendor
Ramp-up to productive output	Rarely	You, in weeks of slow delivery
Your own management overhead	No	Your senior engineers and PM
Context transfer and documentation	Sometimes	Shared, badly
Attrition and re-ramp	No	You, repeatedly
Timezone overlap hours	No	Whoever moves their evening

The honest framing is that the quoted rate is the smallest component of what an offshore team costs you in year one. The largest is usually your own senior people's attention, and nobody prices it because it does not appear on an invoice.

The 5 failure modes

1. the team is given tickets instead of problems

This is the Satellite trap, and it is the most common failure by a wide margin. Work arrives as specified tickets, the team executes them well, and nobody offshore ever owns a user outcome. Velocity looks fine. Quality looks fine. Two years later the team still cannot make a design decision without a call, and the 43% Satellite cohort in the Zinnov-Nasscom data is exactly this pattern at scale.

The fix is unglamorous: hand over a bounded product surface, not a backlog. Give the team a service, a feature area or a customer segment, with the metric it is accountable for, and let it decide how. If nobody offshore can name the number they own, you have a Satellite team regardless of what the contract says.

2. Seniority is bought at the wrong ratio

Teams get costed by headcount and staffed by whoever is available. The predictable result is a team of capable mid-level engineers with no one who has shipped and operated the thing being built.

A team that owns outcomes needs at least one person with real production judgement, and that person's time is where the cost genuinely sits. A cheaper team of six without that person will underperform a more expensive team of four with them, and it will do so while looking busier. If you are also choosing a stack at the same time, our React Native versus Flutter hiring decision framework covers how the talent market should shape that choice rather than follow it.

3. Timezone overlap is treated as a scheduling detail

An India team and a US West Coast team share very few natural working hours. Most engagements handle this by having the offshore team start early or finish late, indefinitely, and then wonder about attrition.

The structural fix is to reduce the need for overlap rather than to extend it. Asynchronous decision records, a written architecture decision log, and genuine ownership of a bounded surface all cut the number of decisions that require a synchronous call. Four hours of deliberate overlap with a team that can decide beats eight hours with a team that cannot.

4. Quality is inspected rather than built in

When the offshore team does not own the outcome, quality gets checked at the boundary: a review queue, a QA gate, a release checklist. Defects go back across a timezone and the loop takes a day. The team learns that correctness is somebody else's job.

Test ownership has to sit with the team writing the code, with the automation to make that practical. Our note on QA and test automation practice covers the mechanics. The organisational point is simpler: a team that cannot release cannot own quality, whatever the process document says.

5. the engagement is structured as a cost centre while the expectation is a product team

This is the contradiction underneath the other four. The contract measures hours, headcount and utilisation, and then everyone is disappointed that the team behaves like a supplier rather than an owner. The Zinnov-Nasscom maturity data describes centres that have outgrown their operating model, and the report names the gap directly: the ecosystem is ahead, the org chart is behind.

Pick one. A cost-centre engagement is a legitimate choice with a legitimate structure, and it should be measured on throughput and cost per unit. A product-team engagement needs outcome metrics, decision rights and a longer horizon, and it costs more per head. What does not work is buying the first and expecting the second.

Captive centre, partner team, or contractors

Dimension	GCC or captive centre	Dedicated partner team	Individual contractors
Time to first commit	Longest: entity, hiring, compliance	Weeks	Days
Fixed cost commitment	High and structural	Contractual, adjustable	Minimal
Ownership potential	Highest at maturity	High with a bounded surface	Low
Retention of context	Strong once past 2 years	Depends on partner attrition	Weakest
Suits	Sustained multi-year roadmap	Defined product scope, uncertain horizon	Bounded, well-specified work
Main failure risk	Stalling at Satellite stage	Treated as a ticket queue	No one owns the whole

The choice is mostly a function of horizon and certainty. A captive centre repays a multi-year, multi-team roadmap and punishes anything shorter. Contractors work when the specification is genuinely stable. The partner team sits between them and is chosen most often when the roadmap is real but the horizon is not yet certain enough to justify an entity. We worked through the first of those decisions in more detail in our comparison of building a GCC against engaging a product partner.

India-specific considerations

Data protection is now an engineering constraint, not a legal appendix. If the team will touch personal data of Indian residents, the Digital Personal Data Protection Act 2023 engineering obligations shape environment design, test data handling and logging from the first sprint. Retrofitting consent handling and data minimisation into a codebase after a year is one of the more expensive mistakes we see.

Intellectual property assignment deserves a real read rather than a template. Confirm that assignment flows from the individual engineer through the partner to you, that it survives an engineer leaving, and that it covers contributions to internal tooling and infrastructure as code, not only the product repository.

Location is a genuine variable now. Bengaluru, Hyderabad, Pune, Chennai and the NCR are all mature markets with different cost, attrition and specialisation profiles, which is why the Zinnov-Nasscom report tracks real estate and hiring across India's top 8 GCC cities separately. Gurugram and the wider NCR sit closer to Europe by timezone than the southern hubs, which matters more than most cost models assume.

Finally, treat AI capability as a hiring reality rather than a marketing line. The same report calls India the number one AI hiring market globally. That means the talent exists and also that it is competitively bid for, so a team you assemble on AI-heavy work will be more exposed to attrition than one built on conventional application engineering.

FAQ

How large is India's GCC market in 2026?

The Zinnov-Nasscom GCC Landscape in India 2026 report counts 2,117 GCCs across 3,728 units as of March 2026, generating $98.4bn in revenue and employing 2.36 million people. That represents 32% growth since FY2021, and 506 Forbes Global 2000 companies operate a centre in India alongside 583 mid-market and 504 private-equity-backed centres.

Why do most offshore engineering teams stall?

Because they are given specified tickets rather than a bounded product surface with an outcome metric. The Zinnov-Nasscom maturity data puts 43% of India GCCs at the Satellite stage, meaning capability at scale without end-to-end ownership. Skill is rarely the constraint; decision rights and accountability structure usually are.

What is a realistic cost benchmark for an India engineering seat?

Published hourly rate cards vary too widely to plan against. Nasscom's FY2025 figures give roughly $48,700 of revenue per employee across the Indian tech industry, and the Zinnov-Nasscom GCC data roughly $41,700 per person. Use those as order-of-magnitude anchors, then add ramp-up, management overhead and attrition costs.

How much timezone overlap does an offshore team need?

Less than most engagements assume, if the team owns decisions. Extending the offshore working day indefinitely to maximise overlap drives attrition. Reducing the number of decisions that require synchronous discussion, through written architecture decision records and genuine ownership of a bounded surface, works better than adding hours.

Should we build a captive centre or engage a partner team?

Horizon and certainty decide it. A captive centre repays a sustained multi-year, multi-team roadmap and punishes shorter commitments because entity setup, hiring and compliance take longest. A dedicated partner team suits a real roadmap with an uncertain horizon. Individual contractors suit bounded work with a genuinely stable specification.

What proportion of India GCCs own outcomes end to end?

Combining the two most mature stages in the Zinnov-Nasscom framework, 44% do: 39% at Portfolio Hub with end-to-end ownership of product, platform and IP, plus 5% at Transformation Hub running AI-led operations with CXO mandates from India. The remaining 56% sit at Outpost or Satellite stage.

Does India have the AI talent for product work?

The Zinnov-Nasscom GCC Landscape in India 2026 report describes India as the number one AI hiring market globally. Availability is not the constraint. Competition for that talent is, so teams built around AI-heavy work carry more attrition risk than teams built on conventional application engineering.

How fast do offshore teams reach full ownership?

The Zinnov-Nasscom report notes that 27% of centres reach Portfolio Hub within five years, and describes the maturity timeline as compressing, with maturity increasingly designed in on day one rather than earned over a decade. Ownership arrives faster when the operating model grants it early rather than staging it.

How eCorpIT can help

eCorpIT is a Gurugram engineering organisation founded in 2021, assessed at CMMI Level 5 and MSME certified, working with AWS, Microsoft, Google, Shopify and Kaspersky. We build dedicated product engineering teams that own a bounded surface and the metric attached to it, rather than a ticket queue, because the second one is what stalls. Our senior-led, multi-disciplinary teams take a defined product area with the seniority ratio the work actually needs, and we design applications aligned with DPDP Act requirements from the first sprint rather than retrofitting them. Talk to us about your engineering roadmap and we will start by mapping what the team would own.

References

1. Zinnov-Nasscom GCC Landscape in India 2026 report - Zinnov and Nasscom, data as of March 2026
2. Nasscom Strategic Review press release - Nasscom
3. Technology Sector in India: Strategic Review 2026 - Nasscom
4. India GCC Landscape Report: the 5 year journey - Nasscom Community
5. Beyond cost centres: the new metrics defining GCC value - Nasscom Community
6. India's ascent as the world's GCC powerhouse - Nasscom Community
7. Technology Sector in India: Strategic Review 2026 - Nasscom Community
8. Offshore software development rates by country and region 2026 - Aalpha
9. Software development rates in India 2026: rate card by role, seniority and stack - Acquaint Softtech
10. Offshore developer rates by country 2026 - Uvik Software
11. Offshore software development rates by country guide for 2026 - Qubit Labs
12. Nasscom Community - Nasscom

Last updated: 21 July 2026.