India’s 170 BFSI GCCs Are the 8% Slice Owning AI Finance

India is home to 170 BFSI Global Capability Centers, roughly 8% of the 2,117 Global Capability Centers the country now runs, and the slice is the dense end of the country’s Global Capability Center map. Sixty of those 170 centers have opened in the past five years, and a fresh framework from Zinnov lays out where the AI bets of global banking and insurance are landing first. The numbers are tight, and the work those centers own is the kind global finance cannot run from anywhere else.

India’s broader GCC sector has long stopped being judged on cost. It owns products, runs global platforms, and makes decisions that used to sit only in head offices.

The Small Slice That Holds the Most

India’s Banking, Financial Services, and Insurance (BFSI) sector runs 170 GCCs and 333 units. That is roughly 8% of all the centers in the country and close to 9% of all the units, a small slice by count, a dense one by footprint. BFSI runs almost two units for every center it operates, a little more than the sector average, and the broader FY2026 picture, per the India GCC landscape report from Nasscom and Zinnov, puts the total at 2,100+ centers, USD 98 Bn in revenue, and 2.36 Mn people.

• 2,100+ Global Capability Centers in India across ~3,700 units (FY2026)
• USD 98 Bn in annual revenue from the GCC sector
• 2.36 Mn people employed across the GCC sector
• 32% sector growth since FY2021
• 170 of 2,117 GCCs are BFSI, holding 333 units

Banks and insurers do not open one office and stop. They spread risk, talent, and platforms across several cities on purpose, and that is exactly how a regulated, resilience-minded industry behaves.

60 of those 170 BFSI centers, more than a third, opened in the last five years, and the arrival rate, with the depth of mandate behind it, is the strongest evidence yet that global finance treats India as a place to build core capability, with names like JP Morgan Chase, HSBC, Wells Fargo, Citi, Standard Chartered, and Barclays building BFSI centers that own real mandates from day one.

Those are the numbers the BFSI slice sits inside, and the 8% by center count, 9% by unit count ratio is what makes the slice dense without being large. The 60 BFSI centers that opened in the last five years explain how 170 centers turned into 333 units.

Six Cities, 289 of 333 Units

For all that growth, BFSI work sits in surprisingly few places. Six cities in India hold 289 of the 333 BFSI units, close to 9 in 10. The remaining 28 units sit everywhere else in the country, and the table below shows how the units are spread across the six BFSI cities.

Bengaluru leads the list with 80, followed by Mumbai with 56, Hyderabad with 51, Pune with 36, Chennai with 35, and Gurugram with 31. The number worth pausing on is Mumbai, which ranks second, ahead of even Hyderabad, in a vertical where most other GCC verticals put the southern tech hubs first. Mumbai’s rank is the visible sign of BFSI’s gravity center, which is where the banks, insurers, and capital markets are headquartered. The rest of the top six splits the work by capability, with Bengaluru bringing engineering and AI depth, Pune and Chennai bringing scale and stability, and Gurugram bringing the National Capital Region talent pool.

Mumbai is where India’s banks, insurers, and capital markets are headquartered, and financial firms like to keep risk, treasury, and regulatory work within reach of the people who answer for it. BFSI bends the GCC map toward its own center of gravity, and that center is Mumbai, with the other five BFSI cities hosting the engineering, AI, scale, and talent work that supports it.

The New Wave Arrived Mature

60 of the 170 BFSI centers opened in the past five years, more than a third of the total. They came for capability, and they came mature, with the kind of mandate the older centers spent a decade earning. The new wave skips the back-office phase entirely, and it is the strongest signal yet that global finance treats India as a place to build core capability.

Six banks sit at the center of this wave. Names like JP Morgan Chase, HSBC, Wells Fargo, Citi, Standard Chartered, and Barclays run BFSI centers in India that own real mandates from day one, across risk, analytics, engineering, and AI, and they hire for product ownership from the start.

The 5-year wave is not a fluke. Zinnov describes it as the strongest indicator yet that global finance treats India as a place to build core capability, with risk, analytics, and AI work as the stated reason for arriving. The named banks at the center of this wave are the same ones that own the risk, treasury, and regulatory mandates of global finance.

The next chapter is whether these centers can turn that mandate into production AI at the speed the parent banks and insurers need it, and with 60 of the 170 BFSI centers under five years old, the recent wave still has more climb ahead of it than the older centers. The next 18 months will sort the ones that own the AI work from the ones that just support it.

Where the AI Work Goes Now

The AI work in a BFSI center now goes to five places, according to Zinnov, and they are not boxes to tick in order. They are the places where owning the outcome matters more than adding headcount. The five steps are the framework, not a checklist, and each one is a place where a BFSI center either owns the AI work or feeds it to a competitor that does.

1. Put AI to work in risk, fraud, and compliance. The biggest shift is AI moving out of scattered pilots and into one fabric that runs across document intelligence, risk scoring, fraud analytics, customer insight, and regulatory reporting. Mature BFSI centers are well placed to own this, because the work sits right next to the risk and compliance expertise they already have.
2. Treat AI model governance as part of the build. In banking and insurance, a model in production is a regulated decision. It carries the same weight as a credit policy, and regulators will treat it that way. The centers that lead are standing up model governance councils with real tooling for monitoring, fairness, and explainability, and building that governance in from the first line of code.
3. Lead the shift to cloud-native core banking and payments. Core banking, payments, and digital channels are moving to cloud-native, microservices architecture, and BFSI GCCs are increasingly the ones leading that modernization. The strongest centers wire security and threat response into the release pipeline itself, so the safest version of the bank is the one being built in India first.
4. Hire for judgment, not just for roles. The talent model has to change with the mandate. The best centers combine deep domain knowledge in risk, treasury, underwriting, and compliance with serious data and engineering skill, putting them together in small cross-functional pods that invest heavily in upskilling on AI and financial regulation.
5. Become the bank’s front door to the fintech ecosystem. The leading centers are turning into the orchestration layer for fintech partnerships, regulatory sandboxes, and pilots across payments, digital assets, RegTech, and ESG-linked products. India’s startup scene is on the doorstep, and the centers that build a clean path from a sandbox pilot to a production system inside the bank turn that proximity into a real advantage.

Each step is also a place where the center either owns the AI work or feeds it to a competitor that does, and the cost of skipping any of them shows up in the regulator’s view, the parent’s roadmap, or both. The centers that move fastest on the governance step in particular will be the ones setting the standard for the rest of the sector.

Governance as the License to Operate

Step 2 in Zinnov’s framework is the one most often skipped, and the one most likely to decide which BFSI centers end up owning the AI work. In banking and insurance, a model in production is a regulated decision, and it carries the same weight as a credit policy. Regulators will treat it that way, and the centers that lead are standing up model governance councils with real tooling for monitoring, fairness, and explainability.

The cost of skipping governance shows up in two places at once, and a separate Zinnov study on AI disruption in India GCCs has warned that more than half of the work inside Indian GCCs sits in commodity and procedural tasks that AI is on track to absorb. The leaders are the ones already moving past those tasks, and governance is the license to operate.

AI has fundamentally altered the pace and nature of change. What used to take a decade is now unfolding in cycles of months. Every new model release has the potential to redraw the boundaries of work, compressing expertise into procedure and procedure into automation. The portfolio itself, the very foundation of a GCC, is now the most exposed layer.

That is Pari Natarajan, chief executive officer and co-founder of Zinnov, on what AI is doing to the work inside India’s GCCs, including the BFSI slice.

The portfolio Natarajan describes is the work mix inside the center, and that mix is what is being rewritten, with a separate Zinnov analysis finding that 45% of India GCC jobs now need expertise and high-end research, compared with 25% a decade ago. BFSI centers that own the governance step now will be the ones still on the right side of that line five years out, and the ones that wait will be buying tools from the centers that did not.

From GE’s Car Loans to AI Engines

The shape of what BFSI centers in India own today is the result of a 30-year arc that started in 1997 with an experiment. GE Capital set up a small office in Gurgaon with about 20 people and a simple mandate: run the parent company’s back-office finance work from India. Most people he consulted told the man running it that the idea would never work beyond a few hundred people.

By the early 2000s the operation had grown to well over 10,000 employees, and by 2005 it spun out as Genpact, a standalone company. The model it proved, that serious, regulated financial work could be run from India, became the template that banks and insurers around the world would follow.

The Gurgaon office is the parent of every BFSI GCC operating in India today, and the 60 that have opened in the last five years arrived mature. They own real work from day one, and the climb that took GE Capital a decade is being skipped entirely. The Zinnov analysis frames the arc in one line: in under thirty years, India went from processing GE’s car loans to running the risk, fraud, and AI engines of global finance.

The 18 Months That Will Decide

The framework and the history both point to a tight window. The same Zinnov analysis that mapped the BFSI slice also says the next 18 months are about how much of the AI work in banking and insurance these centers end up owning, with three forces converging at once: AI moving from scattered pilots into the core fabric of the institution, regulators sharpening their view of what counts as a governed model, and the recent wave of BFSI centers still small enough for the leaders to set the standard.

The shift looks familiar from another industry that has spent the last two years being rebuilt by AI, and automakers rebuilding around AI and MLOps is the parallel pattern, with continuous safety governance as the regulator-facing constraint. BFSI is the same shape on the financial services side, with a faster adoption curve and a tighter regulatory perimeter, and the next move happens in the same six cities, where the centers already sit, with the corner office of the parent bank a step behind. BFSI centers in India are long past the question of whether they matter, and the next 18 months will settle how much.

Frequently Asked Questions

What is a BFSI Global Capability Center?

A BFSI Global Capability Center is an offshore unit set up by a bank, insurer, or other financial firm to run a specific slice of work for the parent company. In India these centers handle risk, fraud, compliance, analytics, engineering, and increasingly AI, and the recent arrivals own those mandates from day one.

How many BFSI GCCs are there in India?

Zinnov’s FY2026 analysis counts 170 BFSI GCCs running 333 units across the country. That is roughly 8% of India’s 2,117 Global Capability Centers and close to 9% of all units, a small slice by count and a dense one by footprint. The same analysis finds that 60 of those 170 BFSI centers opened in the last five years, which is more than a third of the total.

Which cities host the most BFSI GCCs?

Six cities host 289 of the 333 BFSI units: Bengaluru with 80, Mumbai with 56, Hyderabad with 51, Pune with 36, Chennai with 35, and Gurugram with 31. The remaining 28 units sit elsewhere in the country, and Mumbai is the gravity center for the BFSI work because the banks, insurers, and capital markets are headquartered there.

Why are global banks opening BFSI centers in India now?

60 of the 170 BFSI centers have opened in the past five years, more than a third of the total. Zinnov’s analysis describes the new wave as arriving already mature, with mandates across risk, analytics, engineering, and AI from day one. Names like JP Morgan Chase, HSBC, Wells Fargo, Citi, Standard Chartered, and Barclays have built centers in India that own real work from the start.

What are the five steps Zinnov outlines for AI in BFSI GCCs?

The five steps are: put AI to work in risk, fraud, and compliance; treat AI model governance as part of the build; lead the shift to cloud-native core banking and payments; hire for judgment, not just for roles; and become the bank’s front door to the fintech ecosystem, with the regulatory sandboxes and engineering depth to take pilots to production. Each step is a place where the BFSI center either owns the AI work or feeds it to a center that does.

How risky is putting AI into regulated finance?

Zinnov has warned that more than half of the work inside Indian GCCs sits in commodity and procedural tasks that AI is on track to absorb. The same analysis finds 45% of GCC jobs now need expertise and high-end research, compared with 25% a decade ago. The centers that own the governance step set the standard for the rest, and the ones that skip it are exposed to both the regulator and the AI displacement curve at the same time.