GLOBAL CAPABILITY CENTERS
MICROSERVICES FOR THE ENTERPRISE
MICROSERVICES FOR THE ENTERPRISE
How a concept from software architecture explains modern enterprise design
"Microservices are primarily a technical solution to a people problem: allowing different teams to make progress independently without having to coordinate with each other. "
- Martin Kleppmann
November 15, 2026 | RACHIT GARG
Most multinational enterprises are still organised like monolithic software: critical decisions, capabilities, and dependencies concentrated at headquarters.
But a new generation of Global Capability Centres (GCC) is beginning to challenge that architecture. They are taking entire ownership of selective business capabilities, operating with increasing autonomy while being located at standalone premises.
There is a name for this architectural idea in software engineering: microservices.
As someone who began as a software developer, went on to pursue an MBA, and now works with global companies building GCCs, I see an increasingly interesting convergence between software architecture and enterprise design.
Which begs the question: What, exactly, is a GCC?
The textbook answer is "a company-owned offshore centre that serves its parent," usually framed as an evolution of BPO (Business Process Outsourcing). The BPO model gained prominence through cost-arbitrage back offices. But it no longer explains what I see with GCCs.
In 2024, Nasscom projected that India’s GCC sector would reach $99 to $105 billion by 2030. According to the FY26 Nasscom - Zinnov GCC Landscape Report, the sector has arrived four years early: India already generates $98.4 billion across 2,117 GCCs (3,728 units) and employs 2.36 million people.
The work inside these centres looks nothing like a traditional back office and the appeal has moved beyond cost arbitrage. Over 100 new GCCs were established or expanded this year, including Anthropic, Eli Lilly, FedEx, Marriott, and Lufthansa. India now holds roughly 28% of the global GCC AI talent pool (second only to the US), and nearly three in four Indian GCCs are projected to evolve into global portfolio or transformation hubs within five years.
Back offices do not become AI engineering hubs by getting a little better. The current evolutionary model fails to explain how its six-year targets were achieved in just two years. The pace of change hints towards a more structural shift powered by value creation rather than cost arbitrage. To find the right model, we have to leave business for a moment and look at software.
(Engineers: think of a mature GCC as a team in another geography that owns a business capability end to end. Business readers: the next section is about software, but it is really a story about organisational design.)
Early software was built as a monolith: one codebase, one database, deployed as a single unit. That works until the team grows. With hundreds of engineers in one codebase, every change risks affecting other parts, and shipping requires agreement across the board.
Amazon hit this wall in the early 2000s, shortly after the dot-com crash had shown how brutally markets punish companies that cannot adapt. Its key bottleneck was coordination cost: too many people tangled in one system to innovate. Amazonians responded by progressively splitting the monolith into independently owned services and reorganising teams around those services. This became associated with the famous "two-pizza team" model: small, autonomous teams of roughly five to ten specialists, small enough to be fed by two large pizzas.
Each team owned its service end to end, interacting with other services through defined interfaces, or Application Programming Interfaces (APIs). This architectural approach eventually became known as microservices.
The payoff is twofold:
Agility: A team ships without waiting on the rest of the organisation.
Resilience: When one service fails, the impact stays contained.
The underlying principle is: distribute ownership, standardise interfaces and minimise unnecessary dependencies.
As Martin Kleppmann, in his book Designing Data-Intensive Applications, observes, microservices are fundamentally a technical solution to a people problem. By reducing coordination overhead, they allow teams to make progress without centrally negotiating every change.
Replace "codebase" with "corporation." A multinational that concentrates critical functions and decision-making in one or two headquarters is an organisational monolith: tightly coupled, slow to adapt, and one tariff shift or supply shock away from an outage.
In an era of tariff shifts, supply-chain fragmentation, macroeconomic headwinds, and rapid AI disruption, agility and resilience have become survival requirements.
Yet many enterprises remain structurally centralised because enforcing standardised processes at scale historically required extensive management coordination. That meant building deep managerial hierarchies to coordinate work, maintain process consistency, and enforce accountability.
Flat autonomous structures, by contrast, were more naturally suited to exploratory work such as R&D and innovation. Such teams work on self-contained problems, are measured on outcomes rather than process compliance, and operate with minimal daily dependencies on core operations. Because of their loose coupling to the rest of the business, Innovation hubs were among the first business-critical capabilities to break away from traditional enterprise structures as early GCCs.
AI is now making a similar structural shift possible across a much broader range of functions.
By absorbing much of the routine work required to coordinate and control standardized processes (monitoring, reporting, routing, and workflow execution), AI is beginning to reduce organisational coordination costs. As coordination becomes easier, enterprises can hire more specialists than managers, flatten org charts, bring those specialists together in dedicated capability centres, and give them decision-making ownership. Headquarters is freed to focus on overall business strategy, while GCCs own the functional strategy that delivers it.
The internet did the same for code: the barrier to splitting a software monolith was network connectivity, and the internet removed it. For human enterprises, the barrier was coordination, and AI is removing it.
Empirical research supports the case for decentralisation. A 2021 study published in the American Economic Journal: Applied Economics, found that firms delegating more decision-making power to local plant managers outperformed their peers in sectors hardest hit by the Great Recession.
Technology alone , however, cannot create organisational autonomy. Enterprises must deliberately redesign ownership, accountability, and decision-making around the capabilities they distribute.
That is a GCC: A Microservice For The Enterprise.
Engineers learned the hard way that splitting a monolith is easy, and operating microservices is hard. Five practices separate the enterprises that pull this off from the rest:
Build niche talent deliberately. Universities train generalists for an industry, not specialists for your capability. Companies must develop internal ramp programmes to grow their own specialists.
Draw boundaries around outcomes, not tasks. Scope each centre broadly enough to minimise daily dependency on other locations, and narrowly enough to keep its internal mission cohesive.
Define the interfaces and decision rights. Services publish APIs and service contracts. Centres need clear equivalents: explicit decision authority, SLAs, and escalation paths.
Default to asynchronous hand-offs. Message queues exist because software services cannot freeze and wait on each other. Across time zones, people cannot either. Written RFCs and structured state logs beat 9 p.m. status calls.
Build discovery and deliberate redundancy. Any team should be able to find and engage a capability through an internal directory without needing personal connections, and critical capabilities should have planned failover across centres.
Above all, beware the distributed monolith: software services so entangled that you pay the full tax of distribution while gaining zero independence. A GCC that cannot make product or operational decisions without headquarters’ sign-off is the exact same failure.
GCCs are not BPOs that grew up. They are enterprises re-architected for adaptability and endurance. The next generation of GCCs will not be defined by how much work they absorb from headquarters, but by how much ownership they take on. Ultimately, the future of the multinational enterprise may not be one organisation operating across many locations, but many autonomous capabilities operating as one organisation.
For decades, the most ambitious talent had to move to where the decisions were made. What happens to the world when the decisions move to where the talent is?
Sources: Nasscom–Zinnov GCC Landscape Report FY26; Reuters; Kleppmann, Designing Data-Intensive Applications.