Why many India centres stall near forty engineers, and how to break the pattern.

If you run or sponsor a Global Capability Centre in India, you have probably seen the same curve: fast early growth, strong individual contributors, then a long plateau somewhere between thirty and fifty engineers. Boards ask why velocity flattened. Procurement asks why unit economics stopped improving. Your India leadership team works harder, yet throughput barely moves.

Across eighteen programmes we reviewed in depth over the last two years, the ceiling appeared for remarkably similar reasons. It was almost never “we could not attract good people.” It was coordination cost rising in lockstep with headcount, while the interfaces that should have absorbed that cost stayed informal.

The business problem underneath the plateau

At smaller scale, informal habits work. Product owners in London or San Francisco message leads in Bengaluru directly. Architecture decisions happen in a thread. Security exceptions are handled case by case. That is survivable at twenty-five people because everyone still holds the context in their heads.

Near forty engineers, the same habits become tax. Queues form behind a small set of headquarters approvers. Releases wait on individuals instead of cadence. Local managers inherit accountability for delivery without authority over hiring bar, tooling, or escalation paths. Attrition rises—not because pay is wrong, but because the job stops feeling like a credible career arc.

What good governance looks like at this stage

Programmes that break the pattern treat India as part of the same operating system as headquarters, not as a satellite that receives work packages. That means explicit product ownership, documented decision rights, and incident and security processes that mirror what you already expect in your home region.

• 18Programmes reviewedlast 24 months
• 35–50Typical plateau bandIndia engineering FTE
• 90 daysOperating window we co-ownbefore handover playbook

Operational insights from the delivery floor

Hiring ahead of governance is an expensive placebo. If your joiner-to-productive time is rising while defect density is flat, the constraint is not sourcing—it is clarity. The teams that recover fastest reset three things in parallel: architecture guild participation, release train ownership, and a single written map of who can decide what without a meeting.

Shadowing India leaders into customer ceremonies sounds soft; it is one of the strongest predictors we see of sustained velocity. When engineers understand why trade-offs matter, rework drops. When they only receive summaries, they optimise locally—and coordination cost rises again.

The board does not need another slide on “digital transformation.” It needs a credible story for why the next twenty hires will not recreate the same bottlenecks.

Strategic recommendations

First, run a ninety-day operating-system review before you approve the next tranche of requisitions. Second, fund local authority: if India cannot change outcomes within two escalation hops, you are still running a remote workcell, not a GCC. Third, publish a single RACI for product, platform, and security decisions that applies in every region—then instrument adherence the way you instrument uptime.

Closing: where to invest next quarter

If you only remember one point, make it this: scale follows interfaces. Harden the interfaces—decisions, data classes, security, and customer presence—and the forty-engineer ceiling stops being a mysterious law of nature. It becomes a predictable phase you can engineer through, with a playbook your executives can explain without outsourcing language or vague transformation rhetoric.