The Manufacturing Capability Centre: Why the IT GCC Playbook Does Not Transfer to Engineering and Plant Operations

TL;DR: The India GCC playbook was written by and for software companies, and most of its core assumptions break when the parent makes physical things. Engineering work is tied to rigs, prototypes and materials labs that do not travel over a network. Engineering talent sits in different cities from the software belt. The moment a capability centre acquires a workshop or a pilot line it stops being an office and becomes a regulated workplace under the Occupational Safety, Health and Working Conditions Code, 2020, which applies from 10 employees. Ramp is gated by equipment lead times and certification rather than by hiring. Zinnov’s India GCC Landscape Report: The 5-Year Journey, published in September 2024, found the setup rate of engineering, research and development GCCs growing 1.3 times faster than overall GCC setup, so this is a real and growing category. It is simply not the same build.

I have built and run capability centres in India, and I spent years at Eptisa running South Asia operations and the Afghanistan business, which meant carrying commercial responsibility rather than only a people brief. That combination is why I keep flagging the same problem with manufacturing clients.

They read the India GCC material, most of which is written about software companies, conclude the model is proven, and budget accordingly. Then the first engineering hire needs a test bench nobody costed, the first pilot line triggers a factory registration nobody scoped, and the programme is six months behind before anybody has admitted the plan was borrowed from a different kind of business.

There is no universal playbook. Successful organisations balance global standards with local realities, and the local reality of an engineering centre is not the local reality of a software centre.

Why manufacturers are asking the India question now

The commercial pull is real and it is measurable. According to the Press Information Bureau, the Production Linked Incentive schemes cover 14 sectors with an approved financial outlay of ₹1.91 lakh crore, and as on 31 March 2026 had drawn actual investment of over ₹2.40 lakh crore and generated over 14.15 lakh direct and indirect jobs.

The export figure is the one that should interest a board. PIB records cumulative exports under the PLI schemes rising from ₹4 lakh crore as on 31 March 2024 to over ₹15.2 lakh crore as on 31 March 2026.

That is the manufacturing story. The capability centre story runs alongside it and is frequently confused with it. A PLI-backed plant is a production decision. A capability centre is an engineering, design and support decision. They can reinforce each other, and I have seen them justified with the same slide, which is where the trouble starts.

For the shape of the wider base, we maintain a sourced directory in the India GCC Index 2026, which records 2,117 GCCs operating in India, notes that around 66 percent of mid-market GCCs focus on product and platform engineering rather than support work, and sets out the method behind every number.

What the IT GCC playbook actually assumes

Strip the standard India GCC business case back and it rests on six assumptions. Every one of them is reasonable for a software parent.

1. The work product is digital and moves over a network.

2. The talent pool is deep, fungible and concentrated in a handful of known cities.

3. The centre is an office, so the compliance surface is commercial property and employment law.

4. Capacity scales roughly linearly with headcount.

5. Intellectual property is protected contractually, through employment agreements and access controls.

6. The main operating constraint is time zone overlap with the parent.

Hold those in mind, because the argument of this piece is not that manufacturers cannot build capability centres in India. Plainly they can and do. It is that five of those six assumptions need rewriting before the business case means anything.

The five assumptions that break

The work product is not always digital. A CAD model travels. A durability rig, a materials lab, an EMC chamber, a prototype shop and a metrology room do not. Engineering work divides into the part that is genuinely digital, which transfers cleanly, and the part that requires physical apparatus, which requires capital expenditure, lead time and floor space that no software GCC business case contains. The first question for a manufacturing capability centre is not how many engineers. It is which of these activities need equipment, what does that equipment cost, and how long does it take to arrive and be commissioned.

The talent is in different cities. The software GCC map and the engineering talent map are not the same map. Automotive and industrial engineering depth sits around Pune, Chennai and Coimbatore in ways that have nothing to do with where the software industry clustered. Choosing a location by reading a general GCC city ranking is how manufacturers end up paying Bengaluru compensation for engineering skills that were more available elsewhere. Worth noting from our own Index data: only around 5 percent of India’s GCC units set up over the past year went to emerging cities, and two thirds of new GCCs over two years chose Tier I locations. The tier-2 story is smaller than the commentary suggests, which cuts both ways for a manufacturer evaluating a Tier II engineering hub.

The centre may not be an office. This is the assumption that causes the most expensive surprises and I have given it its own section below.

Capacity does not scale linearly with headcount. In a software centre, adding engineers adds throughput fairly directly. In an engineering centre, throughput is gated by test capacity, by validation cycles, and by certification and homologation timelines that are indifferent to how many people you hired. You can double the design team and not move the launch date if the rig is the constraint. Budgeting a manufacturing capability centre on a cost-per-FTE model, which is how almost every GCC business case is built, will misprice it.

Intellectual property protection is regulatory, not just contractual. Engineering design data in defence-adjacent, aerospace, dual-use and certain electronics categories sits inside export control regimes as well as inside your employment contracts. Which drawings can be viewed by which nationals in which location is a legal question with a documented answer, not a matter of access permissions. This needs answering before the org chart, because it can determine what work is permitted in the centre at all. Do not let anyone tell you an NDA covers it.

The sixth assumption, time zone overlap, mostly holds. It is the one thing the software playbook gets right for everybody.

The moment your capability centre becomes a regulated workplace

An engineering capability centre with a prototype workshop, a test lab or a pilot line is not an office with some machines in it. It is an establishment inside India’s occupational safety regime, and the thresholds are lower than most foreign parents expect.

Under the Occupational Safety, Health and Working Conditions Code, 2020, every establishment employing 10 or more employees must apply for registration within 60 days of its existence, and the Code covers health, safety and welfare across all sectors rather than the seven it previously applied to. As the Press Information Bureau’s factsheet on the Code sets out, the factory licensing threshold now sits at 20 workers with power and 40 without.

The obligations that follow are the ones that surprise people. A canteen is required at 100 or more workers. A crèche is required above 50 workers, and it is now gender-neutral rather than a facility for women employees only. A safety committee is required in every factory employing 500 or more workers. Daily hours are capped at 8 and weekly at 48, with overtime payable at double the normal wage rate, and the old 75-hour quarterly overtime ceiling is now fixed by the appropriate government rather than centrally, which makes it a state-by-state question.

Then there is contract labour, which manufacturers use far more heavily than software companies do. The contract labour provisions apply to an establishment with 50 or more contract workers on any day of the preceding 12 months, aggregated across all contractors. Where a contractor fails to pay wages, the principal employer becomes liable to pay those workers directly.

None of this is difficult. All of it is invisible in a business case borrowed from a software GCC, because a software GCC never encounters any of it. The broader picture of what changed sits in our guide to the four labour codes in practice, which covers the 50 percent wage rule, the fixed-term gratuity change and the appointment letter format prescribed by the Central Rules notified on 8 May 2026.

What transfers, what partly transfers, and what does not

The last row deserves emphasis because it is the one manufacturers underuse. Finance and HR shared services transfer more reliably than anything on the engineering side, carry none of the equipment or safety complexity, and are where most manufacturers should probably start. That is the work agile shared services is built around.

Choosing the model, and why manufacturers land differently

There are four routes into India: a wholly owned captive, build-operate-transfer, a vendor-led arrangement, and a managed team without a legal entity of your own. We stay model-agnostic on this and use GCC Compass to work the decision through, because the right answer depends on control requirements, investment appetite, speed to market and governance expectations rather than on which model is fashionable.

Manufacturers tend to land differently from software companies for three reasons.

The capital profile is different. An engineering centre with physical test capability has a capital expenditure component a software centre does not, which changes the calculus on build-operate-transfer, because you are transferring assets and not only a team.

The IP position is different. Where export control or design-data classification is in play, a vendor-led model may be foreclosed regardless of its commercial merits. That determination has to be made first, not tested at contract stage.

The ramp is different. A managed team can put engineers to work in weeks without an entity, which suits a manufacturer that wants to prove the talent thesis before committing capital to equipment. I have seen this used well as a deliberate first phase and badly as an indefinite substitute for a decision.

The mistake I see most often, and it is the same one in every sector, is selecting the delivery model before defining the business objective. The trade-off between what sits at headquarters and what the centre genuinely owns is exactly the kind of decision that goes unmade, which is a pattern we have written about in boardrooms generally: scale rarely fails on strategy, it fails in execution, when a trade-off nobody named becomes a gap nobody owns.

What to move first, and in what order

1. Run the export control determination before anything else. Establish which design data, drawings and software can lawfully be accessed from India and by whom. This is a legal question with a documented answer and it can eliminate options. Doing it after you have signed a lease is an expensive sequence.

2. Start with shared services, not engineering. Finance, HR and procurement transfer cleanly, prove the operating model, build the governance muscle and generate the cost case that funds the engineering phase. This is the least glamorous recommendation in this piece and the one that most improves the odds.

3. Split the engineering scope into digital and physical. List every activity you are considering moving and mark whether it needs equipment. Cost and schedule the physical list separately, with equipment lead times on the critical path. If the digital list alone does not justify the centre, the business case is resting on the physical list and needs proper capital planning.

4. Choose location on engineering talent density, not GCC rankings. Map where the specific disciplines you need actually sit. Then check the labour and compliance position of that state, because overtime ceilings and spread over rules are now set at state level.

5. Design the governance before the headcount. Decide what the centre owns outright, what it recommends, and what stays with headquarters, and write it down. Capability centres fail on unclear decision rights far more often than on talent quality.

6. Scope the regulated-workplace obligations the day a workshop enters the plan. Registration at 10 employees, factory licensing at 20 with power, crèche above 50, canteen at 100, safety committee at 500, plus contract labour aggregation at 50. Put these in the model at the start rather than discovering them at fit-out.

   Frequently asked questions

   Can a manufacturing company set up a GCC in India?

   Yes, and a growing number do. Zinnov’s India GCC Landscape Report: The 5-Year Journey, published in September 2024, found the setup rate of engineering, research and development GCCs growing 1.3 times faster than overall GCC setup, with energy and manufacturing parents among the new entrants. The important qualification is that a manufacturing capability centre is a different build from a software one. Digital engineering work transfers cleanly. Work requiring test rigs, prototype shops or pilot lines carries capital expenditure, equipment lead times and occupational safety obligations that a software GCC business case does not contain. Plant operations do not transfer at all, by definition, because they are co-located with the asset.

   Does the GCC model work for engineering and R&D?

   It works, with three conditions. First, the export control and design-data classification position must be established before the operating model is chosen, because it can rule options out. Second, capacity must be planned against test and validation capability rather than headcount, since throughput in engineering is gated by rigs and certification cycles rather than by people. Third, location must be selected on engineering talent density in the specific disciplines required, which frequently points to Pune, Chennai or Coimbatore rather than to the cities that dominate general GCC rankings.

   What functions can a manufacturer move to an India capability centre?

   The cleanest transfers are CAD, CAE and simulation, embedded and control software, design documentation and PLM management, and finance, HR and procurement shared services. Manufacturing engineering and process planning transfer with periodic physical access to the line. Supplier quality engineering transfers where the supplier base is regional. Prototype build and physical testing transfer only with capital equipment and the compliance obligations that follow. Certification and homologation are often jurisdiction-bound. Plant operations and line supervision do not transfer. Most manufacturers should begin with shared services and the digital engineering scope, and treat physical capability as a funded second phase.

   China plus one: should we build a capability centre in India?

   These are two separate decisions that are often merged into one and should not be. A China plus one supply chain decision is about where production and sourcing sit, and is driven by tariffs, logistics, supplier ecosystems and customer requirements. A capability centre decision is about where engineering, design and support work sits. India can be a strong answer to both, and the Production Linked Incentive data shows meaningful movement on the production side, but the two need separate business cases with separate metrics. Where the supply chain side is the real question, the operational picture sits closer to our work on how GCCs are driving efficiency in logistics and supply chains. Building a capability centre does not de-risk a supply chain, and relocating production does not by itself give you engineering capacity.

   How is a manufacturing GCC different from an IT GCC in practice?

   The differences that matter operationally are equipment, compliance, ramp and IP. An IT GCC needs desks, network and people. A manufacturing capability centre with physical capability needs capital equipment on the critical path, becomes a regulated workplace under the Occupational Safety, Health and Working Conditions Code, 2020 once it has a workshop or pilot line, ramps against validation and certification cycles rather than hiring velocity, and may face export control restrictions on the design data itself. The people model also differs, because contract labour and shift working feature in a way they do not in a software centre, bringing principal employer obligations with them.

   What I would say to a manufacturing leadership team

   People strategy is business strategy in execution, and that is the lens I would apply here. Organisations do not scale because they hire more people. They scale because they build the right structure, leadership capability, governance and culture to support growth.

A manufacturing capability centre tests that harder than a software one, because the failure modes are less forgiving. A software centre that is set up badly produces slower code. An engineering centre that is set up badly produces a validation bottleneck, a compliance exposure and a launch date that slips, and none of those show up in a quarterly headcount report.

So my honest advice is to stop reading the India GCC material as though it were written for you. Much of it was written about software companies by people who have never commissioned a test rig or registered an establishment. Take the parts that transfer, which are real and substantial, and rebuild the rest against your own physics.

Then start with the boring functions. Prove you can run a centre before you ask it to engineer your product.

Where organisations want help structuring that decision rather than being sold a model, that is what our GCC advisory work is for, and the people architecture underneath it is a strategic CHRO conversation from the first week rather than the first hire.

^[1][2][3][4]