India has one visible skilling system and two invisible ones. Policy responds to the visible one.

The Demographic Dividend — Piece 3 of 9

Every number you've seen on India's skill gap was drawn from a survey the IT sector commissioned.

TCS needs to know which campuses produce hirable engineers — so it funds the data. NASSCOM tracks employability because its members hire at scale. Mercer-Mettl runs the Graduate Skill Index because the industry has money riding on the answer. The India Skills Report 2026 puts graduate employability at 56.35%. The Economic Survey says 51%. Mercer-Mettl says 42.6%. The spread matters less than the reason the data exists at all: the sector has skin in the outcome.

Here is what nobody commissioned: the equivalent survey for manufacturing. Or for agri-processing.

India's three economic engines run on three completely different skilling infrastructures. Policy sees one of them clearly — because that one made its gap legible. The other two remain unmeasured. And policy, as a general rule, responds to what it can measure.

The visible system: IT and GCC services

The IT sector's skilling system is visible because failure is commercially expensive. An engineer who cannot perform is a quarterly revenue problem. A campus that produces unemployable graduates stops getting hired from. The feedback loop closes fast.

TCS reskills 100,000 employees annually — approximately one-sixth of its global workforce. Infosys trained 275,000 employees and is bringing 20,000 freshers on board in FY26 with structured AI and digital reskilling built into onboarding. Wipro has put 44,000 employees through AI training. These are not CSR initiatives. They are commercial necessities dressed in corporate communications language.

The government's skilling system works on a different accountability model. PMKVY — the Pradhan Mantri Kaushal Vikas Yojana, India's flagship short-term skilling scheme — reported 54% placement across its training programmes. Independent analysis of the actual data found 22.16% of trained candidates were placed. These are not the same intervention: TCS reskills existing employed professionals at commercial risk; PMKVY trains unemployed first-time job seekers at government expense. The populations and the stakes are different. But the accountability gap is the same point. One system measures trained. The other measures employed and performing — because only one has a revenue line that depends on the answer.

The result is that IT's skill gap is legible, cited, and policy-responsive — and that legibility translates directly into where government money goes. India's skilling budget in Union Budget 2026-27 is ₹9,885 crore — a significant increase from ₹4,520 crore in 2024-25. TCS alone reskills 100,000 employees annually — one company, one sector. The government's budget covers the entire economy. It is calibrated, in practice, to the one engine whose numbers are visible.

The skill gap appears to be 40-50% because IT made it measurable. We do not know what the skill gap is in manufacturing or agri-processing. No institution in either sector has a commercial reason to find out.

The invisible system: Manufacturing

India's largest manufacturers have been quietly solving their skill problem for decades. The solution works — inside the company that built it. The moment a trained worker leaves, or a smaller manufacturer tries to hire that calibre of worker from the open market, the solution disappears. It doesn't travel. And because it doesn't travel, the majority of Tier 2/3 manufacturers — the MSMEs with 150 to 500 workers that are the actual backbone of every industrial cluster — have no access to it at all.

This is what I mean when I call it a Darwinian system. The strongest manufacturers built their own training ecosystems. Their workers perform. Their attrition is lower. Their quality benchmarks are closer to global standards. But the knowledge stays inside the company. Workers who complete these programmes have no transferable certification that another employer can verify on a resume — because the skills were built to that company's processes, not to any portable national standard. An MSME manufacturer in Coimbatore cannot hire a Bosch-trained technician and know what that means. The large companies thrive. The rest make do with whoever the open market or the ITI system produces. That is the Darwinian part — and it is exactly what makes this gap invisible to policy.

Here is what the system looks like inside the companies that built it. Bosch runs a full-fledged Vocational Centre where apprentices enter after tenth grade, spend a year in basic training across all trades, then two years in specialised in-plant advanced skill development. The program produces workers who operate to global precision manufacturing standards. Bosch calls them "multi-skilled technocrats." That is not branding — it is a curriculum designed to produce exactly what a high-precision engineering plant needs.

Bajaj Auto has committed ₹350 crore over three years to its BEST initiative — Bajaj Engineering Skills Training — setting up state-of-the-art laboratories at partner engineering colleges covering Robotics, Automation, Industry 4.0, and Smart Manufacturing. The programme, scaling to 15 centres across India, targets first-generation graduates from Tier 2, 3, and 4 cities specifically. Mahindra runs apprenticeship programs aligned to its own manufacturing standards. TVS and Toyota built the Auto Machinist Technician course through the Automotive Skill Development Council. L&T runs nine Construction Skills Training Institutes that have trained over 3.4 lakh workers across 30 years — one of India's most sustained private vocational commitments in construction. Hero MotoCorp has its own training infrastructure.

I find this pattern across manufacturing engagements consistently: companies that built their own training systems have workers who are measurably more productive, have lower attrition, and operate closer to global quality benchmarks than companies drawing from the open market or the ITI system.

But the three structural reasons this stays Darwinian rather than becoming systemic are worth naming precisely.

First, company-specific skills have no shared credential. The manufacturing sector has no equivalent of the AWS or Azure certification that travels with an IT worker across employers. A Bajaj BEST certificate means something inside Bajaj's supply chain. It is less legible to a manufacturer in Tiruppur who didn't design the curriculum.

Second, because skills are framed to each company's processes rather than a transferable national standard, the best-trained manufacturing workers are paradoxically the least portable — their knowledge is deepest inside the system that built them, and least legible outside it.

Third, and most critically, only companies with revenues in the thousands of crores and dedicated HR infrastructure can run these systems at all. The MSME manufacturer in Rajkot with 200 workers, the garment exporter in Tiruppur with 500, the auto component supplier in Aurangabad with 150 — these are the majority of manufacturing employers in Tier 2/3 India. None of them can build a Vocational Centre. They hire from the open market, find the workers inadequately trained, develop them informally, and lose the best ones to larger companies the moment they are worth poaching.

This is the Darwinian part in practice: the strong survive and build their own ecosystems. The rest churn. And because the MSME cluster — which is 65% of Tier 2/3 manufacturing hiring — generates no measurable data on its skill gap, the gap exists entirely outside the policy system's field of vision.

Manufacturing's best companies solved the skill problem for themselves. The solution did not scale. Policy never noticed, because the gap was never made legible.

The absent system: Agri-processing

Agriculture has 43-46% of India's workforce. Agri-processing — food manufacturing, cold chain logistics, dairy processing, fisheries, food safety compliance — is where that workforce has the most viable path to formal, higher-productivity employment without requiring relocation. The transition from farm to processing plant is shorter in distance and social disruption than any other formalisation pathway for rural India.

The skilling infrastructure for that transition is effectively absent. What exists is designed for a different level of the workforce entirely. NIFTEM — the National Institute of Food Technology Entrepreneurship and Management — exists to produce graduate-level food scientists and food technologists. It is a postgraduate institution producing researchers and managers. It does not produce the processing line supervisor, the cold chain operator, the food safety technician, the dairy plant worker that agri-processing expansion requires at volume.

The ITI system has food processing tracks. They are not updated to modern food safety standards, not aligned to export certification requirements, and not available at the density needed in the agricultural districts where the workers actually are.

The companies that solved this did it alone and invisibly. ITC Agri Business built its own farmer and processor training infrastructure — partly through its well-documented e-choupal network, partly through supply chain development programs that doubled as capability-building. Suguna Poultry runs its own technical training for contract farmers. These companies solved the problem the way large manufacturers solved theirs: internally, quietly, and in a way that doesn't scale beyond their own supply chains.

The consequence is that India's largest workforce sits adjacent to one of its fastest-growing industrial opportunities — agri-processing, valued at over ₹26 lakh crore and growing — without a credible pathway to enter it formally. The workers who make this transition do so through individual enterprise and informal networks. There is no system that could replicate the journey at scale.

What this means for the three engines

The three-engine bet on Tier 2/3 India depends on a skilling system that can deliver three completely different kinds of workers: vocationally trained technicians for manufacturing, processing-competent workers for agri-processing, and graduate engineers with applied skills for GCC services.

What exists is: one sector with a visible, self-correcting, commercially-driven system; one sector with an invisible, Darwinian system that serves its largest companies and abandons the rest; and one sector with no system at all.

This is not a criticism of intent. Government intends to skill India. The MSDE has been allocated ₹9,885 crore for 2026-27 — a significant increase from ₹4,520 crore in 2024-25. PMKVY 4.0 has trained over 27 lakh candidates across 38 sectors. The intent is real. The problem is information, not intention. Policy responds to what it can measure. What it can measure is what the IT sector made legible. The manufacturing gap and the agri-processing gap remain unmeasured — which means they remain, in the policy system's frame of reference, invisible.

What the measurement institutions that could fix this look like — and what a policy response built on all three engines actually requires — is the subject of a later piece in this series.

The next piece is more immediate: what MNCs and GCCs actually find when they arrive in Tier 2/3 expecting the talent the policy promised. They've been reading a map drawn by the only engine that made its gap legible. The terrain looks different on the ground.

Ashwin · freshwin.in