India’s Manufacturing Rotation Is Underway. The Bigger Opportunity May Be One Layer Below the Factory.

Why the manufacturing rotation is the right macro call, why the execution is almost certainly in the wrong names, and where the data actually points.

In Q3 FY26, the RBI's OBICUS survey put Indian industry at 75.6% capacity utilisation. That number looks like a footnote. It isn't. When factory utilisation crosses 75% and holds there, something mechanical happens: the cost of turning away new orders exceeds the cost of building new capacity, and companies that have been waiting start committing. This is the trigger the manufacturing bull camp has been forecasting. It is now confirmed by data. The cycle is live.

The bigger question is whether investors are looking at the right part of the manufacturing value chain. 

The Government Bet You Already Know About

The Union Budget FY27 committed ₹12.2 trillion in central capital expenditure for physical infrastructure, the largest single-year infrastructure allocation in Indian history, going into roads, railways, power grids and manufacturing corridors. Sitting next to that number in the same budget is ₹1.85 trillion in long-term capital loans disbursed directly to states, 23% more than last year,We are seeing significant spends on public infrastructure creating a coordinated demand shock for steel, cement, electrical equipment and project logistics that Indian supply chains have not seen at this scale or speed. Roads and railways alone take 47% of the central allocation, so every switchgear manufacturer, every engineering firm and every project logistics company serving infrastructure is sitting directly in the path of a machine designed to spend as fast as it can.

The demand case is not speculative. It is a government spending programme with a budget attached.

The problem is what private companies are actually doing with their own capital.

Private sector capex in FY26 fell 2.8% compared to FY25, landing at ₹11.9 lakh crore. Not slowing down. Fell. And only 4 out of 18 sectors grew their investment at all. The other 14 were flat or pulling back. The broad industrial boom that every bullish note describes exists on government spreadsheets and almost nowhere else yet.

But Private Companies Aren't Following Yet

Private sector capex in FY26 fell 2.8% compared to FY25, finishing at ₹11.9 lakh crore. Not slower growth. A fall. And only 4 out of 18 sectors grew their investment at all. The other 14 were flat or pulling back. The broad industrial revolution that every market note is describing doesn't exist yet in corporate balance sheets, only in government spending plans.

The 4 sectors that grew are clear about what kind of cycle this actually is. Metals grew 32.6%, because global buyers trying to reduce their dependence on China for steel and aluminium now have India on speed dial and they're spending to meet that demand. Logistics grew 24.8%, because goods have to move no matter what the macro looks like. Power grew 15.4%, because you physically cannot run a data centre or a semiconductor fab on unstable electricity, and every company planning to build something serious knows this is the first bill to pay. Healthcare grew 1%, which barely counts but at least wasn't negative.

Power and crude oil together made up 35.7% of all private capex in FY26. Strip those two out and most of India's economy barely invested in anything. The private capex boom is a real story in four specific pockets, and a fictional story almost everywhere else right now.

The 75% Number That Changes Everything

The RBI's OBICUS survey is probably the most important quarterly survey most investors never read. It tracks what percentage of India's existing factory capacity is actually being used. In Q3 FY26, that number hit 75.6%, up from 74.3% the quarter before.

There's a principle called the 75% threshold, and it works like this: when you're running a factory at 70% of capacity, the smart move is to optimise scheduling and squeeze more output from the same equipment. Nobody calls a capital equipment supplier at 70%. But once utilisation crosses 75% and stays there for two or three quarters, the math flips. Turning away new business starts costing more than commissioning new capacity, and that's the moment capex decisions get made.

The bakery running 7 of its 10 ovens adds a baker. The bakery turning away wholesale clients every Monday because it's running 9 ovens calls the oven company. India just crossed from add-a-baker territory into call-the-oven-company territory, and that shift is mechanical, not a prediction.

Except in May 2026, something happened that complicated the whole thing.

May 2026: The Month That Should Make You Nervous

Corporate order inflows crashed to ₹39,100 crore in May 2026, down from ₹72,400 crore in May 2025. That's a 46% drop in one month, caused by global tariff uncertainty freezing multi-year contract decisions. When the person running a factory doesn't know what her export costs look like in six months because trade policy could flip overnight, she doesn't sign a three-year manufacturing order. Without those commitments, the capex cycle that 75% utilisation says should be starting doesn't start.

This tension, domestic utilisation pointing toward building while global noise freezes the signature, splits the investable universe cleanly in two. Companies needing big long-term contracts to justify new capacity get hurt. Companies one layer down in the supply chain, where the repeat business is more automatic, don't.

Almost nobody is talking about that layer.

Why The Real Money Goes to People Selling Shovels

When India builds a new semiconductor fab or a large data centre, that facility needs things nobody in the headline coverage talks about. Industrial-grade power substations with voltage stability specs completely different from anything residential. Industrial transformers that cost ₹2 crore each and typically run 15 years before needing major service. Specialised cooling systems, contamination-controlled materials handling, water treatment infrastructure, and enterprise software tying all of it together so the plant manager can see every variable on one screen. Each of those suppliers goes through a long painful qualification process once. Then they get repeat orders across every new project, without re-competing for the contract each time.

This is what Levi Strauss was doing. He didn't need to find gold. He needed every miner to need pants. In India's capex cycle, the companies making industrial transformers, precision testing equipment, industrial automation systems, specialty chemicals and the software monitoring complex factory environments don't need to win every factory contract. They need factories to exist. And once a factory has qualified a supplier into its operations, cChanging suppliers can involve lengthy requalification, testing and production-risk processes during which the production line cannot run at full speed. Nobody does that unless they absolutely have to.

The assembler wins a single contract and then has to go win another one. The bottleneck supplier qualifies once and collects for the next 15 years.

The Smartphone Number That Looks Better Than It Is

India's mobile phone exports under PLI hit ₹2.59 lakh crore in FY26, and that got enormous coverage. Some of it deserved: India assembled almost nothing for phone exports five years ago. But gross export numbers hide a problem the headlines never explain. An ₹80,000 phone exported from Chennai counts as ₹80,000 in export data, even if 80% of its components (the display, camera module, processor, battery, circuit board) were imported from Taiwan, South Korea and Japan. India may have created around ₹16,000 of actual value per device. The rest was logistics margin and assembly.

The metric that actually tells you whether India is building a real manufacturing ecosystem or running sophisticated pass-through assembly is localisation rate: how much of a product's components are genuinely made here. India has 12 approved semiconductor projects with roughly $20 billion in planned investment as of September 2026, but the equipment for those fabs is almost entirely imported, specialty chemicals are mostly imported, and the talent pool for chip design is thin enough that optimistic timelines put India's chip ecosystem in competitive shape around 2030. The investable near-term story inside electronics isn't chips. It's packaging, design services, testing infrastructure and industrial electronics, meaning the scaffolding of the chip ecosystem rather than the chips themselves.

And one more thing most rotation arguments completely miss.

IT Is Not Dead. It's Moving Where You Least Expect.

IT services are underperforming right now, and the reasons are real: global tech budgets are tight, AI is compressing the labour arbitrage model that India's IT industry was built on (hiring engineers at 30% of US salaries to do repetitive code-heavy work), and the repricing is probably not finished. But "IT is being repriced" and "IT is dying" are two different things.

The IT companies doing complex AI implementations where work genuinely requires domain knowledge and can't be templated, the ones doing cloud architecture for factories digitising their operations, the ones building industrial IoT connecting factory floor sensors to ERP systems in real time, and the cybersecurity firms protecting increasingly connected manufacturing facilities, those businesses are growing inside the manufacturing boom in a way nobody in the rotation narrative predicted.

Some IT firms will end up selling the operating systems of India's new factories. Some industrial companies will be valued like software businesses because their maintenance contracts generate recurring revenue that equity markets have always paid a premium for. The boundary between those two categories is blurring, and the most interesting risk-reward in this cycle probably lives right at that line.

Just don't let the narrative carry you past the most expensive mistake in every capex cycle in history.

The Trap That Gets People Every Cycle

One thing you really should not do: pay 50x earnings for a company generating 8% return on invested capital because the sector story sounds good. That's a story bet, and story bets in capex cycles are how retail investors fund the institutional exits. The warning signs are consistent across every cycle: capex announced with no funding details, customer concentration where one or two clients make up more than 60% of revenue, receivables growing faster than revenue (clients aren't paying on time), repeated equity issuances that dilute you every time the stock runs up, and order books full of framework contracts that are intentions rather than signed purchase orders with money attached.

Five things worth tracking before buying anything in this cycle: first, order-to-revenue conversion within 18 months, because an order that doesn't convert is just a press release; second, company-level utilisation and not sector averages, because the sector can be at 80% while your specific company runs at 55%; third, net debt to EBITDA below 3x, because a leveraged company in a delayed cycle can run out of time; fourth, domestic value-added percentage for any PLI beneficiary, because the localisation rate is what separates real manufacturing from pass-through assembly; and fifth, whether the valuation multiple is supported by actual return on invested capital or purely by the story the market is telling itself.

The Actual Bet

India's manufacturing rotation will be one of the defining investment themes of this decade. That's almost certainly true. But the companies that define it are probably not the ones in every analyst note right now.

They're the ones building the things that every factory needs to exist, qualifying their way into a recurring revenue stream that nobody has to be smart about once they're in, collecting their 15 years of repeat business while headline companies fight over the next contract win. The investors who get this cycle right will be tracking cash conversion rates, capacity utilisation at the individual company level, localisation percentages as a leading indicator of whether electronics is becoming real manufacturing or staying sophisticated assembly, and which IT companies are quietly winning industrial software contracts that nobody in manufacturing coverage is looking at yet.

Not "sell IT, buy factories."

Sell the story. Buy the bottleneck.