India Funds Its Startups. Why Won't It Buy From Them?

India has spent a decade asking its young people to build. It created funds, grants, incubators, tax incentives, credit guarantees, challenges and accelerators. More than two lakh startups have been recognised.

But when those startups finally build something that works, they run into a far harder question: who will buy it?

The Government e-Marketplace handled purchases worth ₹5.03 lakh crore in FY26. Startups received just over ₹19,000 crore, less than four per cent. Fewer than one in five recognised startups is even registered as a government seller.

This is the contradiction at the heart of India's startup story. We have built a state that is increasingly comfortable funding innovation and remains deeply uncomfortable buying it.

And in energy, healthcare, climate technology, mobility, defence and infrastructure, the government is not merely one possible customer among many. It is often the only first customer large enough to create the market at all.

The number that explains everything

The gap is not evidence that startups build things the government does not need. It is the predictable output of policy design.

Micro and small enterprises cross 47 per cent because a mandate exists. Every central ministry, department and public sector enterprise must source at least 25 per cent of annual procurement from MSEs, with carve-outs of four per cent for SC/ST-owned units and three per cent for women-owned units. Across CPSEs and departments in 2024-25, MSE procurement reached ₹93,017 crore, or 43.58 per cent, comfortably above the floor.

Startups are the only priority category in Indian public procurement that received relief on the entry side and nothing at all on the buyer side. Rules 170 and 173 of the General Financial Rules exempt recognised startups from earnest money deposits and allow relaxation of prior turnover and prior experience. No buyer was ever given a target, a budget line or a reporting obligation to actually purchase.

India changed the rules of entry. It did not change the incentives of the buyer. The result shows in the only long-run figure available: the Standing Committee on Commerce found that ten per cent of recognised startups had transacted on GeM at all, worth ₹14,000 crore across eight financial years.

The room where the problem becomes visible

On 19 August, in an auditorium at IIT Madras, Petroleum Secretary Neeraj Mittal made a promise to a room full of deep-tech founders. Around thirty energy startups would each become eligible for up to ₹2 crore in milestone-linked convertible funding under a new accelerator, MC2+ Ignite, plus pilot sites inside the research centres of the oil and gas PSUs. They would also get visibility that a PSU will procure their product when it is ready. They should not, he said, have to worry about whether a market exists.

The chairmen of ONGC, Oil India, Indian Oil, BPCL, HPCL and Engineers India were in the room. Between them they command one of the largest procurement budgets in the Indian economy; ONGC alone spent roughly ₹62,000 crore on capital expenditure in FY25.

And yet the way to promise thirty founders a customer was to build an accelerator.

That is not a criticism of the programme, which is serious institution-building. It is an observation about the shape of the problem. MC2+ Ignite's own framing says it plainly: early-stage grant funding is easy to come by, while startups struggle to access pilot sites, industry customers and growth capital. Early-stage funding is no longer the only, or even the defining, constraint. Access to customers is.

Two stacks, one of them missing

Over the last decade India assembled arguably the most elaborate public capital stack for startups anywhere outside the United States and China. The Fund of Funds for Startups has mobilised over ₹25,500 crore into more than 1,370 startups through 145 alternative investment funds, and was topped up with another ₹10,000 crore in 2026. The Seed Fund Scheme added ₹945 crore across roughly 300 incubators. The Research, Development and Innovation Fund carries a corpus of ₹1 lakh crore. Layered on top are a credit guarantee scheme, sixty-plus regulatory reforms, tax exemptions, and a dedicated fund inside almost every ministry and PSU. In 2023 the Standing Committee counted 42 ministries, departments and bodies running startup schemes.

What India did not build is a public demand stack. On the buying side there is no target, no dedicated procurement route, no reporting line, and until recently no attempt to change the contract a startup is asked to sign.

What a standard contract does to a young company

The mechanism is in the fine print, and it is the part founders raise first and policy documents mention last. ONGC's standard supply conditions apply liquidated damages at 0.5 per cent per week of delay, up to ten per cent of contract value, and the corporation may require that maximum to be secured by an irrevocable bank guarantee. A performance guarantee sits on top.

Consider an illustrative case: a four-year-old company with ₹6 crore in revenue wins a ₹5 crore order. Ten per cent is ₹50 lakh of exposure, bank-guaranteed, which means margin money locked at a bank. Add the performance security and the working capital gap of a payment cycle, and the order can consume more cash than it generates long before the first invoice is settled.

None of this is malice. A liquidated damages clause is a genuine pre-estimate of the buyer's loss, a principle Indian courts have upheld since ONGC v. Saw Pipes in 2003. For a mature supplier delivering a known product on a known schedule, a ten per cent cap is prudent use of public money. The difficulty is that the identical template is applied to first-of-a-kind technology, where schedule risk is higher and the supplier's balance sheet thinner. One template, two entirely different risk profiles. Defence conceded this in 2025. Civilian procurement has not.

The exemptions that do exist address the wrong end of the process. Waiving turnover and prior experience gets a startup into the tender. It does nothing about qualification criteria written around a reference installation the company cannot possibly have, nothing about the guarantee burden, and nothing about the asymmetry facing the person signing.

Approving a grant through an established programme is often institutionally safer than awarding a purchase order to a three-year-old company, which invites an audit query and a challenge from the incumbent supplier. That asymmetry, far more than any hostility to young firms, explains why so many arms of the Indian state chose to become investors rather than customers. The proliferation of PSU startup funds is not a solution to procurement rigidity. It is a symptom of it.

The natural experiment nobody has read as one

There are two sectors where Indian startups have unmistakably broken through in the last five years, and they are the two sectors where the state changed how it buys rather than only how it funds.

Defence began with money. Innovations for Defence Excellence, launched in 2018, put grants behind startups solving service problem statements. But the decisive move was procurement. The Defence Acquisition Procedure 2020 created a category through which the services can buy iDEX-developed products without competitive re-tendering, converting a prototype into an order. Then in 2025 the Defence Procurement Manual rewrote the contract itself: liquidated damages capped at ten per cent and applied only for inordinate delay, cut to 0.1 per cent per week for indigenisation projects, waived entirely during development, and assured orders for five years, extendable by another five.

Space followed the same logic through different instruments. IN-SPACe became a single-window regulator and ISRO was directed to transfer mature systems to industry. In February 2026 the SSLV technology went to Hindustan Aeronautics Limited in a ₹511 crore, ten-year arrangement. In July 2026, Skyroot's Vikram-1 successfully reached orbit, marking India's first privately developed orbital rocket launch. Asked what the sector still needs, IN-SPACe chairman Pawan Goenka answered in a line: the government has to be an anchor customer. He had numbers behind him. The Department of Defence has ordered 31 satellites from private companies, with ISRO expected to build another 21.

India already knows how to fix this

The most encouraging part of this story is almost entirely unreported. Several states have already built the instruments the centre lacks.

Kerala's live dashboard deserves to be read honestly rather than celebrated: 141 startups, 147 departments, 264 work orders, worth ₹28.05 crore. India's most advanced state-level startup procurement mechanism, running for most of a decade, has moved ₹28 crore. It proves the machinery can be built. It also proves machinery alone does not create demand. Telangana's April 2018 order, meanwhile, is a working civilian equivalent of what the defence ministry built, complete with deemed-approval clocks to stop files from ageing, and it has existed for eight years.

These are not theoretical recommendations sitting in a consultant's presentation. Different parts of India are already testing them. The missing step is to measure what works and take it national.

What India copied, and what it left behind

India's grant programmes are often described as modelled on America's Small Business Innovation Research scheme, and the resemblance is real. Phase I and Phase II, the feasibility grant and the prototype grant, were copied faithfully across iDEX, the Seed Fund Scheme and a dozen PSU programmes.

What was not copied is Phase III, the part that makes SBIR a procurement programme rather than a grant programme. Under US law, a federal agency may award a Phase III contract to an SBIR firm on a sole-source basis, with no cap on value, duration or number, and any agency may do so, not only the one that funded the original research. Competition is deemed to have occurred at the proposal stage.

The 1% Purchase Order Mission

Four changes, none of which needs new money.

One. Earmark at least one per cent of central procurement for DPIIT-recognised startups, with a public reporting line. The MSE experience is the proof of concept: a floor plus a dashboard produced 43.58 per cent against a 25 per cent requirement.

Two. Allow a successful government-funded pilot to convert into an order without restarting the entire tender process. The template already exists twice over, in the Defence Acquisition Procedure and in Telangana's 2018 order.

Three. Create startup-specific contract terms for first-of-a-kind technology. If the Defence Procurement Manual can cut liquidated damages to 0.1 per cent per week and waive them entirely during development, the Manual for Procurement of Goods can carry an equivalent chapter.

Four. Publish ministry-wise and PSU-wise startup procurement every quarter. Nobody today can say what central public sector enterprises buy from startups outside GeM, because nobody collects it. A category that is not counted will not be managed.

A 1 per cent startup procurement target on FY26 GeM volumes would translate into approximately ₹5,000 crore in annual demand, without creating another subsidy or funding scheme. It is payment for products and services the government already needs.

The instruments already exist

Nothing above needs to be invented. DPIIT's notification of February 2026 created a Deep Tech Startup category with recognition for up to twenty years, finally matching the recognition window to the sales cycle of the companies most likely to sell to the government. Railways has relaunched its startup engagement as the Rail Tech Policy 2026. NITI Aayog has produced a discussion paper on adapting the iDEX framework to civilian sectors, and iDEX signed an agreement with EdCIL in 2025 to extend the model beyond defence. MC2+ Ignite has placed pilot sites inside PSU research centres, which is the step immediately preceding a purchase order. The task is migration rather than invention.

India spent its first startup decade teaching its young people to build. It created the grants. It created the incubators. It created the funds. It celebrated the founders.

The next decade must answer a harder question: will India trust them enough to buy what they build?

A country does not create great technology companies only by financing experimentation. It creates them by becoming the first believer, the first reference and, when the product works, the first customer.

Defence has used the instruments. Space has demonstrated them. The states are testing them. Now one of the world's largest government procurement systems can become the world's most powerful market for Indian innovation.