For many deeptech companies, the Indian state is the most logical first customer. It runs the railways, the power grid, the courts, the defence establishment, public hospitals and the largest fleet of public infrastructure in the country. It buys sensors, software, drones, diagnostics, communication equipment and analytics at a scale few private buyers can match. And it has, over the past decade, rewritten a surprising number of its procurement rules to let young companies in.
Yet government procurement for deeptech in India is still widely misunderstood by founders. Some assume the state only buys from large incumbents with ten years of audited turnover. Others assume that registration on a portal leads naturally to orders. Neither is true. The rules now offer startups real relaxations, but each route to a government contract has its own logic, timeline and failure mode, and the company that chooses the wrong one can spend a year in a process that was never going to produce a purchase order.
This article sets out the rules that matter, compares the main procurement routes open to a deeptech venture, and offers a practical reading of when each route fits.
Why the state is a natural first buyer for deeptech
Three features of public demand suit deeptech in particular. Public agencies often face problems with no commercial off-the-shelf answer: monitoring a long rail network, digitising court records, screening at scale in public health. They value domestic sourcing for strategic reasons. And a reference from a government buyer carries weight with other public bodies, with regulated private buyers and with investors.
The demand is real. The courts are a clear example. The Cabinet approved eCourts Phase III in September 2023 with an outlay of ₹7,210 crore, and a Lok Sabha reply in February 2026 put pending cases at 4.77 crore in district courts and 63.66 lakh in the High Courts. PanScience reports that its venture Nyaay AI, which works on judicial workflow automation, has deployments in more than 14 Indian High Courts. That is company-reported, and it illustrates a broader point: where a public institution has a large, measurable workload and a modernisation budget, a specialised technology company can find a buyer.
The marketplace is also large. A PIB backgrounder dated 28 February 2025 reports that the Government e-Marketplace (GeM) handled orders worth ₹4,03,305 crore in the previous financial year and ₹4,52,594 crore in the financial year to that date, with micro and small enterprises taking 37.87% of order value. Startups had fulfilled orders worth ₹35,950 crore on the platform.
The rules that open the door
Four sets of rules shape what a startup can and cannot do.
Relaxation of prior turnover and experience. The Department of Expenditure's Office Memorandum of 20 September 2016 directed central ministries and departments to relax prior turnover and prior experience criteria for startups, whether or not they are micro or small enterprises, subject to meeting quality and technical specifications. The memorandum was also sent to the Department of Public Enterprises to instruct central public sector enterprises to follow it. It notes an exception: for items such as public safety, health or critical security equipment, a procuring entity may retain experience requirements with adequate justification. A Department of Expenditure clarification of 29 June 2020 confirmed that the relaxation applies to all DPIIT-recognised startups. The Startup India portal lists the current provisions as Rule 173(i) of the General Financial Rules 2017 for goods, Rule 1.9(ix) of the Manual for Procurement of Consultancy and Other Services, and Rule 4.5.2 of the Manual for Procurement of Works.
Earnest money exemption. Under Rule 170(i) of the GFR 2017, DPIIT-recognised startups are exempt from paying earnest money deposits. Several states have their own versions; the Startup India portal notes, for example, that Odisha reduced the performance guarantee for eligible startups to 25% of the normal amount.
Preference for local suppliers. The Public Procurement (Preference to Make in India) Order, 2017, as revised in June 2020, classifies suppliers by local content. ELP's summary of the revised order sets out the categories: Class-I local suppliers have local content of 50% or more, Class-II between 20% and 50%, and non-local suppliers 20% or less. Class-I suppliers receive purchase preference: in a non-divisible order, the lowest Class-I bidder within the margin of purchase preference is invited to match the lowest price. Suppliers self-certify local content at the tender stage, and for procurements above ₹10 crore they must provide a statutory auditor's certificate. A September 2020 amendment, reported by Shardul Amarchand Mangaldas, allowed nodal ministries to set minimum local content above 50% and raise the margin of preference beyond 20%.
Limits on global tenders. Rule 161(iv) of the GFR 2017, amended in May 2020, provides that no global tender enquiry shall be invited for tenders up to ₹200 crore, except with approval at the level set out in the Department of Expenditure's consolidated instructions. For a domestic deeptech venture, this removes much foreign competition from the large majority of contracts.
Together these rules mean a recognised Indian startup with genuine local content can, in principle, bid for central government contracts without a turnover history, without earnest money, and with preference over imports. The phrase "in principle" matters. The relaxation is conditional on meeting technical specifications, and specifications are often written around incumbent products.
Government procurement for deeptech in India: four routes to a contract
The table below is this article's original contribution. It compares the four practical routes by which a deeptech venture reaches a government buyer, across the dimensions that determine whether the route will work for a given product and stage.
GeM and the Startup Runway
GeM is the default route for goods and many services bought by central ministries and agencies. The GeM Startup Runway, launched on 19 February 2019 according to PIB, lets DPIIT-recognised startups list innovative products that may not fit existing categories, run market trials with government buyers, and collect time-bound feedback. PIB described the mechanism at launch: once an innovative product receives ratings of 3 or more on a 1 to 5 scale from at least three buyers, it is listed as a regular GeM item. The PIB backgrounder of February 2025 describes a Startup Runway 2.0 category in which startups can list products and services whether or not they hold DPIIT certification.
The weakness of GeM for deeptech is that it is a marketplace, not a sales channel. A listing does not create demand. Founders who succeed on GeM usually have identified the buying officer first, persuaded them of the need, and then used GeM as the transaction mechanism.
Open tenders
Larger or more complex requirements go to tender, typically through the Central Public Procurement Portal or an agency's own e- procurement system. The startup relaxations apply, but tenders are evaluated against technical criteria and, in most cases, awarded to the lowest compliant bidder. A deeptech venture with a better but more expensive solution often loses on L1 price unless the specification captures its advantage. The most effective tender work therefore happens before the tender, in shaping the requirement through demonstrations, pre-bid queries and technical meetings.
Innovation and challenge procurement
Several states and agencies run challenge-based procurement, where a department publishes a problem and selects startups for trials, often with a work order attached. The Startup India portal lists state provisions with very different scales: Andhra Pradesh's policy allows up to ₹50 crore a year of selected products and solutions, with work orders of up to ₹5 crore each, while Kerala's direct procurement allows ₹5 lakh to ₹20 lakh per startup. Haryana treats startups like micro and small enterprises in procurement, and startups quoting within L1 plus 15% may be eligible to execute the contract if they match the lowest price.
The defence sector has its own route. Under the Make-II procedure, the Startup India portal notes that HAL allocates projects with estimated cost up to ₹250 lakh to startups, and BEL offers prototype development projects up to ₹10 lakh with procurement up to ₹5 crore.
The risk in this route is the gap between pilot and purchase. A trial funded from an innovation budget does not automatically create a line in the department's regular budget. Ventures should ask, before starting, which budget would pay for a scaled purchase and who controls it.
PSU and agency partnerships
Central public sector enterprises and technical agencies are often the most sophisticated public buyers of deeptech, and some have dedicated budgets for startup collaboration; the Startup India portal records ₹320 crore allocated by the Ministry of Petroleum for startup collaboration. These relationships usually begin as joint development or pilot projects and can become long-term supply. They also take the longest. A separate PanScience Research piece covers how public sector undertakings can build and buy new technology within procurement and corporate venture capital rules.
Choosing a route: a practical reading
The comparison suggests a simple sequence for most deeptech ventures.
If the product is mature, specifiable and priced, start with GeM and target identified buyers directly. If the product solves a problem the buyer has not yet specified, start with challenge procurement or a PSU pilot, and use the result to shape a later tender. If the opportunity is a large system contract, assume a tender, and invest early in pre-tender engagement and in partnerships with larger system integrators who can meet scale requirements.
Five practical points apply across all routes.
1. Get recognised and documented. DPIIT recognition unlocks the GFR relaxations and EMD exemption. A clear local content calculation, and an auditor's certificate where required, protects Make in India preference. 2. Read the specification exceptions. Safety, health and critical security items may legitimately retain experience requirements. A venture in those areas should plan to partner or to build a track record through smaller orders first. 3. Price for L1. In most tenders the lowest compliant price wins. If the product's value lies in lifecycle cost or outcomes, argue for that in the specification stage, not after the bid. 4. Plan for payment cycles. Government buyers can be slow to pay. Working capital should be planned as if receivables will take longer than contract terms state. 5. Turn pilots into budget lines. Every pilot should end with a written evaluation and a recommendation that the buyer can attach to a procurement proposal.
What the state could do better
The rules have moved further than practice. Relaxations exist on paper, but procuring officers still face audit risk for choosing an unproven supplier, and that risk often outweighs any instruction to relax criteria. Two changes would help more than further relaxations. The first is a predictable path from a successful innovation trial to a regular procurement, so that a pilot evaluated as successful can be scaled without a fresh tender designed around incumbents. The second is publication of procurement projections, which the September 2020 amendment to the Make in India order requires for ministries and departments with procurement above ₹1,000 crore a year. Consistent publication of those five- year projections would let deeptech ventures plan products against public demand.
Conclusion
Government procurement for deeptech in India is more open than most founders assume. DPIIT-recognised startups can bid without turnover or experience history, without earnest money, and with preference over imports for most contracts below ₹200 crore. GeM, open tenders, challenge procurement and PSU partnerships each offer a way in.
But each route fits a different product and stage, and each fails in a predictable way. Listings do not create buyers, tenders reward the lowest compliant price, pilots do not create budgets, and partnerships take years. The ventures that succeed with the state are the ones that pick the route that fits their product, shape the requirement before it is published, and treat every pilot as the first step of a purchase rather than the end of a trial.
Methodology and sources
This article combines official procurement rules and data with an original comparison of procurement routes. The route comparison table and the practical sequence are PanScience Research analysis; ticket sizes and timings in the table are qualitative descriptions, not measured averages. Relaxation rules are from the Department of Expenditure OM of 20 September 2016 (startupindia.gov.in/content/dam/startupindia/homebanners/Relaxation_of_Norms_Startups_Medium_Enterprises_PublicProcurement.pdf) and its 29 June 2020 clarification (doe.gov.in/files/circulars_document/Request_clarification_provision_startups_MSMEs.pdf). GFR rule numbers, state provisions and Make-II figures are from the Startup India public procurement page (startupindia.gov.in/content/sih/en/public_procurement.html). Make in India order definitions are from ELP's summary of the revised order (elplaw.in) and the September 2020 amendment from Shardul Amarchand Mangaldas, 21 September 2020 (amsshardul.com). The global tender limit is from the Department of Expenditure's consolidated instructions of 30 June 2021 (doe.gov.in/files/circulars_document/Consolidated_instructions_Global_Tender_Enquiry_under_Rule_161_iv_2017.pdf). GeM Startup Runway details are from PIB, 19 February 2019 (pib.gov.in/PressReleasePage.aspx?PRID=1565237), and GeM order data from the PIB backgrounder of 28 February 2025 (static.pib.gov.in/WriteReadData/specificdocs/documents/2025/mar/doc202532511101.pdf). eCourts and pendency figures are editor-verified from PIB and a Lok Sabha reply of 13 February 2026 via IANS. The Nyaay AI deployment figure is company-reported by PanScience.