To win the deep-tech race, India needs ‘General Innovation Rules’

India has drawn a bold new blueprint to compete at the frontiers of science. Recent initiatives like the Anusandhan National Research Foundation (ANRF), the Rs. 1-lakh-crore Research Development and Innovation Fund (RDIF), and several deep-tech missions exemplify a commitment to fostering a domestic innovation ecosystem. This strategic ambition is welcome, but only the right financial control regime will make it effective.

The current regime, which includes procurement, fund release and accounting rules, rests on the ‘General Financial Rules (GFR), 2017’ and the ‘Hybrid Treasury Single Account (TSA) model’. The GFR governs public expenditure where outputs are usually specified ex-ante and risks are bounded.

Also read: India can win in AI deployment despite lagging in advanced chips: Neelkanth Mishra


For instance, for goods procurement, it mandates competitive tenders, selection via ‘L1’ or ‘QCBS’ mechanisms, and scheduled payments. The Hybrid TSA guides the payment process under the ‘just-in-time’ principle, ensuring funds are drawn precisely when required to avoid idle balances. This applies to all appropriated funds, granting legal authority to eligible entities to spend up to a ceiling, lapsing at year-end.

This regime works well for conventional government procurement but is structurally misaligned with scientific research funding. Forcing a rulebook designed for building highways to fund quantum research produces perverse incentives. R&D is probabilistic; its outputs cannot be predefined. Accountability must rest on whether the right questions were asked, acknowledging non-linear timelines and unpredictable pathways. To resolve these frictions, India needs a dedicated financial control regime for R&D – what we call the ‘General Innovation Rules’. Three principles should guide this regime:

Firstly, a portfolio approach to undertake high-risk, high-reward R&D. The current project orientation of public R&D generates adverse selection; researchers facing individual scrutiny favour safe, incremental proposals. They choose easily achievable metrics and de-emphasise failed experiments due to perceived reputational risks. For programme managers, this approach forecloses the productive frontier strategy of funding multiple competing approaches. A portfolio approach, on the other hand, accepts failure as a natural outcome of R&D. By not flattening different innovation problems, it allows the deployment of diverse contracting instruments beyond conventional grants to accurately capture risk profiles. DARPA has operated precisely this way since 1958: its programme managers simultaneously fund multiple competing approaches and treat project-level failure as evidence of a functioning selection process. The internet, GPS, and mRNA vaccine platforms emerged from this logic.

Secondly, a doctrine that redefines “public benefit” even if private actors are involved. The reluctance to direct public grants to private entities emanates from a sound theory—public expenditure should produce public benefit, not private profit—but is misapplied in R&D, where the recipient’s identity has become the primary concern. At the translational stage, the most capable performers are typically private actors possessing the required engineering design, rapid iteration capacity, and commercial discipline. Workarounds such as routing funds through intermediaries only create principal-agent problems or a proliferation of Section 8 companies. Israel’s Innovation Authority provides an instructive counterpoint: it operated conditional grants directly to private firms for decades, with repayment triggered by commercial success, helping sustain a high-technology sector that drives Israel’s export competitiveness. Equally, private co-investment is disincentivised. There is no standard mechanism to co-deploy sovereign and private funds within public structures. This is a significant opportunity foregone – Innovate UK’s SBRI Healthcare programme, which issues challenge-based contracts directly to private firms, has leveraged over £680 millions in private investment against £140 millions of public funding.

Thirdly, a milestone-based disbursement framework supported by multi-year commitment guarantees. The ‘Utilisation Certificate’ codifies the tendency to measure inputs over outcomes, tying future funding to prior expenditure rather than scientific progress. This structural rigidity penalises scientifically correct pivots as administrative irregularities. Furthermore, Hybrid TSA’s annuality constraint—where unspent funds lapse—creates a “use it or lose it” pressure. The India Ease of Doing Science Index 2023 found ease of fund utilisation was the lowest-rated research environment dimension for Indian researchers, 31 points behind global counterparts. Because the current funding architecture cannot guarantee allocations beyond a single year, it disfavours retaining long-term teams, contravening a foundational requirement of good R&D: tacit knowledge development.

The government has recognised this, making eleven bounded exemptions to the GFR for scientific ministries and agencies. The ANRF Act, 2023 ameliorates this structurally by explicitly providing for rules derogating from GFR. The operating rules for RDIF were framed under this authority. However, these relaxations remain tethered to the conventional procurement doctrine that the GFR epitomises, with the incompatibility remaining acute in early-stage, grant-based research.

It is time public R&D is accorded different treatment. India’s tax framework already does this – Section 45 of the Income Tax Act, 2025 provides weighted deductions for R&D expenditure, recognising its public good character. That logic has yet to reach the expenditure control side. International practice could point the way for Indian lawmakers. In the UK, the ARIA Act 2022 established a parallel regime for R&D that explicitly disapplied standard UK procurement law. Similarly, Part 35 of the Federal Acquisition Regulation of the US established a dedicated contracting regime for R&D.

Also read: Silicon Valley’s South Park Commons ramps up India deeptech push

Deep-tech R&D is a race of capital deployment between nations. India already has the enabling architecture – the ANRF Act’s explicit provision to derogate from GFR are a functioning pilot of the principle. The evidence these produce should inform a statute of general application, i.e., the ‘General Innovation Rules’ which treat R&D as categorically distinct public expenditure. This new regime should not abandon accountability but redefine it in keeping with the fundamental nature of R&D itself.

(Authors: Varun Agarwal is Co-founder, FAST India and Change Engine; Rohit Saxena is a public policy professional)