India's R&D system asks scientists to develop ideas and industry to deploy them. The two are not often in the same room, which causes weak collaboration when and where it matters.

Walk into the automotive lab at an engineering institute in southern India, and the first thing you notice is the engines. Typically donated by a major Indian OEM (original equipment manufacturer) towards no particular project and no embedded engineer acting as a "mentor." Just dispensable hardware tossed over a wall.

Walls. I have run into them again and again, building labs for universities and watching up close how the system around those labs fails to connect.

In India, a typical collaboration between a company and university goes as follows. The company donates equipment for a lab or writes a cheque. A professor accepts it - either as part of a standalone collaboration with the company, or to satisfy a government requirement to get funds for a research project - and issues a press release.

Then both sides return to their separate worlds. Later, the company complains at a NASSCOM panel that the universities didn't deliver anything useful and produce engineers that are not trained in the right areas. The college complains that the company never engaged.

If you think about it, both are right. The point is that the system was designed to produce just this outcome.

The wall between Indian academia and industry is not an accident of culture or a paucity of money. It is something built into the rules of the game. The most consequential of those rules, arguably, is the Technology Readiness Level (TRL) system.

Where TRL Came From, and What It Became

The TRL is a nine-point scale, originally built at the United States' space agency NASA, that ranks how close a technology is to real-world deployment - from a half-formed idea at level one to a system flying in space at level nine.

Its relevance to this discussion is simple: the way India funds research, structures collaboration, and apportions blame between academia and industry has quietly come to follow TRL's logic. So it is worth understanding where the logic came from.

In the mid-1970s, a NASA researcher named Stan Sadin built a simple scale for judging how close a technology was to being flight-ready. His TRL system originally had seven levels - from basic principles observed (TRL 1) to an actual system proven in space (TRL 7).

By 1995, John Mankins at NASA had expanded and codified it into the nine-level scale used today.

The framework did what it was supposed to: it gave engineers a shared language for risk. Is this thruster ready, or does it need more testing? TRL answered that cleanly.

Then the framework went beyond NASA.

The US Department of Defense adopted it after 2003, following a Government Accountability Office recommendation. The European Space Agency followed. The European Commission built it into Horizon 2020. ISO standardised it in 2013.

India adopted it too, and over the past decade the framework has moved from a niche evaluation tool to a structural feature of how the country funds research. India's Defence Research and Development Organisation (DRDO) uses it in project reviews. The new Research, Development and Innovation (RDI) fund - Rs 1 lakh crore, the biggest innovation-finance experiment in India's history - gates funding at TRL 4 and above.

Somewhere in this diffusion, a measurement tool became an ideology. A thermometer, if you will, for technological maturity became a prescription for how innovation should be organised.

Basic research (TRL 1-3) is academia's job. Deployment (TRL 7-9) is industry's job. And the messy middle (TRL 4-6) belongs to no one in particular. This unowned middle is often called the "valley of death" - where ideas fall to death when attempting a massive leap of progress.

India's academia-industry divide is older than TRL, of course. But TRL gave it a language, a logic, and a veneer of institutional legitimacy that made it look like the natural order of things.

Interestingly, some practitioners at the frontier have already moved past it. Rajat Sethi, co-founder of QuBeats, a Hyderabad-based quantum technology startup, calls the linear TRL approach old school with "zero significance" for how technology actually develops today.

Real work, he says, requires hopping between levels in parallel, pressing a TRL 9 product back down to TRL 4 and rebuilding the stack from a different direction.

The problem is that while practitioners are increasingly leaving TRL behind, the system around them is doubling down on it.

The Customer Mentality

The TRL framework creates a specific kind of relationship between academia and industry, and it is the wrong kind. The relationship is that of customer and vendor.

Industry sits at TRL 7 and looks at what academia produces at TRL 3 or 4 like a dissatisfied buyer. "Why would I consume this? The tolerances are wrong, the cost model doesn't work, the students who built it have graduated and nobody knows how it works."

Legitimate complaints, all of them. But they are the complaints of someone who was never involved in shaping the product.

From the other side, the picture is no better. A professor at a prominent state university described having to protect his best PhD students from being poached by a large automotive company.

The student had done good work on a problem the company cared about. The information was shared with the company as a demonstrator that the university was working on the right areas, in hope of a fruitful collaboration. The company's response was not to fund a joint programme, or embed an engineer in the lab, or co-develop the next phase. It was to simply hire the student. Extract the knowledge. Move on.

Even when engagement happens, it is often transactional. At a conference, I watched a professor from a top IIT present his lab's work like a management consultant - proud of how many industry problems he had solved on contract, and how many projects were delivered.

His lab had machines, all paid for by grants, and it solved specific problems for specific companies on demand. But a consulting model extracts knowledge without building anything durable. The company gets an answer, and the university gets revenue. But what about progressively building shared capability?

And sometimes the engagement is less than that. Large corporations give universities machines and walk away. You end up with a branded lab - a foreign multinational's name on the door, donated equipment inside - but no ongoing joint work.

When resources arrive as donations rather than as part of a working relationship, they produce dependency rather than capability.

The Patient Partner

Is it possible to have a non-transactional partnership that also generates outcomes? That is the interesting question to ask.

Some of my perspectives were shaped by observing at close quarters industry-academia relations in other countries, starting with the University of Michigan, and companies in the US and Germany.

Working at Cummins in the US, I was part of long-running engagements with Rose-Hulman - a small engineering college in Cummins's home town - and with the Southwest Research Institute and Purdue, on the same project.

None of these collaborations produced "outcomes" in the Indian sense. None of the algorithms developed made it into production. Cummins' solutions were in general more robust, whereas the university solutions offered interesting systems insights. This resulted in knowledge transfer in both directions.

Companies in the US and Europe routinely sustain collaborations of this kind for decades without ever using the research outcomes directly in production. The point is not always the output.

The point is the bridge - the ongoing channel through which talent, judgement, and frontier knowledge move in both directions. That is what Indian industry has not learned how to build.

The Empty Side of the Wall

But the problem is worse than a handoff that does not work. In many frontier domains, there is nothing on the other side of the wall to hand off to.

Even if we fix the problem of a transactional mindset, Indian academia would still be looking at a partner who has not developed the technical depth to deliver.

The standard diagnosis of the valley of death assumes both parties exist: academia develops knowledge, industry should adopt it but does not. The reality, in sector after sector, is starker.

The Indian industry partner often does not possess the frontier knowledge. This magazine has reported on this - in automotive electronic controls, in systems engineering, in the gap between assembling imported subsystems and designing them.

Indian industry has one of the lowest R&D spends in the world, resulting in a lack of research depth that makes homegrown companies deficient as the kind of partners and mentors that universities can work with.

Indeed, this lack of knowledge calls into question the whole concept of "reducing the industry-academia gap," because it assumes industry possesses product development knowledge and that it is willing to share with universities. Neither appears to be true, as partly documented in prior articles on this platform.

This goes a long way in explaining why Indian industry is unable to help with the valley of death - that is, help shape the ideas at TRL 4-7 so that they are solving the right problems, and collaborate with academia in solving them the right way.

This is not only an OEM problem, but extends to a lack of knowhow in the ecosystem that has grown totally dependent on imports for key technologies.

The global hardware-in-the-loop testing market, for instance, is entirely foreign: dSPACE, NI, ETAS, Vector, and OPAL-RT.

Embedded software is even more interesting. MathWorks makes a popular model-based development platform called Simulink that enables engineers to generate software that runs the "brains" of a car, plane, or truck - the electronic control unit, or ECU.

This platform now enjoys a monopoly (priced accordingly) in the aerospace, automotive, and other safety-critical sectors. There are dozens of Indian IT and services companies that churn out thousands of engineers who are experts in this platform, but not one of them has created a competitive product.

In quantum technology, India has brilliant professors working on atomic physics and lasers, but no homegrown laser companies that can build the components a frontier lab requires. Professors import lasers because the domestic industry to supply them does not exist.

Jay Mangaonkar, co-founder of QuPrayog, a Pune-based quantum technology startup, has seen the alternative up close. Before starting QuPrayog, Mangaonkar worked at Germany's Physikalisch-Technische Bundesanstalt (PTB), the national metrology institute.

At PTB, he experienced how Germany's precision optics and laser industry works in tight symbiosis with research institutes. Laser companies rely on PTB for development; PTB gets its first customers - early adopters who test and refine new products in real research settings. The industry and academia did not develop separately and then try to connect. They co-evolved. Each created the conditions for the other to advance.

India's situation is the inverse. When grant structures incentivise a university to partner with a large Indian OEM, and that OEM does not have the frontier knowledge, the collaboration is two parties who cannot teach each other trying to learn together.

The college will only knock on these doors, the other side of which are people who do not know how to solve problems.

The question, then, is not just how to connect academia to existing industry. In many frontier domains, it is how to grow the industry that academia can connect to - the Indian laser companies, the Indian HIL testing firms, the Indian precision electronics manufacturers that do not yet exist. And the only way to grow them, as Germany shows, is together with the research institutes, from the earliest stage.

There is a downstream consequence. When the Indian OEM is 'buy, not build,' the deep-tech startup that emerges from a university lab also has no first customer. The startup is left selling to defence labs or government procurement - important channels, but narrow and slow.

And even when a startup does build a product, India's L1 (lowest-cost) public procurement norm penalises first-generation Indian products that are inevitably more expensive than mature foreign alternatives. "There should be someone to cushion that impact," says Mangaonkar, whose own startup is building an indigenous optical atomic clock that will, at first, cost more than a Chinese or European equivalent.

Without an early-adopter mechanism - government as first customer, the way the Defense Advanced Research Projects Agency (DARPA) has operated in the US for decades - the pipeline from lab to market has no exit.

This shapes what gets funded. Venture capitalists in India have learned to back companies at TRL 7 and above, where the technology risk is mostly retired and the question is execution. The harder, more valuable work - the TRL 4-7 stretch where a university prototype becomes a manufacturable product - is where Indian VC capital is thinnest.

The absence of industry at the start of the chain produces the absence of capital in the middle of it.

Financial Rules That Keep the Wall Standing

India's public universities operate under the General Financial Rules (GFR), which govern how every rupee of government money is spent. The GFR was designed for conventional procurement - buying goods through competitive tendering, with clear deliverables and anti-corruption safeguards. Reasonable objectives when you are buying office furniture, crippling when you are trying to build a frontier research lab.

There has been progress. In June 2025, the GFR was amended to let scientific institutions procure equipment and consumables outside the Government e-Marketplace (GeM). This was a real reform, long overdue.

The Anusandhan National Research Foundation (ANRF), the new apex research funding body established under the ANRF Act 2023, also provides some operational flexibility: principal investigators can reallocate funds between consumables, manpower, travel, and contingency within the recurring budget head.

But notice what these reforms fix: the ability to buy things. Equipment, consumables, products.

What they do not fix is the ability to buy knowledge - to pay an external company for system integration, engineering consulting, or collaborative development.

Here is what this means in practice. A professor can now buy a motor, an oscilloscope, and a battery testing rig with less paperwork. What she still cannot easily do is pay an external firm to integrate those 20 products into a functioning lab - to do the systems engineering that makes the equipment useful.

ANRF grant budgets are divided into categories: research personnel, consumables, travel, field work costs, contingency, and "other costs" (which covers software licences, cloud credits, publication fees). System integration is not in the list. Consulting from an external startup is not in the list.

And ANRF's own rules state that industry researchers may participate in projects as "honorary investigators" - but without funding support. The industry person can be in the room, but the grant cannot pay them.

This means that if you used GeM to buy a German product and an American one, you are totally dependent on a usually uninterested overseas vendor to help build your university lab. You cannot hire a consultant to help you put them together in India.

Sometimes one can work around the system and get a lab set up. IIT Kharagpur wanted to develop an ECU. (Read about India's ECU situation.) My team sourced components from specialist vendors in the US, Germany, and Sweden, integrated them into a system no single vendor could have delivered alone, and commissioned the whole thing as a working lab. The real value was in the integration.

We repeated this model at several other institutes with varying success - enough to demonstrate that when a university works with a system integrator, the integrator can fill the knowledge gap that Indian OEMs and Tier 1 suppliers cannot.

But the engineering cost in these cases still has to be buried under product line items. There is no budget head for "integration." We typically listed it as "installation and commissioning," took product A and product B, built system C, and justified the price difference with a sentence.

The workaround is so cumbersome, and the margins so slim, that few serious vendors, let alone startups, are willing to take on the work of helping universities build labs.

In 2024, R A Mashelkar, the former director general of the Council of Scientific and Industrial Research (CSIR), along with Ajay Shah and Susan Thomas of XKDR, published a paper making the case for precisely this shift - from a system that "makes" innovation through vertical government organisations to one that "buys" it by contracting out to whoever can deliver, the way NASA and the National Institutes of Health (NIH) operate in the US.

Their argument is sound. But it remains aspirational. The GFR got a new gate for product deliveries. The gate for people walking through to work together is still locked.

Money is not always a bottleneck, at least for the larger institutes like the IITs. One scheme, the Uchhatar Avishkar Yojana (UAY), alone allocated Rs 250 crore a year for IIT-industry projects, with a standing offer to provide more if needed. What was often missing was the freedom to spend it on the engineering relationships that would have made it productive.

The Metrics That Reward the Wrong Things

There is another lock: the incentive structure inside academia.

The National Institutional Ranking Framework (NIRF) has become the metric that determines how much grant funding a university receives. NIRF's Innovation ranking scores institutions on patents filed, startups incubated, and industry collaboration.

Not unreasonable things to measure. But when institutions optimise for the measurement rather than the activity, the results are perverse.

Universities file hundreds of patents a year, but the vast majority are never commercialised. MoUs (agreements) are signed with multinational automakers who do no R&D in India, announced in the papers, and quietly forgotten. Startups are counted as "incubated" regardless of whether they survive.

The number enters the NIRF submission. The ranking rises. The grant money follows.

TRL tells industry to stay away from early-stage research. The GFR lets universities buy products but not engineering relationships. NIRF rewards the appearance of innovation over its substance.

When Professors Must Build the Industry

The failure of joint R&D to take root has produced a default alternative: professors becoming entrepreneurs themselves.

In many frontier domains, this is not just a fallback - it is how entirely new industries are born.

Oxford Instruments began with a physicist at Oxford building superconducting magnets. IonQ came from trapped-ion research at Maryland and Duke. Delft University's quantum ecosystem has spun out half a dozen hardware companies in the past decade - Delft Circuits, Qblox, QuantWare, QphoX.

When the industrial ecosystem does not exist, the researcher builds it.

Dr Sugam Kumar, who straddles the divide through QUTE Electronics, a spin-off of the Inter-University Accelerator Centre (IUAC), New Delhi, makes the case plainly. Indian conglomerates will not enter deep tech. The market is too small, the timelines too long, and the patience required too great.

The TRL handoff does not motivate a technical person who has taken a technology to TRL 3-4 only to hand it to an industry player who lacks the depth to take it further. The only path, Kumar argues, is for technical people to spin off companies and carry products to TRL 7-8 themselves.

The government has encouraged this. Incubators have proliferated as a result. India's growing quantum startup ecosystem is proof that it can work.

But this path succeeds only with the right scaffolding. Delft had QuTech. Stanford had Terman's deliberate strategy and a mature VC ecosystem. The Netherlands had a government that understood deep-tech timelines. These professor-entrepreneurs did not go alone. They had institutional support, experienced co-founders from industry, and capital that understood the difference between a five-year and a five-month payoff.

In India, the scaffolding is inadequate. Rajat Sethi of QuBeats sees what is happening instead. Professors are guarding ideas, reluctant to share with students or peers. The deeper problem, Sethi warns, is that "building out a company is an extremely hard full-time job, maybe even several lifetimes of work. And you cannot be balancing between two oars."

Professor-led startups are both a symptom of weak industry participation and the starting point from which new deep-tech ecosystems emerge. The answer is not to discourage them but to build the infrastructure. Think of a translation institute, the system integrators, the patient capital. Such infrastructure shares the burden and lets the professor do what she is best at while the company-building happens alongside, not on top of her teaching load.