Episode 163 · Deep tech · 33 min

Talent was never the bottleneck

Ankur Capital's deep science report argues India's constraint has moved. The research is good and the talent is not in doubt — but a tech-bio startup with lab data still cannot find a 500- or 1,000-litre tank anywhere in the country to prove it at scale. Their evidence that the ground is shifting anyway: for the first time, in FY24, more than half the patents filed in India came from Indian entities.

SA
Suraj and Vishal
Investment team, co-authors of the deep science report, Ankur Capital · with Vishal Krishna
Talent was never the bottleneck — episode thumbnail
33:09
Said in this episode
▶ 13:21
50%+
Patents filed in India by Indian entities
First time ever: in FY24, over half of all patents filed in India came from Indian people and Indian corporations, as reported in Ankur Capital's report.
▶ 10:36
TRL 4–5
Where Ankur Capital writes its first cheque
After early prototype development, with lab-level data validating the idea — never pure R&D, which the fund treats as a grant-and-pre-seed journey.
▶ 6:58
500–1,000 L
The pilot run India can't host
The scale-up experiment a tech-bio startup needs after building in the lab; the panel says there is no real facility with that capacity in the country.
▶ 21:18
70% vs 34%
Corporate share of R&D, China vs India
Quoted by an audience founder, not from the report — treat as an approximation of the gap rather than a verified figure.
▶ 0:36
222
People who sat through the report's live session
The moderator's marker of rising attention on deep science: 222 stayed for an hour and a half on the online session before this packed room.
The brief

The argument in sixty seconds

The claim from Ankur Capital's two report co-authors is that India's deep science constraint has migrated. Technology, Suraj argues, is now available to everybody, and talent is not the question — good research is happening in the universities. What is missing is everything between the bench and the market: commercialisation, industry participation, and above all infrastructure, because a startup that has proven something in a flask cannot find a 500- or 1,000-litre pilot facility to run it at scale. China's asset values, he says, are simply what happens when a country builds that layer first. The capital gap is the argument's second half: founders SaaS-ify their stories and quote ARR because the people writing cheques do not speak TRL, so a company that has genuinely derisked its technology has no vocabulary in which to say so. Ankur enters at TRL 4 or 5 — early prototype, lab-level data, never pure R&D — and funds the pilots, trials, regulatory filings and multi-geography patents that stand between a prototype and a product. The signs they offer that this is turning are unglamorous and specific: median cheque sizes rising, family offices and India's first corporate venture arms appearing, and Indian entities crossing half of all patents filed in India — which they insist is step one of defensibility, not proof of it. The unfinished work sits upstream, in a research system where high-quality work still fizzles out before it reaches the real world.

Worth your time if you are

Lab founders with data and nowhere to scale it
PhD students weighing a company against a paper
Investors trying to price pre-revenue science
Family offices and corporate venture arms new to deep tech
Policy people building India's research-to-market rails
Episode map

Where the conversation travels

Every block is a chapter, coloured by what it's about. Click any of it to jump straight to that minute on YouTube.

01Cold open: a bigger room than last year 0:00 The moderator opens a packed session on Ankur Capital's annual deep science report — 222 people had sat through an hour and a half of the X live — and notes the visible jump in turnout and in returnees from the US and UK who want to build in India. 02Why a scientist came home 1:38 Vishal, one of four co-authors, explains that he went to the US for a PhD intending to be a full-time scientist but always believed Indian technology would be globally competitive — and that the RDI fund and a less piecemeal capital stack have since made the bet concrete. 03The map: thesis, IP, capital, convergence 3:10 The moderator walks the report's structure — the India thesis and borderless science, defensible IP built locally, the state's appetite for backing deep tech, and the converging forces across AI and compute, energy, tech bio and space — before the lab-to-market journey and the funding graphs. 04Talent isn't the gap — pilot plants are 5:13 Asked which of technology readiness, market pull and geopolitical tailwinds breaks first, Suraj says technology is available to everyone and talent is not in doubt; what is missing is commercialisation, industry participation and infrastructure — there is no facility for a 500- or 1,000-litre pilot run, which is what China built before its assets were worth anything. 05Why deep tech gets SaaS-ified 8:39 Deep science revenue arrives late and lumpy, yet founders describe themselves in ARR because capital providers and other stakeholders do not speak the language of technology readiness levels and technology derisking. 06Ankur enters at TRL 4 or 5 10:07 The fund typically writes its cheque after early prototype development, when lab-level data validates the idea, and avoids pure R&D — then funds the three or four gates before commercialisation: customer pilots, clinical trials where regulated, regulatory filings and IP filings across geographies. 07Half of India's patents are now Indian 11:40 For the first time, in FY24, more than half of all patents filed in India came from Indian entities — a number the panel treats as step one of knowledge defensibility rather than proof of it, since intellectual property is also know-how, and the Indian filing is only the start of the global route. 08The median cheque is rising 14:08 The graph they most like shows median funding size climbing and more early-stage money flowing in; Suraj recalls being the only investor turning up at university campuses when he joined in 2021, and now sees portfolio companies raising Series A and B rounds. 09Patient capital, family-office shaped 15:55 Where VC funding is not available or the company is not yet VC-ready, family offices are stepping in with more understanding and more patience — the kind of capital that long R&D cycles and lumpy revenue require, and a sign the ecosystem is maturing. 10What got cut, and how 2026 gets sliced 17:09 The authors wanted five pages on every technology area and settled for a thousand words; the 2024 report cut deep tech by underlying technology and this year by application area, with AI inferencing — cloud and edge inference costs, models, architectures, memory transport — as the current obsession and drug discovery a candidate for 2026. 11India's first corporate venture arms 19:36 Alongside family offices, the panel flags the rise of Indian corporate venture capital — Indigo Ventures launching, a portfolio company raising from a publicly listed entity — and corporates investing from balance sheets into companies with no revenue on the day of the cheque. 12Why Indian industry doesn't fund science 20:34 A founder building plastic-degrading enzymes asks why Indian corporates fund so little R&D against China's much higher corporate share; the answers are inertia, a growing base of translational research centres and grants, and foreign conglomerates setting up JVs that want startups as first customers but do not yet know how. 13Nobody agrees what a TRL means 23:50 A scientist from ARTPARK points out that a venture centre and an IIT may define technology readiness levels differently and that the deep tech world lacks the shared vocabulary SaaS founders take for granted; the panel's answer is that definitions are emerging by sector, and that founders must know what stage they are raising for and from whom. 14Early screening, and AI that touches things 26:31 Asked what changes human-technology interaction over the next ten to fifteen years, Suraj points to early disease screening built on newly discovered biomarkers and Indian non-Caucasian data sets, while Vishal wants AI that holds its own task autonomously and acts in the physical world. 15A university for batteries, a system for papers 29:01 A desalination founder and educator asks what would fix the pipeline; the moderator answers with China building an entire university around battery manufacturing, R&D and recycling, while the panel argues India needs graduate researchers who treat companies and patents as outcomes and more industry-academia translational centres.
Takeaways

Ideas to carry out of this hour

01

The bottleneck moved from the lab to the tank

Suraj's position is that technology is available to everyone now and Indian talent is not the constraint — good research is happening in the universities. The gap is everything after: commercialisation, industry participation and, most concretely, infrastructure. A tech-bio startup that has built something in the lab and wants a 500-litre or 1,000-litre run to see whether it holds has nowhere in the country to do it, which is why new policy is aimed squarely at that hole. His comparison is blunt: Chinese assets are valued where they are today only because China built the infrastructure that could produce them.

02

Founders SaaS-ify because capital cannot speak TRL

Deep science revenue is late and lumpy, yet founders keep translating themselves into ARR. The reason is not vanity but audience: capital providers and other stakeholders do not talk the deep tech language, so a founder cannot say 'I am at this technology readiness level and I have derisked this much' and expect to be understood. To get partners on board you revert to a vocabulary that does not suit a science-based or hardware company — and then get judged by it.

03

The cheque starts where the prototype ends

Ankur typically invests at TRL 4 or 5 — after early prototype development, when there is lab-level data validating the idea — and deliberately refrains from funding pure R&D, which it treats as a different journey served by grants and very early pre-seed money. What the money buys is the crossing itself: pilot testing with potential customers, clinical trials for regulated products, regulatory approvals and filings across geographies, and IP filings across multiple geographies. Three or four steps, all of them before a rupee of product revenue.

04

Half of India's patents are Indian — and that is only step one

For the first time ever, in FY24, more than 50% of the patents filed in India were filed by Indian entities — Indian people and Indian corporations. The panel reads it as a signal that founders are thinking about defensibility and about the sequence of filing in India first and going global after. But they refuse to over-read it: patents are not the only place technology is defensible, know-how matters as much, and the number tells you the ecosystem has taken step one, not that it has built a moat.

05

The capital stack is finally diversifying

The graph the authors like most shows the median funding size rising, with more early-stage money and the first real signs of growth-stage money as portfolio companies close Series A and B rounds. Underneath that, the sources are changing: family offices funding companies that are not VC-ready, India's first corporate venture arms appearing with Indigo Ventures, and a publicly listed Indian company writing a cheque into a portfolio startup. Globally it is corporates who are most active in this segment; India is just starting, and the patient money is what long R&D cycles have always needed.

06

Corporate India still doesn't buy its own science

An audience founder building plastic-degrading enzymes puts the gap in numbers — a large majority of China's R&D comes from the corporate sector against a much smaller Indian share — and asks how startups get past the hesitation. The panel does not claim a fix: there is inertia, and this is the first generation making the argument at all. What is changing is the surrounding scaffolding — translational research centres, initiatives and grants, public money doing far more than five years ago — plus foreign conglomerates setting up Indian JVs who want to work with startups and simply do not know how. The prize is moving from a logo on a year-end deck to being someone's first customer.

07

Nobody in India agrees what a TRL means

A scientist from ARTPARK notes that a venture centre and an IIT will classify technology readiness levels differently, and that the SaaS ecosystem has a rich public conversation about stages while deep tech has almost none. The panel concedes the definitions are not yet hardbound in India and are emerging sector by sector — a readiness level means something different in bio than in robotics. Their practical answer puts the burden back on the founder: know what you are raising for, know which step comes next, and pick the source of money that gets you there.

08

A research system that optimises for papers leaks value

The education gap the panel names is not a shortage of quality research but its evaporation: high-quality work fizzles out by the time it reaches the real world. What they want is graduate researchers who think about starting companies and filing patents rather than papers as the terminal outcome, and far more translational research centres where industry-academia projects actually run — Suraj recalls a campus where every professor had one, and an alumni meetup where he saw how much that had faded. The counter-example is a country that decided battery was a core sector and built an entire university around its manufacturing, R&D and recycling.

The numbers, drawn

What the episode measures

Every figure below was said on air — timestamps included, caveats kept.

Conversation share

portion of the hour spent on each theme
Deep tech & hardware · 24%Venture capital · 20%Fundraising · 12%India macro · 10%Education & skilling · 10%Regulation & policy · 9%
Deep tech & hardware24%
Venture capital20%
Fundraising12%
India macro10%
Education & skilling10%
Regulation & policy9%
Computed from the chapter map of this episode.

Who pays for the research

% of national R&D from the corporate sector
China70India34
Figures as quoted from the floor by an audience founder, not from Ankur Capital's report — indicative of the gap the panel then discusses rather than a sourced statistic.▶ 21:18
Worth keeping

Lines that stay

I don't have to just say that I came back to India because I wanted nice dosas every day. There's a lot of globally competitive technology that is going to be built, and I want to be part of it.

— Vishal, Ankur Capital ▶ 2:55

Assets from China being valued at whatever numbers today is only because they built the infrastructure that could actually produce that level of assets.

— Suraj, Ankur Capital ▶ 7:12

When I joined Ankur back in 2021, I used to be the only person going to these universities. Investors used to be panned, and I used to be the one listening to all of that. Now there are more people to listen to it.

— Suraj, Ankur Capital ▶ 15:12

We have a lot of intelligence, artificial intelligence, but we don't necessarily have durable AI that interacts with the physical world.

— Vishal, Ankur Capital ▶ 28:37

There's a lot of high-quality research. It kind of fizzles out by the time it reaches the real world.

— Vishal, Ankur Capital ▶ 31:16
Clips that travel

Short on time? Start here

Lab founders with data and nowhere to scale it

Talent isn't the gap — pilot plants are

The episode's core argument: technology and talent are settled questions, and the missing 1,000-litre tank is the one that isn't.

5:13 → 8:39 · 3 min ▶ Watch clip
Founders pitching science to generalist investors

Why deep tech founders start talking ARR

The language mismatch between TRL and ARR, then exactly where a deep tech fund's first cheque lands and what it pays for.

8:39 → 11:40 · 3 min ▶ Watch clip
Anyone building a defensibility story around IP

Half of India's patents are Indian now

The FY24 crossover, why patents are only a signal, and where know-how sits in the moat.

11:40 → 14:08 · 2 min ▶ Watch clip
Investors trying to price pre-revenue science

The median cheque and the arrival of patient capital

Rising median rounds, family offices funding what VCs won't yet, and a fund partner's memory of being alone on campus in 2021.

14:08 → 17:09 · 3 min ▶ Watch clip
PhD students weighing a company against a paper

A university for batteries

The sharpest exchange of the session: what an education system built for commercialisation looks like, versus one that ends at the paper.

29:01 → 32:33 · 4 min ▶ Watch clip
Glossary

The jargon, unpacked

Deep science / deep tech
Companies whose advantage comes from hard scientific or engineering work — long R&D cycles, physical products, and revenue that arrives late and lumpy.
TRL (technology readiness level)
A 1-to-9 scale for how far a technology has come from idea to deployed product; Ankur Capital invests around 4 to 5, meaning an early prototype backed by lab data.
Translational research
Work that deliberately carries a laboratory result towards a real-world product, usually through industry-academia collaboration — the layer the panel says India most lacks.
Pilot facility
The intermediate plant where a lab process is run at 500 to 1,000 litres to test whether it survives scale; India has almost none, so startups stall between bench and factory.
CVC (corporate venture capital)
A company investing its own or a venture arm's money into startups; the panel flags India's first examples appearing, against a global norm where corporates are the most active backers of deep tech.
RDI fund
The Indian government's research, development and innovation financing vehicle, announced after the first edition of the report — cited as evidence the capital stack is no longer piecemeal.
ARR
Annual recurring revenue, the SaaS yardstick that deep science companies are pushed to quote because their investors have no other shared metric.
Edge inferencing
Running a trained AI model on local hardware rather than in the cloud; cutting its cost is one of the specific technology bets the report calls out.
Connections

If this resonated, go here next

Full transcript

The whole conversation, searchable

133 segments

Auto-generated captions, lightly cleaned. Click a timestamp to open that moment on YouTube.