“Geopolitics Has Taught Us We Must Build Resilience And Redundancy”
Vishesh Rajaram, Founding Partner at Speciale Invest, speaks to BW Businessworld on India’s deep tech momentum, the promise of sovereign technology, the right way to think about the semiconductor push, and why the country must build its own energy, GPS, and AI capabilities. Excerpts:How has the conversation around deeptech changed from your vantage point over the last two years?
In the last two years specifically, what’s changed is there’s more confidence. There is a deeper realisation of two very important things. First is that India can build deep and frankly, build it well. Not only in consumer products or software, but in critical hardware and hardware-software systems. There was a real question mark earlier on whether this was possible at all, and that has now changed. The second big change is capital availability. I’ll give a lot of credit to the RDI programme, which is looking to broad-base the availability of capital for commercialising science and technology. That tailwind and that policy push is clearly making more financial investors take note of this sector and consider investing.
We’ve heard about Rs 1 lakh crore Research, Development, and Innovation (RDI) Scheme, and more recently the Fund of Funds 2.0 notification. What does that do to a VC firm like yours? Does it change your fundraising strategy and your LP base?
It most definitely changes the LP base for all funds looking to invest in deep tech. The RDI programme, as they’ve publicly stated, allows them to invest in funds up to 50 per cent, which means they’re effectively providing matching money for all the private capital these funds can raise. That makes RDI a very significant, stable, long-term LP potential for venture funds like ourselves, which is a very big deal. It sort of mimics A*STAR in Singapore or DARPA in the US, where they are effectively taking the capital problem off the table by saying they will bring up to half the money, provided of course there’s a rated return on the upside.
A lot of leaders we’ve spoken to have raised the question of why institutional investors aren’t coming into startups in a bigger way. What are they looking for, and what’s holding them back?
For the longest time, they’ve been large investors in the public market. Deep science and tech comes with a slightly longer gestation period where risk is more tail-end. What’s changing is that there are now successes to show that this is possible out of India. That should get more of them to consider it. But anything in the private space, deep tech or otherwise, does come with illiquidity, it’s a 10 to 12-year product you’re investing in. From a diversification standpoint, yes, they should allocate some capital towards this. I’m sure it will happen over time, given that the sector itself is very new and upcoming.
If you look at the Indian deeptech ecosystem right now, where is the greatest opportunity for VCs?
It is still early days but if you look at what’s here and now, I’d clearly say spacetech is here and now. Defence manufacturing and semiconductors — one, because there’s a large and growing Indian market, and two, because geopolitics and supply chain pressures mean you need to diversify and have certain capabilities or parts of the value chain within the country. Advanced manufacturing, in robotics and in precision engineering focused on the aerospace sector, is also growing and has the potential to expand into large markets. These are all clearly active interests in the venture capital world right now. If I roll forward a few years, which is our job, to think and estimate ahead of time, I would put nuclear right in the centre of it. We have to solve the energy crisis, both from the standpoint of a growing need and the cost of energy. Nuclear is one clear avenue to broaden the availability of energy and bring its cost down and when that cost comes down, it opens up a whole plethora of other opportunities.
Realistically, in the current decade, what will be Indian deeptech’s role in the global ecosystem?
I think about it in two parts: India inward, and India outward. On the India inward side, energy security and energy transition is a big block. How do you generate more energy? How do you reduce its cost? How do you reduce dependence on fuels? That’s a big, big problem statement worth solving for in the decade ahead — and only growing from there. On the India outward side, I’d put manufacturing and space right in the middle. Both leverage India’s knowledge capacity and capability to deliver to global markets. Space is a global market, our Indian startups are looking to address global demand, not just domestic. Similarly for the manufacturing play — CNC machining, precision engineering, robotics — these are all solid cases to build from India for the world. Health is also a big piece, it’s been there for 20 years and will be there for the next 200. The complexity is only increasing, and interventions leveraging new technology — think protein unfolding, faster and more accurate drug discovery for cancer — are big problem statements that India has the knowledge capital and CRO/CDMO infrastructure to solve.
With geopolitical tensions becoming the new normal, do you see fresh interest coming into Indian deeptech? Do you see the capital picture stabilising?
I think it’s early. It is far from stabilising. It will get incrementally better and then normalise. We will see more global capital coming to participate in the Indian technology ecosystem. It will happen in due course. The global order is definitely resetting. Where it will all land, time will tell. Some of this also has to do with how interdependent or less interdependent we become and it almost seems like the less interdependent you are, the better standing you are in. A fund like us reads macro but operates in micro. Macros are there, but they don’t define what we do on a day-to-day basis.
Electronics and IT Minister Ashwini Vaishnaw talked about India Semiconductor Mission (ISM) 2.0 with a focus on design earlier this year. Every expert off-record has been saying we haven’t done enough on semiconductor design — fabs got all the glamour in ISM 1.0. What does ISM 2.0 need to get right, and how do we get capital to the people who actually need it?
One of the things in India is we are always impatient, we want everything to happen quarter-on-quarter. Deep science and tech unfortunately needs a little more patience. I actually have a very positive outlook on ISM because I’ve invested in semiconductors since 2008 and 2010, in a time when ISM didn’t exist. I know the pain of funding your own IP, funding your own tape-outs with equity capital, and just how binary and hard it is to raise capital in that environment. What ISM and more specifically the DLI scheme has done beautifully well is bring down the barrier to build. You get access to design tools, you get access to IP, and you get a certain amount of grant money that allows you to hold people. Two of our companies actually raised capital under the DLI and that’s possibly led to about a dozen companies coming up across the ecosystem.
Now, what can be done better? The quantum of capital needs to grow as the market grows. Earlier it was a couple of million dollars per startup, which was good initially but ISM 2.0 should possibly increase that quantum, because some people are working on slightly larger problems. I fully endorse the focus on design, because that’s where startups are going to play. OSATs, packaging, fabs — those are for larger organisations. The missing link that makes all that infrastructure useful is having enough people designing chips. India has that talent base of designers. That’s the right place to focus — to incubate and encourage more companies to come out.
From what startup founders are telling you, what specifically needs to improve about the DLI scheme?
The scheme operates on a reimbursement model, companies spend first and recover it later. And not everybody is equity-funded enough to go that route. Maybe the equivalent of co-investing with equity-like capital alongside other investors would help. Or simply expanding the pool of capital available. It’s good to get companies started, but some of them will need more as they scale. Venture capital funds will do their part, but there could be additional support mechanisms within the design-linked element that could further enable them.
When we spoke to your partner about a year or so ago, the view was that Speciale wouldn’t be going after LLM-scale AI. What is your thesis now, given how much AI is dominating deeptech funding conversations?
Within the larger AI scheme, there will be about five to ten companies in the country that aim to build LLMs. They will need fairly large amounts of money, and they should build them — LLMs that are very uniquely specific to India, with voice and regional language capabilities built in. We haven not done anything there yet, and we keep looking, but we also know it needs a large amount of capital to create. The more obvious play for a firm of our fund size is the small language model game, domain-specific models, where you’re leveraging deep expertise in a particular sector: aerospace, space, healthcare, drug discovery for cancer. You’re using your own data to build or fine-tune a small language model that works brilliantly well for your specific use case, one that outperforms large general models for your particular needs. The third area we’re spending time on is what I’d call the non-linear jump from LLMs — world models. LLMs are largely trained on language and text. The physical world is multi-dimensional, three-dimensional at least. How do you operate in that environment? It is a very different thought process, and there’s significant conversation happening globally on this. We are trying to see if there’s relevance for something like that in the Indian context for power, telecom, robotics — physical AI concepts.
US export controls on Anthropic’s latest models recently generated a lot of discussion in India around the need of a sovereign AI stack. Should India be in the LLM race? Sarvam is there, but the debate continues.
India is already in the LLM race. I don’t think it is a question we should even be asking ourselves. Given the stature this country has and the vision we have for ourselves, I don’t want to be asking whether I should be in the race for anything. I am in the race. I don’t think we can sit back and say someone else is going to build something and we will just use it. The last few years of geopolitics have taught you that you’ve got to figure yourself out. You’ve got to build resilience and redundancy into yourself, given the scale and size of this country and the aspiration that comes with it.
Pages:
[1]