Multi-stage venture capital firm Lightspeed, which has backed startups such as Acko, Zepto and PineLabs, received nearly 1,400 qualified applications for its deep-tech fast-track programme, India Ascends, and selected just 12 – a roughly 1% acceptance rate, says partner Dev Khare. In a conversation with Ayanti Bera, he discusses the rising interest among young founders in building deep-tech companies, growing policy support for the sector, and why “AI for India” remains under-appreciated. Excerpts:
What does the response to Ascends tell you?
There’s a lot of excitement around this category. It’s still early, maybe the first five years of deep tech in India, and really the first two or three years with AI. But the excitement is palpable. There’s even a 15-year-old in the cohort, and the average age is about 22. These are young founders with mission orientation, much like the deep-tech founders we have already invested in at Pixxel Space, Airbound, and Sarvam AI.
What kinds of companies are you seeing in deep-tech?
Since Lightspeed is a global firm, we look at the sector from around the world. Globally, we see AI-based drug discovery, space tech, drones, aviation technology, semiconductors and defence. In the US, there are companies working on nuclear fusion and even deep-sea mining. What’s different now is that we’re getting founders who are not executives who’ve been in industry for 20 years. These are much younger folks.
There’s also a whole geopolitical angle to this that makes funding sources and customers available now, that weren’t necessarily there 10 years ago. For example, governments and military have procurement mandates and experimental budgets for defense equipment such as unmanned drones, unmanned submarines, unmanned tanks, robotics etc.
What stood out about the 12 companies selected?
They seemed very audacious, and the founders felt deeply technical, and that’s what you need in deep tech. You usually won’t find non-technical business leaders starting these companies. In the case of foundation and frontier models, many are researchers coming out of labs. Others have been building and tinkering in college labs or on their own. These founders immerse themselves into something highly technical that will happen sometime in the future. While at some point, you need business leadership, but in deep-tech, invention comes first.
How do you quantify success over five to seven years?
There are non-commercial milestones. Proof of concept is one — does the thing even work? It may not be commercially or economically viable yet, but it should function. Then you look at design partners, regulatory approvals or certifications, especially in defence or drug discovery. Financing milestones matter too. Deep tech usually requires a syndicate of investors because there’s a lot of capital needed over long periods. Multiple investors often come in over time.
How do you think about timing and risk?
The art lies in not picking companies that will take 20 years to commercialise. That’s more for governments, foundations or grants. Venture capital works best when technologies are within roughly a 10-year time frame. Sometimes we don’t get it right. For example, fusion has been venture-backed for the last 20 years and still hasn’t fully happened. So, timing is critical.
Is there more competition for early deep-tech deals?
If you zoom out, India still has a limited number of firms doing solid Series A and seed investing. There are less than 10 venture capital funds that do solid series A’s, maybe 10 to 20 funds that do seeds. In total, there’s maybe 30 organizations that do seed in series A. In a country of this scale, that’s a very small number. Within deep tech, the number is even smaller. So yes, there’s competition, but there’s also room.
Any AI themes that you are excited about?
AI for India. Everyone talks about AI from India for the world. But what about AI built for Indian consumers and businesses? Voice AI, financial advisory, consumer applications — these are meaningful areas. There’s also a narrative that India is behind because we don’t have an OpenAI. But there are things happening in India that aren’t happening elsewhere.