Why are India’s most important startups still hard to value?
ADVERTISEMENT

An international school in Dubai has a problem. New parents can send their children on the school bus or drive them in the family car. Over the last two years, the share of children taking the bus has fallen from 40% to 20%, and more families drop out every month. The school now wonders whether to run buses at all.
The decline feeds itself. As fewer children ride, fewer riders live in each locality, so routes get longer. Longer routes mean longer commutes, and more parents opt out.
Deeptech investing in India is caught in a similar loop. Most investors do not believe a big exit is possible, so they do not pay up as startups build deeper technical moats. Without that reward, fewer founders and investors take the risk. Fewer ambitious companies get funded, and big outcomes become less likely, which seems to confirm the original doubt.
The timing could not be worse. Software is no longer the only game in town. Advances in AI and hardware are moving some countries quickly towards the future we once imagined for the next millennium. Physical AI, humanoids, spacetech, robotics, AI data centre components, biotech and semiconductors will not have markets limited to India. These companies will compete globally for both capital and customers.
Break the loop
India’s venture industry needs to break this loop by valuing deeptech startups on the technical moats they build. The reasons to do so are simple: the exits can be enormous and large amounts of capital can be raised to build globally competitive companies without breaking the cap table.
A deeptech company makes progress by solving hard technical problems. Each problem solved deepens its moat and should raise its value, often well before revenue arrives. If investors cannot value that progress round by round, founders and early backers have little incentive to take the risk.
Most of India’s alternative investment industry is not set up to do this. Few funds have the capital pool to back moonshots, and fewer have the expertise to assess a technical moat. Founders raising in India rightly complain that a typical investor meeting spends 5 minutes on technology differentiation and 45 minutes on go-to-market.
The structural gaps are real. The US spends about $1 trillion a year on technology R&D, $750 billion of it from corporations. Top talent from those research centres and hyperscalers gives investors a strong signal, and it helps explain the mega seed rounds there. In India, founders will mostly remain young, enthusiastic first-principles thinkers, with less exposure to global opportunity and timing. Frugal engineering helps, but producing results will still take large amounts of capital that is not available today.
The cost of inaction is high. This loop threatens to push tech talent towards countries with a greater appetite for risk, and to deprive India of indigenous technological progress.
Alternative asset investors are driven by returns, so evidence of returns is what will change how they value these companies. That evidence now exists.
China, US show the way
China has long been generous towards technology risk, and its early bets on autonomous vehicles and robotics are paying off. Its deeptech investments are adding to national competitiveness and producing some of the largest investor returns it has seen. Over 2025–26, the biggest valuation jumps came in three places: private rounds for AI model companies, Shanghai listings of GPU makers, and humanoid robotics.
● Unitree priced its IPO at $9 billion and closed its first day of trading in August 2026 at about $48–50 billion.
● Moonshot AI (Kimi) went from a $4 billion valuation at the end of 2025 to over $30 billion in July 2026. Meituan, China's equivalent of Blinkit, led its $2 billion round.
● Three hard-tech VC funds launched in December 2025, each reported at over ¥50 billion (about $7 billion).
The US is moving the same way. Hardware and physical AI now make up nearly one-third of all US venture capital outside pure-play large language model companies. Funds are growing to write much larger cheques for deeptech. In January 2026, a16z raised $15 billion across a family of funds, including a $1.1 billion Machine Age Fund dedicated to physical AI and hardware infrastructure. Talent leaving hyperscalers is raising seed cheques of $500 million.
India has early signs of its own. Pixxel, Sarvam, Skyroot.
Indian investors need to reset their expectations for this new world: huge risk, in exchange for huge returns. Many are handicapped by a lack of global context for benchmarking technology risk. This is no longer about valuing a uniquely Indian story like quick commerce. Technology has to be assessed against what the best exit could be globally.
That means building resources to answer questions like:
1. Is anyone else in the world building something similar? Who and what resources do they have? Where does this fit in someone else’s solution set?
2. How difficult is it to build, and what precisely are the hardest parts?
3. Is the IP defensible?
4. Is this a genuine breakthrough or an inflated claim, and can technical consultants help tell the difference?
5. What is imported and what is indigenous? This matters especially in spacetech.
Then, as a startup proves more, investors should reward it with a globally competitive valuation. Recognizing these proof points will need a different lens to understand. The era of the unicorn is over. The decacorn decade is next, and India’s investors should get ready for it.
(Chandra is Managing Director at Arkam Ventures and the author of two books: The Moonshot Game: Adventures of an Indian VC and Tightrope to the Moon: How Mega Founders Win the Startup War. Views are personal.)