Why public capital isn’t enough to save India’s deeptech

What does “deeptech” mean, and how is it different from ordinary software startups?

Deeptech refers to startups built around challenging scientific or engineering technology. Their products may involve advanced AI, hardware, defence systems, space technology, or other technologies requiring substantial research and testing. The central question is not only whether customers want the product, but whether it can be made to work reliably.

Ordinary software can often be developed, tested, and changed quickly with relatively limited physical infrastructure. Deeptech products may need laboratories, specialised equipment, regulatory approval, manufacturing, or field trials. The article highlights development cycles of 8–12 years, far longer than many software businesses need to launch and scale.

That longer path creates technology, time, and exit risks. Public grants can help prove the science, but private investors still worry about waiting years for growth or a sale. This explains why funding is increasingly concentrating on already-proven AI and hardware winners rather than earlier, untested deeptech ideas.

How much money is flowing into India’s AI and deeptech sector, and what happened to the number and size of investment rounds?

India’s AI and deeptech investment reached $2.1 billion across 289 deals in 2025. Deeptech’s share of total venture capital and private equity activity also rose to roughly 15%, from 4% in 2016. This shows that the sector has gained substantial investor attention.

The 2026 figures through August are striking. Companies raised $2.22 billion across 179 rounds, compared with $861 million across 287 rounds during the same period in 2025. Funding therefore rose about 2.6 times, even as the number of rounds fell by nearly 40%.

The average investment cheque increased from roughly $3 million to more than $12 million, a four-fold jump. This pattern suggests that private capital is backing fewer, larger, and more established winners. Investors are becoming comfortable with proven companies, but not necessarily with the uncertain process of derisking new deeptech ventures.

Why can public funding reduce the risk that a deeptech product will not work, but fail to solve the problems of long development times and uncertain exits?

Technology risk asks whether a product can work at all. Government grants and lab-to-market funding can pay for research, testing, and development, so the state is well placed to absorb this early uncertainty. That support can turn a scientific concept into a functioning product.

Time risk is different. Deeptech development may take 8–12 years, while a typical fund has a 10-year life. Even patient fund-of-funds capital only partly fixes this mismatch. Investors may need to wait longer than their fund structure allows before a company becomes valuable or ready for sale.

Exit risk is weaker still. Tracxn counted 104 acquisitions and 46 IPOs among 1,821 funded companies. That equals roughly one exit for every twelve funded startups over a decade. Without credible routes to acquisitions or public listings, government first-loss support may not persuade private investors to underwrite the sector.

What are the RDI Scheme and Startup India Fund of Funds 2.0, and how are they intended to attract private investment?

The Research, Development and Innovation Scheme is designed to create a Rs. 1 lakh crore corpus over six years. The Union budget allocated Rs. 20,000 crore for it in FY 2025-26. Its purpose is to support research, development, and the journey from laboratory technology to commercial products.

Startup India Fund of Funds 2.0 works through SEBI-registered alternative investment funds. SIDBI is mandated to deploy Rs 10,000 crores through these funds, with government investing alongside private investors. Private investors make up the majority of each fund’s corpus, so public money is intended to leverage additional capital rather than replace it.

The first Fund of Funds invested through 145 AIFs, which together put more than Rs. 25,500 crores into over 1,370 startups. That produced leverage of about 2.5 times. The new programmes could attract private money by sharing technology risk, but they will work best if they also address long development periods and weak exits.

What happens to private investment when government money supports startups but does not create customers or viable ways for investors to exit?

Government support can reduce the cost of backing risky technology, but money alone does not create demand or an eventual sale. Private investors need evidence that a startup can earn revenue and that their fund can return capital through an acquisition or IPO. Without those signals, public funding may not change their basic decision.

The article warns that state capital at this scale could inflate valuations or displace the private investors it is meant to attract. It may also encourage “policy-shaped” startups that chase allocations rather than markets. A large public fund can therefore increase financing without increasing conviction in the underlying businesses.

The result may be a split market. Proven AI and hardware winners receive larger private cheques, while earlier companies remain difficult to fund. The article points to procurement, strategic corporate investment, and credible exit buyers as necessary complements. Without them, India could get plenty of capital but limited confidence in deeptech’s long-term commercial prospects.

Why could a government procurement commitment from defence, space, or railways help a deeptech startup more than an equally large grant?

A grant helps a deeptech company answer a technical question: can the product work? A procurement commitment answers a commercial question: will a serious customer pay for it? That distinction matters because private investors price future revenue, not only scientific achievement.

For example, a Rs 200 crore commitment from defence, space, or railways could support a startup’s product adoption and demonstrate demand. It may help the company raise a Series B round because investors can see an anchor customer, expected revenue, and a real deployment pathway. The government becomes more than a financier; it becomes an early buyer.

The article proposes advance market commitments, outcome-based public purchase contracts, and faster IDEX-style procurement. These tools could reduce both market uncertainty and time to commercialisation. They would also give private investors stronger evidence that a startup can grow, while creating the revenue and customer validation that grants alone cannot provide.

How do venture capital funds make money, and why do fund lifetimes, startup development cycles, acquisitions, and IPOs matter to their decisions?

Venture capital funds collect money from investors, place it into startups, and seek returns when those companies grow and are sold or listed. The fund’s success depends on a small number of investments producing large gains. Investors therefore assess not just technical promise, but the likelihood and timing of future exits.

Deeptech creates a timing problem. Companies may need 8–12 years to develop, while funds typically have 10-year lives. A startup can remain promising but still be too early for an acquisition or IPO when the fund needs to return money. Patient capital helps, but it does not fully remove this structural mismatch.

Exit data also shapes decisions. The sector had 104 acquisitions and 46 IPOs among 1,821 funded companies, or about one exit per twelve startups over a decade. Those odds make investors cautious. More corporate buyers, strategic acquirers, and public-market pathways would make deeptech funding easier to justify.

Key Facts:

📌 Deeptech includes advanced AI and hardware companies.

📌 Deeptech development cycles can last 8–12 years.

📌 Public funding can help prove whether deeptech technology works.

📌 India’s AI and deeptech investment reached $2.1 billion in 2025.

📌 2026 funding reached $2.22 billion through August.

📌 Average cheques rose from roughly $3 million to over $12 million.

📌 Deeptech development cycles run 8–12 years.