Nikhil Kamath warns startups of a 2027 funding squeeze as AI spending outpaces revenue

Startups should consider securing cash while they still can, Zerodha co-founder Nikhil Kamath said in an NDTV World Summit interview covered on October 22. He warned that fundraising could get harder in 2027, arguing that AI’s enormous investment needs have run ahead of enterprise revenue—and that sustaining the gap will require more capital.

A financing warning, not a verdict on AI

Kamath estimated that roughly $1 trillion a year is flowing into AI, against perhaps $100 billion to $200 billion in enterprise revenue. Those are his estimates, not independently verified industry totals. His concern was how investors would keep financing the buildout before the hoped-for business returns arrived.

Investors are counting on productivity improvements and large economic benefits to justify that spending, he said. But Kamath questioned whether those gains were already visible: “I don't know if I've seen that happen on the ground yet.” His argument leaves the eventual payoff unresolved rather than declaring that AI cannot deliver it.

He pointed to foreign investment, including money from the Gulf, as support for American AI companies in the previous funding cycle. Replacing that funding at the same scale each year could prove difficult, he argued. He also cited rising US Treasury yields as a potential source of higher borrowing costs.

If somebody is able to raise a lot of capital today and keep it, probably not a bad idea.

Nikhil Kamath, speaking at the NDTV World Summit

India’s opening: operate models rather than fund the frontier

Kamath drew a distinction between developing frontier models—the most advanced AI systems—and running existing models for customers. He doubted Indian companies could compete fairly in frontier development when rivals in the United States and China have much greater access to risk capital.

Instead, he sees an opportunity in inference: running a model to produce answers or other outputs. Specifically, he suggested serving companies that want to use open-source models while keeping their data private. That places the proposed business closer to operating AI for customers than financing the development of a new leading model.

Lower infrastructure costs underpin his case. Kamath estimated that building a megawatt of data-center capacity in India could cost 30% to 40% less than in the West. He separately said electricity was significantly cheaper than in several other countries.

Cautious on funding, bullish on listed shares

His caution about startup finance did not extend to all investments. In the same interview, Kamath said he was 90% allocated to equities. He described valuations as attractive across several sectors, citing quality banks, IT-services companies and pharmaceutical businesses available at 11 to 13 times earnings.

Kamath also challenged how the benefits of rising asset prices are distributed. Wealth inequality cannot fall if wages fail to keep pace with asset-price inflation, he argued, suggesting government intervention may be necessary. He supported considering an inheritance tax, but said its design should not force founders to sell company equity to meet taxes on unrealized gains.

On trading fairness, he backed two changes aimed at limiting advantages over retail investors:

• End co-location arrangements that allow some traders to execute orders faster than retail clients.
• Intersperse algorithmic orders with retail orders so automated trading does not receive an execution edge.