'Raise Capital Now': Nikhil Kamath's AI Bubble Warning — and What Indian Startups Should Actually Do

This week, one of India's best-known investors said what many founders have been privately worrying about.

Speaking at the NDTV World Summit and in interviews this week, Zerodha co-founder Nikhil Kamath warned that AI valuations are showing signs of a bubble — and backed blunt advice for startups: raise money now, keep enough cash to last a year or two, and spend carefully, because next year could be bad for funding.

His name has been trending on Google ever since. For founders, the question is practical: is he right — and if so, what should an Indian startup actually do?

In a recent post, I looked at the trillion-dollar bet Big Tech is making on AI infrastructure. This post looks at the same story from a founder's desk.

What Kamath actually said

Based on reporting from his remarks this week:

AI spending far exceeds AI revenue. Kamath estimated that roughly US$1 trillion a year is going into AI, against perhaps US$100–200 billion of enterprise AI revenue — leaving a gap of US$700–800 billion that investors have to fund every year. These are his own estimates, not audited figures, but they're broadly in line with what analysts at J.P. Morgan and Sequoia have argued.
"It always feels like this time is different." He compared today's enthusiasm for global AI companies to past bubbles, in which investors priced in very optimistic outcomes years ahead.
Capital could get harder to find. With US bond yields rising and borrowing costs climbing, he warned that raising fresh money to support current valuations could become much harder in the coming months.
His advice to founders: he publicly agreed "100%" with a post urging startups to raise enough money to stay comfortable for a year or two, because next year could be tough.
India's opportunity is different. He argued that value will increasingly go to companies that own local data and context, that open-source AI models will spread, and that Indian companies should focus on serving businesses rather than competing with US and Chinese labs on giant frontier models.

Kamath also spoke positively about the Jio Platforms IPO this week, saying demand far exceeds supply — and noted that his investment platform, Rainmatter, took part in its anchor book. It's worth keeping that in mind: he's not a bystander in India's tech markets. But his warning on AI valuations is consistent with what many independent analysts are saying.

Is he right?

On the big picture, the evidence supports his caution:

The spending-revenue gap is real. Big Tech is spending around US$660–780 billion on AI infrastructure this year, while leading AI labs earn a fraction of that.
Doubts are growing. When OpenAI's annualised revenue came in about US$18 billion below widely reported figures earlier this month, AI-related shares fell sharply.
Enterprise returns are uneven. McKinsey's 2026 survey found only about 6% of companies are seeing significant financial value from AI.

But a funding squeeze isn't certain, and India isn't Silicon Valley:

India's capital markets are holding up. Strong demand for IPOs, including Jio's, shows that public investors still have appetite for profitable tech businesses.
Indian startups mostly aren't building frontier models. A correction in US AI-lab valuations wouldn't hit an Indian SaaS company the same way — though global venture funds that back Indian startups would feel the pain and could become more cautious everywhere.
Bubbles don't burst on a schedule. "Next year could be bad" is a forecast, not a certainty.

The sensible conclusion isn't panic. It's preparation.

What Indian startups should actually do

1. Raise if you can — but optimise for runway, not valuation

If good capital is available on reasonable terms, take it. A slightly lower valuation with 24 months of runway beats a headline valuation with 12. Avoid terms that look harmless today but hurt later — such as aggressive liquidation preferences or ratchets that punish you in a down round.

2. Plan for 18–24 months without new money

Build a base case where you don't raise again for two years. What would you cut? Which hires would wait? Which experiments would stop? Having that plan ready means you can act calmly if the market turns.

3. Get serious about unit economics

In a tight market, investors fund revenue, margins and efficient growth, not just user numbers. Know your customer acquisition cost, payback period, gross margin and burn multiple — and show they're improving.

4. Don't be a thin wrapper around someone else's AI

If your product is mostly a user interface on top of a foundation model, you're exposed: the model provider can copy you, change prices or cut you off. Kamath's point about local data is the key: build a moat from proprietary data, deep workflows, domain expertise and distribution, especially in Indian contexts — languages, regulations, payments and the way Indian businesses actually work.

5. Watch your AI compute bill

For many AI startups, model and compute costs are the fastest-growing expense. Use open-source and smaller models where they're good enough, route only the hardest tasks to expensive frontier models, cache aggressively, and avoid locking into a single provider.

6. Sell outcomes to businesses

Enterprise customers are under pressure to prove AI's value. Startups that can show measurable savings or revenue gains — and price on outcomes — will keep winning deals even when budgets tighten.

7. Diversify your sources of capital

Beyond venture capital, consider revenue-based financing, carefully structured venture debt, strategic investors from your industry, and government-backed options such as the Fund of Funds for Startups and the Startup India Seed Fund Scheme. Each has trade-offs, but having options reduces dependence on any one source.

8. Use India's natural advantages

India's public digital infrastructure (UPI, Aadhaar, ONDC and more), cheap data, a deep talent pool and a lower cost base are real advantages. Products built on these foundations — for Indian users and for global customers — can be far more capital-efficient than their Silicon Valley equivalents.

9. Don't call yourself an "AI company" unless you are

Not every startup needs to be an AI startup. If AI is a feature rather than the core of your business, say so. When sentiment turns, companies whose value depends on the AI label are hit hardest.

10. Keep your house in order

Clean books, a clean cap table, good governance and regular, honest investor updates matter more in a downturn. Investors back founders they trust when times are hard.

What investors should watch

Revenue quality: recurring, growing and profitable — not one-off pilots.
Dependence on a single AI provider and exposure to rising compute costs.
Burn multiple and runway across the portfolio.
Valuations versus fundamentals, especially for companies priced as if AI will deliver years of perfect growth.

Final thought

Nikhil Kamath's warning isn't a prediction that AI will fail. It's a reminder that money and technology run on different clocks. AI may transform the economy over the next decade — but funding markets can tighten in a single quarter.

The startups that come through the next cycle won't necessarily be the ones that raised the most or shouted "AI" the loudest. They'll be the ones that raised sensibly, spent carefully, built real moats and sold real value. That advice was true before the AI boom, and it will be true after it.

Are you raising, cutting or waiting? How is your startup preparing for a possible funding squeeze? Share your view in the comments.

Sources: Reporting on Nikhil Kamath's remarks at the NDTV World Summit 2026 and in interviews, by CNBC-TV18 (via TradingView, 21 October 2026), NewsBytes, The Economic Times, Hindustan Times, NDTV, Inkl, StartupTalky, Superpower Daily and Pragativadi; Google Trends (India) trending searches, October 2026; Axios and CNBC on OpenAI's reported revenue (October 2026); Futurum Group on hyperscaler capital expenditure; McKinsey, The State of AI: Global Survey 2026; J.P. Morgan and Sequoia Capital on the revenue needed to justify AI investment; Startup India and SIDBI on the Fund of Funds for Startups and the Startup India Seed Fund Scheme.