Whenever I talk to VCs, I tell them that I don’t think thin-wrapper AI products from India are venture scale products. Unless they are world class. Because the cost of inference is the same whether you serve customers in India or the US.

Your CAC may be lower if you acquire users here, but the opex remains the same. The opex of the tier-1 companies from India that have gone public in the recent past is a lot of offline costs. Hence business models that have not even worked in the US and Europe has crushed it here. In South East Asia, everyone is asking how can we like Swiggy, Zomato and Zepto. Yes, Zoho has also won, but they have been grinding for decades and they win on lower costs.

So in AI, you either stay small, bootstrap and build something like Basecamp. OR go big. Own the underlying asset. Roll up hospitals. Call centres. Cut costs by using AI. Take more risks.

Now everything has changed with deepseek and the step change of lower cost and the ability to adopt open source. Low LTV, price sensitivity of users, and lower free to pay conversion in India doesn’t matter as much when CAC and opex is low too. Yes, you are still competing with the US AI companies, but they will always price higher to compensate for their high CAC. So you can win on cost again.