Starting an AI company is not for the faint-hearted. Even though there is an AI boom and everyone is ready to invest in AI companies, the VC ticket sizes everywhere, when compared to the Silicon Valley in the West, are minuscule.
The situation is far worse in India. Here, nobody is ready to invest in research and development startups.
Recently, AIM made a ballpark estimate of the cost of building an AI research startup in India. For the seed stage, a funding of $5 million or $10 million is a decent amount, but comparatively very less to build something like OpenAI, or even Perplexity AI.
To put this in perspective, Perplexity AI had raised $15 million in its seed funding round when it was just a six-month old company. Besides, Perplexity did not spend time training its own AI model, but instead chose to build a fantastic product around the existing LLMs in the market.
For building an estimated 7 billion parameter foundation model out of India, it is said that the cost for the compute alone would be close to $2 million. Accounting for all of these, an ideal amount to do a lot of foundational work in AI at the seed stage is anywhere around $7-15 million.
Assuming a seed fund of $10 million, an AI startup in India can last for around two years without raising any more funds and not running inference. But for Perplexity, the whole point is to provide inference to customers. Plus, it has not trained any models of its own.
So, how would Perplexity fare in India had the founder not moved to the Bay Area to start it?
Building an AI Startup in India is No Joke
When speaking with AIM, Vishnu Ramesh, the founder & CEO of Subtl.ai, a company which is coincidentally also building a ‘private Perplexity for enterprise’, said that several times, he has also evaluated the thought of starting his company in Hyderabad, instead of the US.
Born in the Bay Area, Ramesh came back to India at the age of three.
“The deeper question is why does the Bay Area get so much money when startups come out,” said Ramesh. He explains that it is more to do with the fact that Google, Meta, or OpenAI have come out of the same place, which gives investors a sense of confidence when investing in startups born there. “Once India has its Google moment, I think the scene should change,” he added.
CEOs such as Aravind Srinivas of Perplexity AI are an inspiration for Indians building AI. His vision is to build a product which is in direct competition to Google. Graduating from IIT Madras and leaving OpenAI to start one’s own company surely takes a lot of courage.
The company is one of the latest to become an AI unicorn with its last funding round raising a $63 million round led by Daniel Gross, NVIDIA, and Jeff Bezos. This points to the fact that even though the seed round was not that high, with the right product, the company was able to make a dent in the market.
This brings us back to what Abhishek Upperwal, the CEO and founder of Soket AI Labs, told AIM. He said that funding scenarios and sizes are just enough to manage for AI research within a startup.
“Yes, there are fewer funds that are available as compared to any foreign markets, but I also believe that we can maybe make-do with that particular fund and then ultimately grow in scale after the seed stage,” said Upperwal.
It is Probably the Right Time to Build in India
For comparison, TWO.AI has released a product called Geniya, which can browse data from the internet using Google. Pranav Mistry, the founder of TWO.AI, said that like Perplexity AI, Geniya is still in public beta and users can try it out, but it has not achieved as much traction as Perplexity yet.
Surprisingly, the company has introduced two products, another being SUTRA, with just $20 million in funding.
Similarly, PAiGPT is another product similar to Perplexity AI’s search. The app’s USP is its ability to fetch real-time information on various topics and current affairs, ideal for the UPSC exam preparation. The company is bootstrapped and has invested around $1.2 million funding till date.
There are several new AI startups researching the field and focusing on building foundational models for AI. Perplexity, for instance, partnered with AI providers to integrate capabilities into its own AI services, which is probably what a lot of India AI companies should focus on.
The only thing that lacks for something like Perplexity AI in India is the proof of concept that something can stem out from here to the world, which is slowly but surely changing with Bengaluru becoming the Silicon Valley of the East.
Moreover, it is also tougher for companies in India to garner support from heavyweights like Jeff Bezos and NVIDIA, which Perplexity had. It is high time for investors to realise that these startups can be built within the country, so that the talent remains within India and builds for India.
