In the fortnight since commerce and industry minister Piyush Goyal’s address at the Startup Mahakumbh, we have seen a range of comments, rants, potshots and some hot takes — everyone has an opinion on why the Indian startup ecosystem has not cracked the deeptech code.

Before going deeper into the issue — what exactly is deeptech and why is it such a hot topic right now after all?

There is no hard definition for deeptech, and this is perhaps why it’s been so poorly written about in recent days. But we will try to break it down.

As far as Inc42’s thesis is concerned, deeptech governs areas that are considered pure-play tech — the stuff that becomes intellectual property (IP) aka the building blocks for the tech industry at large. But the definition varies depending on who you speak to.

A lot of new-age tech is crammed under the deeptech umbrella — from foundational models to proprietary artificial intelligence and machine learning algorithms, to original IP creation in robotics, industrial systems, electronics systems and chip design, green energy and climate tech solutions, healthcare systems, aerospace (drones and spacecraft), connected and autonomous vehicles, and everything that can be clubbed together as futuristic.

Some call it frontier tech, others deeptech and many others simply AI.

Indeed, within the sprawling landscape of deeptech, the incredible rise of generative AI and consequently the fortunes of model makers such as OpenAI and Google have sharpened the focus on AI and GenAI models. And it is arguably this gold mine that India is missing out on right now. In many ways, AI models will shape the various other segments within deeptech.

So India needs its own GPT or Gemini goes the cry. And that’s where we are today with the whole debate on sovereign models.

Deeptech In A Nation Of Consumers

One doesn’t need to think too long to know that Indian startups are perhaps not the frontrunners in the deeptech innovation race. Instead, as Goyal lamented, the focus has been on building businesses that are adjacent to retail business models — dukaandari or running a shop, as the minister phrased it.

Let’s not talk about specific companies here, but the largest Indian startups by valuation and market cap are not companies that have created the fundamental pieces of the internet ecosystem, but those that have leveraged internet access to offer services through apps and software products. In many cases, they were simply clones of Silicon Valley startups that set the trend.

No one can fault founders for looking for large outcomes in the consumer market — India is after all the most populous country in the world, and that means the most number of consumers in any one country. If good business means solving big problems, then consumer startups definitely targetted the largest audience base.

Nothing wrong with that approach, and not everyone can be expected to build the next AWS or Google or Nvidia or OpenAI, some do need to build a Zomato or Flipkart or Swiggy to sell to the masses. The other part of the complaint around copycat or clone startups is somewhat more valid.

Having said that, it’s not like deeptech is dead in India — it’s been slow to take off unlike and too timid compared to other sectors. And it’s worth examining why and what needs to change to fix that.

We can debate endlessly on whether Goyal’s description is perhaps a bit unfair on some Indian startups that have become global examples in this very dukaandari. But we cannot say that he was completely wrong either.

Amid all this, there’s a whole lot of finger-pointing and blame games going around. But time to go beyond that and see where India truly stands.

A short aside: The timing of Goyal’s remarks cannot be overlooked. It coincided with sweeping changes in the US with regards to GPU exports and a tariff-led trade battle with China and other nations. Naturally, this created a sense that the world of AI was going to be split according to geopolitical alliances and territorial maps.

Goyal made a number of visits to the US earlier this year to discuss the tariff situation, and it seems that it is largely in this context that the issue of India’s deeptech capability has taken on a greater significance.

A bilateral trade agreement has been proposed by India, but at the moment, this is not close to fruition. In the event that such a trade agreement is not struck, India would be very much left on its own when it comes to AI. And hence the talk of sovereign AI.

But, what even is sovereign AI?

Sovereign Or Global? The Trillion Dollar Question

It sounds incongruous — sovereign being a term of French origin from the early 14th century and AI of course being quintessentially 20th century — but there’s a reason this has become en vogue.

“If you had asked me just six months ago, I would have said there’s no need for India to develop a ‘sovereign model’, but a lot has changed in that time. And now I definitely believe that we need to take control of our destiny,”........

© Inc42