India is a country driven by ambition, not just in pursuing significant goals, but in choosing to achieve them through a path of self-reliance.

From ISRO’s remarkable achievements to the development of a powerful nuclear arsenal and the launch of the ‘Atmanirbhar Bharat’ and ‘Make in India’ initiatives under Prime Minister Narendra Modi, the baton of self-reliance has been passed on across generations and sectors. 

However, the pathway for any state or organisation to make significant advances in these technological fields begins with research. 

Take the IndiaAI Mission, for example. Over ₹10,000 crore has been allocated for innovative startups to develop AI models, using advanced hardware. 

While capital and access to resources can lead to tangible results, making these outcomes truly worthwhile, especially if they are to outperform or at least compete with Western technology, requires substantial research. 

The Case for Research

“It’s pretty amazing that so many people in the tech industry and the tech press don’t understand the difference between research, technology development, and product development,” Yann LeCun, chief AI scientist at Meta, said in a post. 

According to him, product development typically has a horizon of three to 12 months, while tech development has a horizon of one to two years. In contrast, research has horizons spanning two to 10 years, sometimes even longer.

“I work on research. The stuff I focus on tends to be three to five years ahead of what AI pundits are currently obsessed with,” he added, reinforcing how research is key to building future-proof technology. 

China’s DeepSeek serves as a compelling example. Its success wasn’t celebrated simply because it was “homegrown” technology; it earned global recognition because it genuinely outperformed Western models through innovative architectural breakthroughs.

So, how is the country’s research ecosystem positioned to make such achievements today? 

While startups selected by the IndiaAI Mission are being led by some of the brightest minds in the industry, several argue that the initiative, for now, is mainly focused on proving that India can build LLMs rather than creating a functional and useful one that can solve legitimate problems in the country. 

While the innovations under AI models today, largely including the Transformer architecture, were built in Google’s research arm, a similar situation cannot be expected with Indian tech companies. 

Many large IT service companies have in-house AI R&D labs and venture funds. Yet, these efforts are often focused on enhancing their enterprise services and client solutions rather than developing cutting-edge models. These IT firms invest in R&D, but their core business remains providing services. 

Read More: Indian IT Doesn’t Seem to Care Enough About IndiaAI Mission

This brings us to those whose core mission is research—the country’s leading academic and research institutes. Here, many institutes undertake research projects aimed at building technology with a less myopic intent. 

While the IndiaAI Mission provides the necessary resources for selected startups to build AI models, it is essential to focus on premier institutes in the country, where some of the brightest minds have emerged. 

The Need for Top-Tier Research Institutes

“I think the biggest constraint for research in India is funding,” said Balaraman Ravindran, who is the founding head of the Wadhwani School of Data Science and AI (WSAI) at the Indian Institute of Technology, Madras (IIT-M), in an interaction with AIM. 

“If I need to have a student who is going to run highly compute-intensive experiments, I basically have to find a collaborator in Google, OpenAI or Microsoft to run those experiments for us,” he added, citing the challenges with infrastructure supply in India while acknowledging the efforts of the government in subsidising GPUs. 

In an earlier interview with AIM, Ravindran highlighted this point. “We don’t really have any university to research at the scale of a Tsinghua (University) or an MIT.” 

“Even though our corporates make a lot of money, they don’t invest the kind of money into fundamental research. They’re happy to do more applied research. So it’s not just a question of blaming academia or industry; it’s all around,” he added. 

For example, in 2023-24, IIT-M received funding of ₹584.87 crore for various state-sponsored projects. This was in addition to the ₹513 crore raised that year from alumni, corporate partners, and donors. However, the total amount still falls short and appears modest compared to the funding received by leading universities in the United States and China. 

During the same period, China’s Tsinghua University received approximately $5.43 billion (~₹46,480 crore), while the Massachusetts Institute of Technology (MIT) in the United States received around $2.04 billion (~₹17,460 crore).

“We seem to have put our bets on things with more short-term turnaround, where we can see the results quickly, rather than investing in long-term research,” Ravindran said. 

Regarding the talent threat in India, Ravindran said that perhaps it is predicated on the assumption that the country lacks sufficient resources to actually instruct students on how to build a model using GPU credits.

“By the time you’ve learned to develop a good model, I will likely have exhausted all my GPU credits for the lab, whether for the month or the year,” he added. 

Ravindran also pointed out that while India has high-quality institutions such as the IITs, BITS, and several IIITs, the country still lags in terms of scale. 

He added that regions such as California and countries like China have many more institutions functioning at a similar level. For example, China probably has over a thousand such schools. 

“We are fairly lagging behind in the volume, not the quality of people,” he added. 

In the QS World University Rankings 2025 for Engineering and Technology, only six Indian institutes—namely IIT Delhi, Bombay, Madras, Kanpur, Kharagpur, and IISc Bangalore—are in the top 100. In contrast, more than 20 institutes from the United States are ranked in the top 100. 

Currently, India has a total of 23 IITs, far fewer than the number of top-tier universities present in a country like the United States. 

There are 8 Ivy League colleges in the United States, and as of 2025, 187 U.S. institutions have achieved the R1 (Research 1: Very High Research Spending and Doctorate Production) status by the Carnegie Classification of Higher Education, recognising them for very high levels of research activity and doctorate production.

That said, today, many of India’s top-tier universities have undertaken efforts in both research and the development of AI and machine learning. 

Besides, Ravindran observed that much of the emerging talent in India is developing expertise through on-the-job learning. He highlighted that people across the country are doing remarkable work in building various AI models and use cases. 

Still, most of this learning happens in the course of their professional work rather than through formal research pathways. 

“That basically stops the research pipeline somewhere because they don’t get into the research pipeline. They are busy building their products and talking to customers,” Ravindran said. 

He also pointed out that countries such as China began focusing on AI more than 10 years ago, aiming for long-term outcomes. 

“This was maybe 15 years ago—the Chinese premier was already talking about AI. And I recall seeing him with (Stuart) Russell and (Peter) Norvig’s textbook (Artificial Intelligence: A Modern Approach) in the background—not some pop-science book on AI, but an actual textbook on AI. That’s a signal. They were thinking about AI in serious terms way before most other countries caught on,” Ravindran said. 

“Of course, the US, Europe, and Canada have also invested a lot, but the kind of focused investments that China made, that’s what’s allowed them to catch up.”

The post India Needs More IITs Than AI Startups appeared first on Analytics India Magazine.