Inside IAIRO, India’s ISRO for AI

On December 30, last year, the Gujarat government approved something India has been debating for more than a decade, but never quite managed to build: a serious national AI research institution. The Indian AI Research Organisation (IAIRO) began operations on January 1 as a Section 8 non-profit inside the GIFT City in Gandhinagar. 

Despite the low-key launch, the intent is anything but small.

IAIRO is being set up as a public-private partnership between the state government, the Indian government through the IndiaAI Mission, and the Indian Pharmaceutical Alliance (IPA).

For the first five years, IAIRO will operate on a budget of roughly ₹300 crore, split equally between the state, the Centre, and the private partner. IPA alone is putting in ₹25 crore in the first year. The pharma group includes companies like Cipla, Torrent and Sun Pharma. 

This is not a startup incubator. It is not a skills programme. 

IAIRO is being positioned as a national research institution for AI, closer in spirit to ISRO than to anything India has tried in software before.

The official note describes it as a multidisciplinary AI hub that will conduct research, build products, generate IP, train people, and work with industry and startups. It will run on a hybrid compute model, using its own GPUs along with India AI Cloud. 

Not Just About LLMs

On paper, it could sound like another government AI centre. What makes it different is who is behind it and what they are trying to fix.

Amit Sheth, a professor of computer science and engineering at the University of South Carolina, who is leading IAIRO, has spent decades in AI research in the US. He said the idea did not start in Gandhinagar. It started two years ago in Delhi, in a meeting with Prime Minister Narendra Modi. 

Sheth had taken the proposal for Ekagrid University that was being coordinated by Shail Kumar and involved some of the top AI leaders of Indian origin, for developing a world-class research university with an AI-first strategy. That plan did not go through. And the problem did not go away. 

India produces a large number of AI users and service engineers. It does not produce frontier AI researchers or people who can build deep systems at scale.

Sheth is blunt about the gap.

“Jensen Huang has pointed out that nine of the world’s top 10 AI institutions are in China, over 70% of papers at leading AI conferences now come from China, and that supports ecosystems with deep research capability capable of producing organisations like DeepSeek. India currently lacks this depth. IAIRO exists to deliberately build it,” he said.

That sentence captures what IAIRO is really about. This is not about chatbots or coding tools. It is about whether India has the kind of research base that can create the next DeepSeek—not just use it.

An ISRO Moment for AI

From those early discussions, a position paper titled Sovereign AI for India’s Strategic Autonomy made its way to key decision-makers, including S Krishnan, secretary at MeitY; Abhishek Singh, CEO of IndiaAI; and Abhay Karandikar, secretary at the science and technology department. What followed were three goals that now define IAIRO.

The first is talent. Not broad skilling, but building a deep pool of AI researchers, scientists, and system builders capable of genuine frontier work.

The second is original IP: foundation models, applied platforms, and domain AI systems that can become Indian-origin global products.

The third is what Sheth calls a prototype-to-product ecosystem—turning papers into companies.

That is why IAIRO is being set up as a national platform rather than a single lab. Sheth describes a growing network that already includes senior researchers with h-index between 70 and 120, technical leaders from Google, Apple, DeepMind, OpenAI, IBM Research and Amazon, a reverse brain-drain cohort of Indian researchers returning or working part-time from abroad, founders who have built IP-driven companies, and investors across stages.

Layered on top of that is a formal programme called From Breakthrough to Breakout, bringing together entrepreneurs-in-residence, applied labs, and venture creation tracks.

Along with Sheth, IAIRO was founded by prof Ramesh Jain, the founding director of UC Irvine’s Institute for Future Health; prof Dev Niyogi, UNESCO Chair for AI, water and cities at the University of Texas at Austin; prof Sanjay Chaudhary from Ahmedabad University; investor and entrepreneur Juhi Bhatnagar; and Selvam Velmurugan, senior technical advisor at BlinkRx.

The board includes Ajai Chowdhry, co-founder of HCL; Sharad Sharma, founder of iSPIRT Foundation; IndiaAI’s Singh; and P Bharathi, secretary of the science and technology department.

“The ambition is not incremental improvement. It is to create, for AI, what ISRO created for space: a concentrated national capability that compounds talent, IP, and execution over decades,” Sheth said.

A Different Model Than the US & China

The first real test of that ambition will come in the sectors IAIRO has chosen to start with.

Unlike big labs in the US that chase general-purpose models, IAIRO is deliberately starting with enterprise and mission-critical AI. The logic is simple: India cannot afford to burn billions training broad LLMs that do not solve local problems.

“From day one, we are building AI systems that can power discovery, decision making, and operations inside strategic sectors,” Sheth said.

The flagship domain is pharma, which is exactly why the IPA is the anchor partner. Indian pharma already spends more on R&D than any other domestic industry, but it remains largely stuck in generics. IAIRO is bringing in people who have built AI systems for drug discovery, clinical trials, and intelligent manufacturing in the West.

“This programme spans the full stack: from AI-enabled discovery platforms, to clinical trial intelligence, to manufacturing optimisation and quality systems,” he said.

The second pillar is sustainability, and what Sheth calls ‘Digital Earth’. This is about climate, weather, agriculture, energy, and resilience.

India’s economy is intensely weather-sensitive. Floods, fog, heatwaves, and shifting rain patterns directly hit infrastructure, crops, and supply chains. IAIRO wants to build AI systems that can deliver hyperlocal prediction, digital twins of cities and ecosystems, and decision tools that governments and companies can actually use.

“Future proofing infrastructure and investment risks to weather and climatic extremes is possible with AI-based novel weather and environmental predictive tools,” Sheth said.

The third pillar is health. The idea here is not just hospital AI, but continuous, everyday health, starting with crises like diabetes.

“IAIRO is reinventing healthcare, replacing episodic care with continuous, everyday health support that helps people understand what is happening, adapt gradually over time, and engage with medical care more effectively,” he said.

Behind all of this sits IndiaAI. IAIRO is being created as a Focused Research Organisation under the IndiaAI Mission, again using a PPP model. The compute, policy backing, and national coordination will come from there.

“We consider ourselves lucky that the Indian Pharmaceutical Alliance stepped in as an inaugural private partner,” Sheth said, noting how reluctant Indian corporates usually are to fund real research.

Working From -1 to 0

One of the most misunderstood parts of IAIRO is its startup plan. It is not an incubator like the ones at IITs. It does not write early cheques to every idea.

IAIRO is designed to work at what Sheth calls ‘-1 to 0’. The emphasis starts with ideas that are nationally relevant—essentially solving sovereign problems—and only then moves to the technical depth of the solution.  

This happens before a company even exists. Founders are selected first. Problems are discovered and validated. IP is created inside IAIRO. Only then are companies spun out, with deep technical and ecosystem backing already in place.

“We operate at −1 to 0, not 0 to 1,” he said. “We focus on creating a small number of deeply technical, AI- and IP-led companies rather than supporting large volumes of early startups.”

There will also be a second funnel for existing startups that need serious technical reinforcement to reach a breakthrough level.

IAIRO will not usually fund TRL 1 to 3 startups with equity cheques. Instead, it absorbs the risk by paying for research, people, infrastructure and validation until the science is real enough for venture capital.

It is a very different model from most Indian programmes that hand out grants and hope for the best.

So, is IAIRO the DeepMind of India?

Sheth says no—not in the near term, at least.

“DeepMind operates at a completely different financial scale. While we have high ambitions, the current level of investment in IAIRO is far more modest, multiple orders of magnitude lower than Big Tech, and nowhere close to even relatively new players like DeepSeek,” he said.

That is why the focus is on mission-critical enterprise AI rather than generic models. It is also why the goal is to start reversing the brain drain over a long cycle, not overnight.

Creating World Class Talent

Another important reality is that India is significantly behind and has real distance to cover. China began its Thousand Talents program nearly 15 years ago, and today an estimated 80-90% of Chinese students and experts trained in the US have returned to build in China. 

IAIRO is an early step towards creating a similar momentum for India—providing a serious institutional pull to start reversing the flow, with the clear understanding that this is long-cycle work. Better late than never.

Why does India publish so little cutting-edge AI research? Sheth argues it is not a talent problem. It is an ecosystem problem.

“Breakthrough research comes from institutions, not individuals,” he said. “We did our undergraduate education in India, but our serious research training, PhDs, and early high-impact work happened in the US. Not because Indians are better there, but because the ecosystem there makes excellence far more likely.”

Over the last few years, Sheth’s group in the US has published 40-60 papers annually. “More than half had Indian students working as interns with my team as co-authors. Many of them went on to do a PhD with me or join other top, fully funded PhD programmes,” he said.

IAIRO is trying to recreate that loop inside India with funded labs, strong mentors, ambitious peers, and a culture of publishing and building.

The government partnership is what makes this possible. It also creates risk.

“The risk is bureaucracy. The upside is nation-scale capability,” Sheth said.

Private capital and most Indian corporates simply are not wired to bankroll research that takes a long time to pay off. That kind of long-horizon ambition has always belonged to the state. But IAIRO is trying to keep things sharp, running as a hybrid with a third of its funding coming from private sources.

What Makes Sheth Think It Will Work?

Sheth believes this model can actually hold, as the foundation is already being laid with unusual seriousness.

“IAIRO has already attracted exceptional expertise across every role required to build a world-class AI research and deep-tech ecosystem,” he said, pointing to senior researchers with global publication records, engineers who’ve worked inside the biggest AI labs, experienced programme leaders, and an embedded investment network designed to move ideas from labs to real deployments.

On paper, the Gujarat government may have announced IAIRO as another PPP in GIFT City. In reality, however, what is actually taking shape is something far more ambitious. It is a bet that India can finally build a home for serious AI, where frontier research, real-world product building, and national missions don’t live in silos but under one roof. 

Whether ₹300 crore is enough to pull that off is still up for debate. But what stands out is the intent. For the first time in a long time, India’s ambition is not modest.

The post Inside IAIRO, India’s ISRO for AI appeared first on Analytics India Magazine.