India fluttered a few feathers when it launched the ambitious IndiaAI Mission on March 7, 2024 – two months after China’s AI chatbot DeepSeek-R1 was launched. A knee-jerk reaction or a well-thought-out strategy, it’s hard to tell. But most experts called it the much-needed nudge for the Indian administration to give the country an AI edge.

Despite the ₹10,372-crore IndiaAI Mission significantly boosting national compute capacity, the country still faces capital limitations and a shortage of deep, high-end AI research needed to compete with global leaders. Experts, too, have repeatedly raised concerns over the actual impact of the mission. Yet the government maintains that a homegrown DeepSeek-class model is within reach, citing its rapid distribution of thousands of top-tier chips to startups, universities, and researchers — aimed at enabling a breakthrough model in under 10 months.

According to Stanford HAI’s Global AI Vibrancy Index, which ranks countries on metrics such as patents, funding, policy and research, India lags far behind the US and China in many key areas. Both China and the US already have a four to five-year head-start, having made heavy investments in research and academia and developed AI for military applications, law enforcement and homegrown large language models (LLMs). 

Plus, India’s AI mission is worth a trifling $1bn compared with the staggering $500bn the US has earmarked for Stargate—a plan to build massive AI infrastructure in the US, or China’s reported $137bn initiative to become an AI hub by 2030.

In such a scenario, how fast the country gets its own LLM is a question everyone seeks an answer to.

Not Quite Enough

While the GPU capacity is secured with the Mission already pooling 34,000 GPUs of the planned 38,000 GPUs, the actual usage of these high-end machines has been slow to take off. Excluding the handful of startups developing foundational models, there is a noted lack of large-scale, high-end research projects that require continuous use of massive GPU clusters. This indicates a mismatch between the infrastructure now being deployed and the research community’s ability to leverage it effectively.

Raising serious concerns about various aspects of the mission, serial entrepreneur K Ganesh, who is also the promoter of Bigbasket, Bluestone, Homelane, and Portea Medical, said that so far, none of the programs of IndiaAI Mission have shown results. “There haven’t been any models produced from GPUs from the grants. Almost all of the grants are purely computational. Hence, [I] do not see any impact yet.”

He agreed that while the GPU capacity is secure, the actual usage of these machines has been slow to take off, “because many GPU providers are new datacentres still facing teething issues.” According to Ganesh, GPU compute is only one piece of the puzzle. For India to reach a globally competitive level, the country’s top players will need to raise their game across talent, operational maturity, and risk appetite, he said.

Challenges remain, particularly in achieving population-scale AI adoption and developing indigenous LLMs, said Anurag Dua, partner, government and public sector, EY India. “Above all, a focused and coordinated R&D effort, backed by adequate funding, is essential to spur innovation and create affordable technologies capable of scaling across India,” he added.

Dua advised that in the long run, a balanced approach, consisting of industry-led innovation and government-led adoption for the public good, would be the right approach to help achieve the economic and social objectives.

Another area where the mission has struggled to make progress is closing the AI talent gap. A 2024 Deloitte-Nasscom report, while declaring that India is positioned to become a global AI powerhouse by 2030, states, “However, to truly harness this potential, the focus must shift towards not just quantity but the quality of AI talent. By reskilling the existing workforce and fostering new talent through robust government-academia-industry collaboration, we can ensure a steady pipeline of professionals ready to lead AI-driven innovation.”

The Talent Tussle

Despite initiatives to broaden AI education in India, a substantial skills gap remains, with only a small portion of the workforce trained in advanced AI. Low broadband penetration in rural India further widens this divide, limiting both the reach and equitable adoption of AI solutions nationwide.

Considering the talent gap, the IndiaAI Mission launched the FutureSkills pillar to expand AI education at multiple levels. The initiative has reportedly enrolled about 8.6 lakh candidates in AI and related skilling/training programmes. There’s also a plan to build 570 AI & data labs in non-metros to decentralise education and AI infrastructure. However, there’s no official figure as to the completion of these labs.

Shortage of deep‑AI researchers, PhD programs, and high-level mentorship persists outside elite institutions. Abhishek Upperwal, CEO and founder at Soket.AI, one of the startups selected by the IndiaAI mission to build LLMs, said, “There are very few people in India with firsthand experience building 100-billion or trillion-parameter models. Most of those experts are abroad, and we can’t realistically expect access to them for developing our own LLMs. What we can do is build capacity step by step.” He suggested that a researcher who has experience training a 1-billion-parameter model and understands language modelling can, over time, figure out how to scale to 100-billion-parameter systems.

“Even in my own team, no one has built a 100-billion-parameter model before. We began with a 1-billion model, we’re currently working on a 7-billion model, will move next to a 40-billion model, and eventually graduate to 100 billion,” Upperwal said.

Experts have also raised concerns over the absence of a dedicated, comprehensive AI-specific data protection law, raising questions about how the mission will ensure the ethical and secure use of large, sensitive datasets being developed under the AIKosha platform. Though the DPDP Rules, 2025, were notified on November 14, bringing the framework to life, it is said to be less stringent and comprehensive compared to “gold standard” regulations like the European Union’s General Data Protection Regulation (GDPR). Critics debate its effectiveness due to broad government exemptions and a less independent regulatory body.

The Story So Far

Apart from onboarding 12 startups across two phases and enabling their access to compute, datasets, and incubation support, the IndiaAI Mission has also carried out several early groundwork initiatives.

It has acquired 34,000 GPUs, empanelled service providers to host and manage the hardware, rolled out data/AI labs, training programmes, and fellowships to grow AI talent nationwide. The AIKosha platform — one of the seven pillars of the mission — has been made operational, providing researchers access to datasets and models.

Work on tools, frameworks, and oversight mechanisms for responsible AI deployment have begun, though details are not out in public yet.

While the intent of the Mission seems to be in the right place, the lack of a pan-India strategy, resource allocation, and timely innovations may prove to be a dampener in achieving the overall impact.

“Because something of this scale and type is being tried for the first time, the processes weren’t known to anyone. Initially, there was a lot of confusion. I hope things get better and smoother with time,” Upperwal said.IndiaAI Mission did not respond to AIM’s queries until the time of publishing.

The post A Year Later, Is IndiaAI Mission Delivering Enough? appeared first on Analytics India Magazine.