In the run up to the Indian AI Impact Summit, which has drawn over 100,000 registrations, Abhishek Singh, additional secretary, MeitY and CEO of the IndiaAI Mission speaks to Ojasvi Gupta about the country’s approach to AI infrastructure, regulation, data policy and foundational models. Excerpts:
What outcomes should we expect from the AI Summit?
We have crossed 100,000 registrations. A declaration will be issued on February 19th Participating countries will contribute to it.
Is China participating?
Yes, a vice-minister-level delegation is expected.
Is there a particular focus on rural and under-served communities in the AI approach, given PM’s recent interactions with AI startups and tech companies
The focus is on bridging the digital divide. AI should empower those who are currently outside the mainstream digital ecosystem-rural populations and underserved communities.
There is an ongoing debate globally about AI regulation versus innovation. Where does India stand?
We are not regulating AI technology itself. We are addressing harmful use cases. If an AI application enables deepfake-based financial fraud, creates communal disharmony, interferes with democratic processes, facilitates child sexual abuse content, or causes reputational harm, then the government cannot ignore that. For instance, labeling AI-generated content in harmful contexts is about protecting ordinary citizens. These technologies have become so realistic that distinguishing real from synthetic content is increasingly difficult. Citizen protection is the responsibility of the government.
Last year’s India AI Mission allocation was not fully utilised. What is the current status?
Last year, we consciously decided not to procure GPUs directly. Instead of putting government money into buying hardware, we chose a model where the private sector invests in infrastructure and the government subsidies end usage. As a result, private companies have invested in around 38,000 GPUs, amounting to roughly Rs 20,000 crore. Our actual expenditure was around Rs 800 crore, and this year we have received a 25% increase in allocation, taking it to Rs 1,000 crore. This model ensures that capital investment comes from industry, while government funds support startups, researchers and developers who use that infrastructure.
What can we expect from India AI Mission 2.0?
Major global AI investments are being driven by private capital. Even large-scale initiatives abroad like the Stargate announcement are funded by private companies. Governments create enabling ecosystems; they do not try to outspend private players. Our role is to provide policy clarity and confidence, through measures such as AI governance guidelines, the Digital Personal Data Protection Act, the data centre policy, and usage subsidies. That ecosystem encourages investment in AI infrastructure and applications. Details of future steps on AI Mission 2.0 will be announced soon.
What’s the progress of the foundational AI models which have been approved so far?
Twelve foundational models have been approved and are under implementation. Several are preparing launches and announcements. Further approvals depend on infrastructure availability, particularly GPUs.
What is the status for GPU procurement
GPUs are being sourced from multiple providers, including NVIDIA, AMD, Intel, AWS and Google TPU infrastructure. Additional bids will be opened soon.
There is growing discussion around data localisation and sovereign cloud. How does the government view this?
The Digital Personal Data Protection Act clearly defines current data localisation requirements. Laws can evolve, but companies must comply with the law as it stands at any given time. Similarly, the 20-year tax holiday for data centre investments is designed to drive fresh capital into India. The aim is to position India as a competitive AI and data infrastructure hub.