India’s AI leap: bridging the gap – Crypto News

The Cabinet meeting of 7 March, chaired by the PM, which approved the comprehensive national-level IndiaAI mission with a budget outlay of ₹10,371 crore, should allay the fears. The money allocated to help build the all-critical GPU infrastructure through a public-private partnership model addresses the infrastructure gap that many feared would hold back India’s AI efforts. It completes the building blocks that are needed to make India a hub for AI research and innovation.

Three things are crucial for progress in AI research. One is AI talent – which the government is trying to address by getting the best minds in academia and private sector involved. The IndiaAI Innovation Centre, which will undertake the development and deployment of indigenous LMMs, and the earlier announced three AI Centres of Excellence in premier technology institutes will pull together the AI brainpower required for the research projects.

The second critical requirement is humungous amounts of data and here India starts off with an advantage. The country generates the second largest amount of digital data in the world, only behind China, because of Digital India initiatives, mobile broadband affordability and availability and finally the numerous social media and other platforms on which citizens communicate and transact daily. The AI Mission will create the IndiaAI Datasets Platform to make non-personal data easily available for AI research and innovation.

The third crucial thing is computing power – and this was where India had a disadvantage for long. AI research requires extremely high computing power, with tens of thousands of the latest and most powerful GPUs harnessed together. The US chip company, Nvidia, produces the most powerful AI-focused GPUs and has 80% market share. Its latest generation chip – the H200, unveiled a few months ago – costs between $25,000 and $40,000. (Its rivals AMD and Intel have the remaining 20% but their chips have lagged Nvidia’s in power).

Global AI firms such as OpenAI, Google, Meta, Amazon, Microsoft, Anthropic and others have raised resources and not only taken up most of Nvidia’s AI-focused GPUs but also booked its future products. This year, the capital spending on AI compute infrastructure by Silicon Valley Tech giants is estimated to cross $32 billion. 

Nvidia, however, also offers its own GPU-cloud infrastructure, the Nvidia DGX cloud, which is an AI supercomputer in the cloud to those who do not wish to build their own compute infrastructure. So, for India, the choice would be to hire compute capacity in one of the US-based AI compute infrastructure that is available.

Building a domestic, highly scalable national AI compute infrastructure with latest generation GPUs, which would be available to the Indian AI research ecosystem as well as start-ups, and which offers AI as a service as well as pre-trained models, will fill this gap, providing the necessary condition for Indian innovation to flourish. 

The public-private partnership model that the government is proposing is likely to work particularly well as two big business groups are already partnering Nvidia to build AI compute infrastructure in the country and they would be interested in working with the government because it would provide immediate utilization for any infrastructure built. 

The ₹10,372 crore that the government has budgeted may need to be augmented over a period of time but it is enough to start with now.

Of course, the Cabinet clearance and the allocation for AI compute infrastructure will not automatically make India an AI super-power. The AI talent for research and innovation needs to reach a critical mass. The AI compute infrastructure will need some time to be built and become available for researchers and start-ups. 

Most important of all will the regulations and laws that now need to be drafted in detail and passed. This is where considerable attention and effort of policy makers will be needed now. For example, there is a need to ensure that non-personal datasets are clean and available without too much bureaucracy to the researchers, innovators and start-ups. 

Apart from the Digital Personal Data Protection Act (DPDP Act), that is already in place, the lawmakers will need to pass more AI data regulations, ensuring that these maintain the delicate balance between being available to innovators while preventing them from being used for malicious purposes. 

Finally, the rules for Safe AI, Responsible AI and Ethical AI will need serious discussions and debate before being passed.

But well begun, they say, is half done. And a good beginning has been made by the AI mission that was cleared.

Mahesh Makhija, Technology Consulting Leader, EY India and Rakesh Kaul Punjabi, Partner, Technology Consulting, EY India

Milestone Alert!
Livemint tops charts as the fastest growing news website in the world 🌏 Click here to know more.

Unlock a world of Benefits! From insightful newsletters to real-time stock tracking, breaking news and a personalized newsfeed – it’s all here, just a click away! Login Now!

Published: 11 Mar 2024, 12:50 PM IST