India needs AI that no one else can switch off: Andrew Ng

India should build its sovereign artificial intelligence (AI) strategy ensuring no outside entity can cut off its access to the technology, said Andrew Ng, managing partner at AI Aspire and adjunct professor at Stanford University. He argued the country can pursue such a goal through both homegrown models and technology developed elsewhere.

In an interview with The Economic Times Digital, Ng said supporting open-weight models, whose trained parameters are available for users to download and run, offer an economical way to secure continued access. He also backed India’s efforts to develop its own models, including those built by Sarvam AI and others.

“What most nations need is not to own every piece of AI, but ensure no one can deny them access,” he said.


Ng’s argument places “continued access” at the centre of the sovereignty debate. In his view, the country where a model was developed matters less than whether governments and businesses can keep using it without depending on another entity’s permission.

“So whether India built it or not, either way seems fine. Investing in building it seems fine,” he said, expressing support for open-source and open-weight models.

Ng said the growth of India’s AI model ecosystem is encouraging. Asked about the push for homegrown technology and greater self-reliance, he specifically welcomed Sarvam’s work.

“I’m very encouraged by Sarvam’s progress. I think it’s a great effort,” he said.

Other countries are also investing in research and building expertise in training models, according to Ng. However, nations with fewer resources than India are relying more heavily on open-weight models, which he sees as an inexpensive way to reduce the risk of losing access.

He referred to open-source networking software powering the internet to explain the value of collectively-developed technology. The contributions from developers across countries, he said, created software others could continue using without its creators withdrawing access.

Ng argued international support for open models could serve a similar purpose, allowing countries to benefit from research and development beyond their own borders while retaining access to the resulting technology.

“When the global community comes to fund research and fund development of these open models, it gives everyone equal access that no one can take away,” he said.

Asked what India should prioritise to capture AI’s benefits and manage its risks, he placed training before everything else.

“There is a lot India could do, but if I had to prioritise one thing, it would be training people and giving them the skills to build, use and understand AI,” he added.