India praises DeepSeek, a Chinese AI lab, and intends to use local servers to house its models

In an unusual move for Chinese technology in India, India’s IT minister hailed DeepSeek’s advancements on Thursday and announced that the nation will host the Chinese AI lab’s massive language models on local servers.

In response to criticism of New Delhi’s own AI spending, which has been significantly lower than that of many other nations, IT minister Ashwini Vaishnaw remarked on Thursday, “You have seen what DeepSeek has done—$5.5 million and a very, very powerful model.”

Citing national security concerns, India has prohibited around 300 Chinese-affiliated applications and services since 2020, including WeChat and TikTok. According to India’s stringent data localization regulations, the platform must store and handle all Indian users’ data domestically for DeepSeek to be approved to be hosted there.

Also Read: How to use DeepSeek R1, a competitor to ChatGPT, on a computer or smartphone

At an industry conference, Vaishnaw stated, “Open-source models on Indian servers can address data privacy issues regarding DeepSeek.”

The new AI computing facility in India is probably where DeepSeek’s models would be housed. Substantially, 13,000 of the facility’s 18,693 graphics processing units (GPUs) are Nvidia H100 GPUs, while over 1,500 are Nvidia H200 GPUs, substantially doubling its original goal. According to the minister, the facility is expected to start operations “in the coming days,” with about 10,000 GPUs ready for deployment.

Additionally, the facility would provide computing services to Indian businesses at great prices. According to Vaishnaw, high-precision computing would be discounted by 47% and regular AI computing by 42% from market rates.

The minister’s comments were made the day after DeepSeek’s app was removed from the Italian app stores of Google and Apple after the data protection authority in Italy stated that it was investigating the Chinese company’s use and storage of personal information belonging to Italians.

Also Read: Chinese startup Moonshot AI’s Kimi k1.5 model outperforms OpenAI-o1 after DeepSeek-R1

This week, the computer sector was rocked by the unveiling of DeepSeek’s R1 “reasoning” model, which was developed on a supposedly minimal budget. As a result, chip giant Nvidia’s market capitalization dropped by $600 billion. The model has rapidly drawn harsh criticism and generated contentious discussions about copyright concerns, U.S. export restrictions, and the need for further funding for AI initiatives.

In addition to hosting AI models from other countries, India is working to promote the development of AI models and associated technologies domestically. According to Vaishnaw, “Major chip designers are willing to work with India to develop indigenous GPUs.”

Over the following two to three years, Vaishnaw predicted that $30 billion would be invested in data centers and hyperscalers in India. According to a Bloomberg story from last week, Reliance, one of the largest corporations in the nation, is to construct what could be the largest data center in the world in the city of Jamnagar. It will have a capacity of three gigawatts. 

At least six significant developers, in our opinion, are capable of creating AI models in four to six months, with a more optimistic estimate of six to eight months. The most crucial element in building a strong AI ecosystem, according to Vaishnaw, is a shared computing facility.

Also Read: Alibaba Reveals Improved AI Model, Says It Beats Competitor DeepSeek-V3

Additionally, the computer center will help India’s larger AI projects. According to Vaishnaw, 18 AI-driven apps with a focus on learning disabilities, climate change, and agriculture have been chosen for early financing.

Using what Vaishnaw called a “hub-and-spoke model,” India would set up a regulatory agency to supervise the growth of these AI efforts, enabling various institutions to work together on safety protocols. He declared, “Our models will remain open and application-focused.”