On Thursday the Minister of India praised Deepseek progress and said that the country will host the great linguistic models of the Chinese laboratory to the national servers, in a rare opening for Chinese technology in India.
“You saw what Deepseek did – $ 5.5 million and a very very powerful model,” said Minister Ashwini Vaishnaw on Thursday, responding to the criticisms that New Delhi received for his investment in AI, which was much less than Many other countries.
Since 2020, India has forbidden more than 300 apps and services related to China, including Tiktok and WeChat, citing national security problems. The approval to allow Deepseek to be hosted in India appears according to the storage and processing platform of all Indian users at national level, in line with the rigid location of the data of India.
“Data privacy problems regarding Deepseek can be addressed by hosting open source models on Indian servers,” said Vaishnaw in a sector conference.
Deepseek models will probably be hosted on the new AI calculation structure in India. The structure is powered by 18,693 graphic processing units (GPU), almost double its initial goal – almost 13,000 of these are NVIDIA H100 GPUs and approximately 1,500 are NVIDIA H200 GPU. About 10,000 GPUs are ready to be lined up and the structure should start the operations “in the next few days”, according to the minister.
The structure will also offer elaboration services at steep discounts to companies in India. Vaishnaw said that the computer science would be offered with a 42% discount on market rates and high precision calculation would be discounted by 47%.
The minister’s observations come a day after the deepseek of the same name was removed from the App and Google stores in Italy, after the data protection regulator that country said he was asking how the Chinese company was using and storing the personal data of Italians.
The release of the “reasoning” R1 of Deepseek, based on a presumably modest budget, has sent shock waves through the technological industry this week, causing the drop in the market of the Nvidia chip of $ 600 billion. The model quickly had intense control and aroused bright debates on copyright issues, US export checks and how even more money should be paid in artificial intelligence efforts.
In addition to hosting foreign artificial intelligence models, India is also trying to guide the development of artificial intelligence models and technology related to its turf. “The main chip designers are willing to work with India to develop indigenous GPUs,” said Vaishnaw.
Vaishnaw estimated that India would see investments of $ 30 billion in hyperscaler and data center in the next two or three years. One of the largest conglomerates in the country, Riliance, is planned to build what could become the largest data center in the world in the city of Jamnagar, with a capacity of 3 Gigawatt, reported Bloomberg last week.
“We believe that there are at least six important developers who can develop artificial intelligence models in six-8 months on the external limit and four to six months on a more optimistic estimate. A common calculation structure is the most important component for the creation of a robust artificial intelligence ecosystem, “said Vaishnaw.
The computer structure will also support the largest artificial intelligence initiatives of India. Vaishnaw said that 18 applications were selected based on AI focused on agriculture, climate change and learning difficulties for initial funding.
To supervise the development of these artificial intelligence initiatives, India will establish a regulatory body using what Vaishnaw has described as a “Hub-and-Rake model”, allowing several institutions to collaborate with safety paintings. “We will keep our models open and focused on the application,” he said.
