On Thursday, India’s Information Technology Minister, Ashwini Vaishnaw, expressed excitement about the advancements made by a Chinese company called DeepSeek. He announced that DeepSeek will have its large language models stored and operated on servers within India—an unusual move given the strict policies India has had towards Chinese technology.
During his speech, Vaishnaw highlighted that DeepSeek has managed to create a highly effective AI model with a budget of $5.5 million. This statement came as a response to criticism towards the Indian government regarding its comparatively lower investments in artificial intelligence technology when stacked against other countries.
Since 2020, India has taken significant steps to limit Chinese influence in its tech sector by banning over 300 apps and services that are tied to China, including popular platforms like TikTok and WeChat, due to concerns about national security. However, the new agreement with DeepSeek suggests that to host its services in India, the company has to ensure that it keeps and processes all data from Indian users within the country. This requirement aligns with India’s strict rules on data storage and privacy.
Addressing concerns about data privacy, Vaishnaw mentioned that these worries could be resolved by using open-source models hosted on Indian servers, which would provide a higher level of transparency and security.
DeepSeek’s AI models are expected to run on a new facility in India that’s equipped with a staggering number of graphic processing units (GPUs). This facility will have nearly 19,000 GPUs, which is an increase from the initial goal of about 13,000. The GPUs are high-powered units, primarily from Nvidia, and approximately 10,000 of these are ready for use as the facility is set to begin operations soon.
This facility aims to offer computing services at lower prices than what’s typically available. Vaishnaw shared that standard AI computing services will be offered at a 42% discount compared to the current market rates, while those requiring higher precision will be available at a 47% discount.
His comments came shortly after an important event: DeepSeek’s application was removed from both Apple and Google’s app stores in Italy due to the country’s data protection authority questioning how the company handles Italians’ personal information.
DeepSeek recently released a unique AI model known as the R1 “reasoning” model, which has garnered a lot of attention in the tech world. This release significantly impacted Nvidia, a major player in the chip industry, causing a loss of $600 billion in its market value. The new model has sparked discussions around several critical issues, such as copyright, export regulations, and the need for increased funding in AI development.
Besides allowing foreign AI technologies to operate in India, the government is also keen to boost local development of AI solutions. Vaishnaw mentioned that several major chip designers are interested in collaborating with India to create homegrown GPUs.
He predicted that there would be a $30 billion investment in large-scale data centers and technology over the next two to three years. A notable project includes Reliance, one of India’s largest companies, planning to build what could become the biggest data center in the world, located in Jamnagar, with an extraordinary energy capacity of 3 gigawatts.
Read Also: DeepSeek Hits Top Spot On US PlayStore
Vaishnaw expressed hope that at least six major AI development companies could create new AI models within a timeframe of six to eight months, or potentially four to six months with an optimistic forecast. He stated that a shared computing facility is crucial for building a strong AI ecosystem in India.
In addition to supporting AI development, the computing facility will help advance various initiatives across different sectors. Vaishnaw noted that 18 AI applications targeting areas like agriculture, climate change, and learning disabilities have already been chosen for initial funding.
To manage and supervise these AI projects, India plans to create a regulatory body. Vaishnaw described this as a “hub-and-spoke model,” which means that various institutions will work together to create safety procedures. He concluded by emphasizing the importance of keeping AI models open and focused on practical applications.
