India lauds Chinese AI lab DeepSeek, plans to host its models on local servers

India’s IT minister, Ashwini Vaishnaw, commended the progress of DeepSeek on Thursday and announced that the country will host the Chinese AI lab’s large language models on domestic servers, marking a rare collaboration between Indian and Chinese technology.

Highlighting the success of DeepSeek’s powerful model and $5.5 million investment, Vaishnaw defended India’s AI investments, which have faced criticism for being comparatively lower than other countries.

Despite India’s ban on over 300 Chinese apps and services since 2020, including TikTok and WeChat, DeepSeek’s hosting in India is contingent on adhering to strict data localization requirements, ensuring all Indian users’ data is stored and processed domestically.

Vaishnaw proposed addressing data privacy concerns by hosting open-source models on Indian servers, with DeepSeek likely to be hosted on India’s new AI Compute Facility equipped with a large number of GPUs.

The facility will provide computing services at discounted rates to Indian firms, with standard AI computing offered at a 42% discount and high-precision computing at a 47% discount.

Following the controversy surrounding DeepSeek’s R1 model and its impact on Nvidia’s market cap, India is also focusing on developing indigenous AI models and technology, collaborating with major chip designers for GPU development.

Vaishnaw anticipates significant investment in hyperscalers and data centers in India, with plans for a massive data center in Jamnagar by Reliance. The country aims to foster a robust AI ecosystem through a common compute facility and support for AI-driven applications in various sectors.

To oversee AI initiatives, India will establish a regulatory body using a collaborative “hub-and-spoke model,” ensuring open and application-focused development of AI technologies.