Summary: Union Minister for Electronics and Information Technology, Ashwini Vaishnaw, has confirmed that the AI assistant DeepSeek will be hosted on Indian servers. This decision aligns with India’s strategic focus on enhancing domestic AI infrastructure and promoting data sovereignty.
DeepSeek, developed by a Chinese startup, has garnered international attention for its cost-effective and efficient AI capabilities. The model was developed in just two months for under $6 million using Nvidia’s H800 chips. It has achieved significant success, even surpassing OpenAI’s ChatGPT in downloads on Apple’s App Store.
Minister Vaishnaw praised DeepSeek’s rapid development and affordability, highlighting it as an example of frugal innovation. He noted that the use of “brain” in developing such models is commendable, emphasizing the importance of efficient resource utilization in AI development
This move to host DeepSeek on Indian servers is part of India’s broader initiative to bolster its AI capabilities under the IndiaAI mission, which involves a $1.25 billion investment aimed at promoting AI startups and developing domestic AI infrastructure.
By hosting DeepSeek locally, India aims to ensure better data control, enhance security, and foster the growth of its AI ecosystem. This aligns with the government’s vision to position India as a leader in AI technology and innovation.
India is strengthening its artificial intelligence (AI) infrastructure with significant investments in the India AI Compute Facility, acquiring 18,693 GPUs to power large-scale AI model development. Union IT Minister Ashwini Vaishnaw announced that Chinese AI platform DeepSeek will be hosted on servers in India to address concerns related to data privacy and cross-border data transfer, as reported by Moneycontrol.
Open-Source AI Model Hosting
Speaking at the Utkarsh Odisha Conclave on January 30, Vaishnaw emphasized that DeepSeek, like Meta’s LLaMA, is an open-source model that can be hosted within India. He highlighted that this step would mitigate privacy concerns by ensuring that AI models operate within Indian infrastructure.
He further stressed the importance of model optimization, advocating for AI initiatives to remain open and application-driven for broader technological progress.
India AI Compute Facility: Driving AI Growth
The India AI Compute Facility is leading the charge in AI research and development by procuring 18,693 GPUs, of which 15,000 are high-end, including 1,480 H200 units. Around 10,000 GPUs are currently active, providing computing resources to researchers, startups, and enterprises.
“This initiative is highly cost-effective, with expenses being significantly lower than one dollar, subsidized by the Government. We are ensuring support for four years, but true AI value will emerge through efficient algorithms and quality datasets,” Vaishnaw stated.
Indigenous GPU Development in Progress
To strengthen India’s technological capabilities, the Government is collaborating with global partners to advance GPU and chip development. Vaishnaw disclosed that leading chip designers are willing to co-develop GPUs with India.
“We are building a robust chip design ecosystem, and discussions are underway with MeitY for GPU co-development. More details will be shared soon,” he said.
Focus on Data Sovereignty and Security
India remains cautious about cross-border data movement, particularly in relation to China. Hosting AI platforms like DeepSeek on Indian servers aligns with the nation’s commitment to data security and localization.
Major players such as Jio Platforms, CtrlS Datacenters, Locuz Enterprise Solutions, and NxtGen Data Center are actively contributing to the country’s AI mission by supplying GPUs.
Concluding his remarks, Vaishnaw underscored AI’s transformative potential in India, highlighting the crucial role of high-quality datasets and algorithmic advancements in unlocking its full capabilities.
Check out more on Startup news and Startup Funding News. Stay connected with us across all our social media platforms: Facebook, Instagram, LinkedIn, X (formerly Twitter), and Join Our Community.
