Indian Minister of IT praised Deepseek’s progress on Thursday, and said that the country would hold a large -scale language model of China in a domestic server in a domestic server at an unusual opening of Indian Chinese technology.
“I’ve seen DeepSeek -it is a very powerful model of $ 5.5 million,” said Minister ASHWINI VAISHNAW on Thursday, for investing in AI, which is much less New Delhi than many other countries. In response to criticism that was criticized.
Since 2020, India has cited national security concerns and has banned more than 300 apps and services related to China, including Tactoku and Wechat. Approval to allow DeepSeek in India seems to be a condition for storing and processing all Indian users in accordance with the strict data logging requirements of India.
“Data privacy issues about DeepSeek can be dealt with by hosting an open source model on an Indian server,” VAISHNAW stated at an industry meeting.
Deepseek models may be hosted by Indian new AI calculation facilities. The facility is equipped with a 18,693 graphic processing unit (GPU), almost twice the first target. About 13,000 are NVIDIA H100 GPUs, and about 1,500 is NVIDIA H200 GPU. According to the Minister, we are ready to develop about 10,000 GPUs, and the facilities are planning to start operations in the coming days.
The facility also provides a sudden discount to Indian companies. VAISHNAW stated that standard AI computing was offered at a 42 % discount on market rates, and high -precision computing would be 47 % discounted.
The Minister’s remarks come the day after the DeepSeek’s app is excluded from Italian Apple and Google’s app store.
The release of DeepSeek’s R1 “Reasoning” model has been built based on a modest budget, sending shock waves to the high -tech industry this week, reducing the market capitalization of a huge NVIDIA of chips. The model was immediately undergoing a fierce scrutiny, causing a fierce debate about copyright issues, US exports, and even more money for AI’s efforts.
Beyond the host of foreign AI models, India is also trying to promote the development of AI models and related technologies on its own lawn. “The major chip designers are inevitable to cooperate with India to develop indigenous GPUs,” said Vaishnaw.
VAISHNAW estimates that India is looking at investment of $ 30 billion and data center in the next two to three years. Bloomberg plans to build a report last week, one of the largest Konglomarit in Japan, which could become the world’s largest data center in Jamnagar.
“We believe that at least six major developers can develop AI models in 6-8 months due to the outside restrictions can develop 4-6 months with more optimistic estimates. This is the most important component for creating a robust AI ecosystem, “said Vaishnaw.
Computing facilities also support India’s wider AI Initiative. VAISHNAW stated that 18 AI -led applications focused on agriculture, climate change, and learning disabilities have been selected for initial funds.
In order to supervise these AI Initiative development, India will use what we called “Hub and Spoke Model” to establish a regulatory organization, so that multiple organizations can cooperate with the safety framework. 。 “We will open the model and keep the focus on the application,” he said.
