DecodePc
DecodePc

India is ready to introduce its own reasonably priced safe and secure indigenous artificial intelligence model. This was stated in New Delhi at Electronics Niketan by Shri Ashwini Vaishnaw, Union Minister for Electronics & Information Technology, Railways, Information & Broadcasting.

The Union Minister told the media that India’s AI model is an appropriate move because India is a trusted country. This will help India become a more dependable leader in developing ethical AI technology in the future.

Supported by an advanced shared computing infrastructure, the India AI mission is now progressing towards the customization of indigenous artificial intelligence solutions tailored to the Indian context, utilizing Indian languages. He stated that scientists, researchers, developers, and coders are engaged in the development of various foundational models in this context. Given the current rate of progress, the Union Minister expressed optimism that the Indian AI model is expected to be completed within six months.

The Union Minister, Shri Ashwini Vaishnaw, stated that “The economic philosophy of our Prime Minister is characterized by a strong emphasis on inclusivity.” He advocates for the accessibility of modern technology for all individuals, with the aim of economically empowering those at the base of the socioeconomic pyramid.

The artificial intelligence model is commencing its operations with a computational capacity of approximately 10,000 GPUs. The remaining 8,693 GPUs will be incorporated shortly. Initially, it will primarily benefit researchers, pupils, and developers. The technical collaborators involved in the mission have conveyed significant confidence in the initiative’s capacity to achieve its objective of democratizing access to computing, and to do so at a highly competitive rate. In contrast to global models, which incur computational costs ranging from $2.50 to $3.00 per hour, the computational expenses associated with India’s AI model will amount to less than 100 rupees per hour, following the application of a 40% government subsidy. The appealing semi-annual and annual programs will enhance affordability.

Within a period of ten months following the start of the India AI Mission, the Ministry of Electronics and Information Technology has garnered an exceptional response and established a sophisticated and resilient common computing facility comprising approximately 18,693 Graphics Processing Units (GPUs). This amount is approximately nine times greater than that of the Open Source Model DeepSeek and roughly two-thirds of the capacity possessed by ChatGPT. In response to media inquiries, the Union Minister stated that DeepSeek may be hosted on Indian servers following comprehensive security assessments, thereby enabling programmers, developers, and designers to benefit from its open-source code.

The government prioritizes the safe and ethical deployment of artificial intelligence models. In articulating this commitment, the Union Minister declared that India is in the process of establishing an Artificial Intelligence Safety Institute, which will employ a techno-legal framework.

Following eight concurrent initiatives to guarantee data privacy coupled with ethical audits of algorithmic efficiency constitute the key safety-related projects in this sense.

  • Machine Unlearning at IIT Jodhpur
  • Synthetic Data Generation from IIT Roorkee
  • NIT Raipur’s AI Bias Mitigating Strategy
  • Framework for Explainable AI: Minecraft Technologies, Pune’s Defence Institute of Advanced Technology
  • Privacy Enhancing Techniques: IIT Delhi, IIIT Delhi, IIT Dharwad & Telecom Engineering Centre, TEC
  • AI Ethical Certification Framework (created at IIIT Delhi & TEC from Tool Nishpaksh)
  • Civic Data Labs’ developing tool parakh, AI algorithm auditing framework
  • The Amrita Vidyapeetham & Telecom Engineering Centre AI Governance Testing Framework

The solid basis for Democratic AI development is Common Compute Facility.

Rapid development of a massive computing infrastructure under the India AI Mission exceeds world standards. Now housing 18,693 GPUs—including 12,896 H100, 1,480 H200, and 7,200 MI 200300 units—the facility greatly surpasses the intended aim of 10,000 GPUs. DeepSeek was trained on 2,000 GPUs, but ChatGPT needed 25,000 GPUs, thus putting this capacity into perspective is somewhat different. Along with accelerating research, model training, ethical AI algorithm development, and Indian AI ecosystem creativity promotion, this enormous computer capacity will enable

A shared computing tool has been set up, making it accessible to everyone involved. There are now about 10,000 GPUs on hand, and technical partners have shown faith in the capacity of the mission to provide world-class artificial intelligence solutions. Approved, this facility will shortly be operational for general usage.

Empanelment of cloud-based AI services

Approved by the Indian government with a Rs 10372 Cr budgetary investment in March 2024, the India AI Mission The aim was to close the gaps in the current AI ecosystem and let India to become a centre of AI technological and application development. Under seven main pillars, India AI Mission is being carried out with one of the main pillar facilitating the availability of 10,000 GPUs for AI Compute Infrastructure.

The Ministry of Electronics and Information Technology (MeitY), via IndiaAI Independent Business Division (IBD), released a Request for Empanelment (RFE) on August 16, 2024, empanel AI services on cloud and offer the services to academia, MSMEs, startups, research community, governments, public sector agencies and other entities approved by IndiaAI via the CPP portal.

Proposals were put in by 19 bidders comprising Managed Service Providers (MSPs), Cloud Service Providers (CSPs), and Data Center Service Provider. Following preliminary investigation per pre qualifying criterion 13 bidders were invited to present to the technical evaluation committee. Ten bidders turned out to be qualified for opening of financial bids according to the technical assessment. Opening on January 22, 2025, the financial bids of the ten technically qualified bidders were The ten bidders below have provided their commercials for several types of artificial intelligence compute units (GPUs) specified.

  • CMS Computers India Pvt Ltd
  • Ctrls Datacenters Ltd
  • E2E Networks Limited
  • Jio Platforms Limited
  • Locuz Enterprise Solutions Limited
  • NxtGen Datacenter and Cloud Technologies Private Limited
  • Orient Technologies Limited
  • Tata Communications Limited
  • Vensysco Technologies Limited
  • Yotta Data Services Private Limited

Intel Gaudi 2, AMD MI300X, MI325 X, NVIDIA H100 (PCIe, NVL and SXM), NVIDIA H200 (PCIe, NVL and SXM), NVIDIA A100, NVIDIA L40S, NVIDIA L4, AWS Inferentia2 and Aws Tranium have been stated by these bidders as varied category of artificial intelligence compute units (GPUs). Additionally provided by bidders are eight GPU modules of AMD and NVIDIA.

 

Under empanelment, the bidder quoted the L1 rates for a certain category of compute unit will be affiliated first and all other bidders in that category of the compute unit will be eligible for empanelment should they agree to match the L1 bid.