MeitY opens continuous empanelment for GPU procurement under IndiaAI Mission

This follows the August 2024 GPU tender where the government sought applications for procuring 10,000 or more GPUs

The Ministry of Electronics and Information Technology (MeitY) has launched a continuous empanelment process for AI cloud service providers, allowing firms to apply on an ongoing basis to supply GPUs and related AI compute services under the IndiaAI Mission.

This move, announced in Request for Empanelment (RFE) document on February 21, aims to significantly boost India's AI computing capacity, building on the initial procurement of nearly 19,000 units finalised in January.

Unlike previous tenders, which were based on a one-time bidding process, the new system allows vendors to join anytime, provided they meet the lowest rates established in earlier rounds.

"IndiaAI would renew the empanelment every quarter inviting fresh proposals from the empaneled agencies for discovering any revised rates. Empaneled agencies shall submit a revised financial proposal which can be same or lower than the existing L1 rates," the RFE said.

The empanelment is open to data centre providers, cloud service providers (CSPs), managed service providers (MSPs), and system integrators that meet the eligibility criteria.

Google will likely provide its Tensor chip in this round of the empanelment process, Moneycontrol previously reported. While the chip was considered in the earlier round, it did not receive necessary internal and legal approvals.

The empaneled providers will need to supply AI compute instances, with at least 70 percent of GPUs offering FP64 (double precision) capabilities. Floating point 64 or FP64 is a format that uses 64 bits to represent numerical values, and which allows higher precision and accuracy in computations.

Like the 2024 August tender, the government has set a minimum requirement of 1,000 AI compute units per provider.

Also read: IndiaAI Mission: MeitY invites applications from industry for procuring 10,000 GPUs

Applicants will be evaluated based on technical qualifications, financial strength, installed GPU capacity, and the diversity of AI compute models they offer.

Additionally, the RFE states that the bidders must provide a self-service portal for users to access GPUs, track utilisation, and apply for subsidised AI compute services.

In the IndiaAI Mission, the government will subsidise GPU access through a coupon-based system up to 40 percent.  This model was previously announced alongside the January procurement but is now being integrated into the empanelment framework.

This development comes at a time when the government, through the IndiaAI Mission, has started working on developing indigenous AI models, including large multimodal models (LMMs), domain-specific AI applications, and foundational model.

The empanelment process is expected to intensify competition among Indian cloud providers, as new firms can challenge existing L1 bidders by offering lower rates.

Also read: IndiaAI GPU procurement sees record discounts as firms slash prices by up to 89%

Existing L1 bidders include Jio Platforms, NxtGen Data Center, Locuz Enterprises, E2E Networks, CtrlS DataCenters, CMS Computers, Orient Technologies, Tata Communications, Vensysco and Yotta Data Services.

The last date for the first round of applications under this continuous empanelment process is April 30, 2025, with evaluations set to begin in May 2025. However, new firms will be able to apply in subsequent rounds.