Over 17K GPUs Installed Under IndiaAI Mission: Report

India has reportedly completed installations of over 17,300 graphic procession units (GPUs) under the IndiaAI Mission as the country advances towards its adoption of artificial intelligence.

An ET report said that this was revealed in a last week meeting by IndiaAI CEO Abhishek Singh with the cloud service providers (CSPs). The meeting was held to review the progress of GPU installations and their integration with the IndiaAI compute portal. 

The report further said that two of the ten selected CSPs namely Jio Platforms and CtrlS Datacenters are yet to deploy their GPUs while others including Yotta, NextGen Cloud, and E2E Networks have made progress in the installation of GPUs. 

In a statement to Inc42, E2E Networks said it has deployed and made available all the committed GPUs. The homegrown cloud computing company had committed to install about 1,256 GPUs which includes, 700 H100s, 256 H200s and 100 A100s, L40s and L4s each.

“We’re proud to be the only provider that made 100% of empanelled GPUs available on day one for the IndiaAI mission. While others have taken POs approach, we committed upfront—driven by our belief in the IndiaAI mission and the future of AI. We ensured immediate availability of GPUs and zero delays,” said a spokesperson from E2E Networks.

Cloud computing services provider Cyfuture has also been added to the empaneled list of CSPs lately. The company has been provisioned to provide 1184 GPUs and has commissioned 80 GPUs out of them so far. The company said that the Centre has given six months to Cyfuture to fulfill their commitments. “We’re confident of procuring and commissioning this within 3-4 months time,” said Anuj Bairathi, CEO at Cyfuture. 

Other CSPs part of the empanelled list by the Centre includes Locuz, Orient, Tata Communications, Ishan, Sify and Netmagic. 

Inc42 had also sent queries to Yotta, NextGen Cloud, Jio Platforms and CtrlS Datacenters. The story will be updated based on their responses. 

The report also said Jio Platform was in the process of procuring 752 Nvidia H200 GPUs and 268 AMD MI300X GPUs, while the company’s API integration process with the computing portal was in progress and is scheduled to be completed by the end of the third quarter of 2025.

CntrlS has not confirmed the installation of GPUs but shared purchase orders. The cloud computing platform claims to have completed the integration of the computing portal. 

Yotta Data Services has completed the installation of 50% of its committed GPUs. The company was provisioned 4,096 GPUs by the Centre and it has concluded the process of installing and commissioning 1,504 Nvidia H100 GPUs to Sarvam along with an additional 108 as a redundant buffer, and 200 Nvidia H100 GPUs to Bhashini. 

The remaining 2,592 Nvidia H100 GPUs are scheduled to be delivered to SarvamAI by July 10. These GPUs are under testing by the Yotta and Nvidia team at its data centre in Navi Mumbai. Yotta’s API integration has been completed.

NextGen Cloud has installed 51% of its lot of GPUs and the remaining will be installed and commissioned by the cloud computing company by June 30.

While Ishan Technologies, another CSP is provisioned to provide 1,000 GPUs under the IndiaAI mission.

IndiaAI mission has an outlay of more than INR 10,000 Cr has approved the proposal to import 34,333 GPUs across its first two rounds. It aims to build a cloud computing platform for startups and researchers to help them train their artificial intelligence models. This would further help the country to build its own large language model (LLM). 

India’s AI Dream

Under the scheme, the Centre has shortlisted Soket AI, SarvamAI and Gnani.ai to build India’s own LLM. While Sarvam and Soket are building LLMs focussed on text, Gnani.ai is working on its potential Voice AI foundational model. Besides, it is also crafting a 70 Bn parameter model with text-to-speech capabilities. 

In an earlier interaction with Inc42, founders of these AI startups shared that availability of GPUs is still a major bottleneck. While GPUs are an essential part of developing LLMs, better infrastructure, including efficient data pipelines, low-level GPU access, and flexible training configurations is also another challenge faced by the above AI startups.