Decoding IndiaAI: Inside India's Rs 10,372-crore plan to boost AI infrastructure

On March 7, the Union Cabinet approved the IndiaAI mission

The Union Cabinet on March 7 approved the ambitious Rs 10,372-crore  IndiaAI initiative whose ambit encompasses creating computing infrastructure for the country, AI centres for multi-modal LLMs, 10,000 graphics processing units and more.

The approval of this initiative can be traced back to early 2023, when the government created seven committees of experts to come out with a comprehensive roadmap for policy push on AI.

Based on the deliberations of the working group, the government came out with a 180-page report that echoes most of the initiatives that have been approved under the India AI mission.

Let's take a look at the most important initiatives in detail, going by the details that are present in the report of the IndiaAI working group.

 Support for computing and AI chips

AI chips, distinguished by their unique features, offer a substantial speed and efficiency advantage—tens or even thousands of times faster than central processing units (CPUs)—for both training and inference of AI algorithms. State-of-the-art AI chips not only outperform state-of-the-art CPUs in terms of efficiency but also prove to be significantly more cost-effective. These chips fall into three classes: graphics processing units (GPUs), field-programmable gate arrays (FPGAs), and application-specific integrated circuits (ASICs).

In light of these benefits, the IndiaAI mission proposes government support for the development of AI chips, through a collaborative approach through a public-private partnership (PPP) model. This was also recommended by the IndiaAI working group. The expert group had recommended that the Indian government create a computing infrastructure of 24,500 graphics processing units (GPU) at 17 centres to enable innovation in the emerging tech area in the country by startups and academia. Under the proposal, about 14,500 GPUs have been recommended for training AI models and high-performance storage, while the rest 10,000 are for AI inferencing. A GPU centre is a group of computers with GPUs on each node to train neural networks for image and video processing.

Marketplace for AI as a service

As it was recommended in the report of the AI working group, to enhance AI accessibility, AI Marketplaces will be leveraged to build service models for AI as a Service (AIaaS), Infrastructure as a Service (IaaS), and Platform as a Service (PaaS), offering pre-trained models to users. Open-source AI frameworks and libraries, supported by diverse vendors and communities, will be utilised to promote vendor-agnostic AI development, fostering innovation and collaboration.

Funding for startups

The startup financial vertical of the IndiaAI mission envisages funding support for deep-tech AI startups. This follows an announcement by Finance Minister Nirmala Sitharaman on the same line, as part of the Interim Budget 2024 last month.

The report of the IndiaAI workings group provides more detail into how this funding may look like. The working group has proposed an AI Ignition Grant Program to support 500 AI/deep tech startups at the product development stage with a grant of Rs 30 lakh each.

Non-personal data platform

Under IndiaAI, the government aims to streamline access to non-personal datasets to entrepreneurs and startups. "A unified data platform will be developed to provide a onestop solution for seamless access to non-personal datasets to Indian startups and researchers," a release said. In its report, the working committee had recommended that that the India Dataset Platform be monetised. “Data providers can set pricing models, such as one-time purchases, subscriptions, or licensing fees, based on the value and uniqueness of their datasets,” the report read. It is important to note that the idea of monetising the sharing of non-personal data was dropped by the Indian government after it received criticism over it. The government last year issued a draft policy on the India Datasets Platform which had withdrawn the monetising plans.

IndiaAI FutureSkills

This program which was launched last month by the Ministry of Electronics and Information Technology, also falls under the ambit of the IndiaAI mission. This too finds its source in the report of AI working group report. Under the program, the government will promote entry into AI programs and AI courses in undergraduate, masters-level, and Ph.D. programs.  The FutureSkills program will include a model curriculum for AI and repository, a framework that categorises courses and programs as per focus areas (for instance technology specific like algorithms, LLMs, and infrastructure specific such as GPUs, accelerators etc). The program also envisages training for faculty in AI.

"Recommendations like making it mandatory for educational institutes to re-skill their teaching staff with the new trends & research in the field of AI at least once in every two years. Periodic incentives for training of faculty can also be considered," the AI working group report had recommended.

Centre for application development and indigenous models

As part of the mission, the government is also proposing two pivotal verticals: the IndiaAI Innovation Centre and the IndiaAI Application Development Initiative. The IndiaAI Innovation Centre is envisioned to play a crucial role in the development and deployment of indigenous Large Multimodal Models (LMMs) and domain-specific foundational models within critical sectors. Simultaneously, the IndiaAI Application Development Initiative is geared towards promoting AI applications in critical sectors by addressing problem statements sourced from Central Ministries, State Departments, and other institutions.