How IIT Madras Just Cracked AI’s Biggest Barrier with Kompact AI, No GPUs Needed

How IIT Madras Just Cracked AI’s Biggest Barrier with Kompact AI, No GPUs Needed
IIT Madras and Ziroh Labs establish Centre of AI Research, unveil Kompact AI

In a major leap for India’s artificial intelligence (AI) ecosystem, IIT Madras has partnered with California-based deep tech startup Ziroh Labs and the IIT-M Pravartak Technologies Foundation to establish the Centre of AI Research (CoAIR). The highlight of this collaboration is the unveiling of Kompact AI, an AI platform that enables foundational models to run on everyday CPUs, removing the need for expensive GPUs.

This development could change how AI is accessed and used across the country, especially by startups and small businesses that often lack access to costly GPU infrastructure.

Kompact AI: Making Advanced AI Work Without Expensive Chips

Kompact AI is a new platform built to run large AI models on everyday CPU-based machines. Developed by Ziroh Labs with support from IIT Madras, the system can handle complex models like Meta’s Llama 2, Alibaba’s Qwen2.5, and DeepSeek—without the need for costly GPUs. Instead, it works efficiently on standard processors like Intel Xeon, already used in many office laptops and desktops.

Kompact AI aligns with India's "AI for All" mission, aiming to make AI more accessible, cost-effective and scalable.

In a recent live demo, Ziroh Labs showed Kompact AI in action on a regular laptop fitted with an Intel Xeon processor. The platform successfully ran queries on high-performing AI models, proving that powerful AI is now possible without specialised hardware.

To date, 17 AI models have been optimised to run through Kompact AI. These include some of the most demanding language models available today. Ziroh Labs and IIT Madras jointly carried out performance and accuracy tests to ensure reliability. It has also been tested on popular chips from both Intel and AMD, confirming that it can operate across a wide range of systems already in use.

Traditionally, models of this size required powerful GPUs, such as those offered by Nvidia, which often cost anywhere between INR 10 to INR 20 lakh. These chips are not only expensive but are also hard to source as they remain frequently out of stock due to global demand. For most startups in India, this made it difficult to work with cutting-edge AI.

Kompact AI removes this barrier. It is designed to run offline and doesn’t depend on cloud infrastructure, making it a practical solution for regions with limited or unstable internet. For a country like India, with thousands of startups and many operating in Tier II and III cities, this could be a game-changer.

By cutting hardware costs and simplifying AI access, Kompact AI may open the door for more innovation and real-world adoption of AI across sectors like education, healthcare, logistics, and governance. It shows that high-performance AI can be both accessible and affordable.

A Step Towards Accessible and Inclusive AI

The Centre of AI Research, jointly established by IIT Madras and Ziroh Labs, aims to make India a global hub for accessible AI. The goal is not just to develop cutting-edge technology but also to make sure that it reaches a wider base, including startups, students, and enterprises that cannot afford costly infrastructure.

India has over 1.5 lakh startups, and a large number of them are building tech-based solutions but struggling to keep up due to infrastructure costs. With this new development, the playing field is starting to level out. It can surely support more AI-based innovation, enable research and development at universities, and create new opportunities for businesses in Tier 2 and Tier 3 cities.

This initiative also strengthens India's position in the global AI race. Instead of being dependent on costly imports and cloud subscriptions, Indian companies can build and deploy AI models locally, reducing costs and improving data privacy.

A Big Step for India’s Tech Future

This launch is indeed a big-time win for India's technology and startup sectors. In a world where access to hardware limits who gets to innovate, Kompact AI removes the barrier.

It shows a shift in how India is deciding to solve its own AI challenges, not by running after what is trendy but by focusing on cost-effective, scalable solutions that can work well in the Indian conditions.

While global companies dominate the high-end AI race, platforms like Kompact AI open the doors for grassroots AI innovation. With IIT Madras and Ziroh Labs leading the way, this could mark the beginning of a strong new phase in India’s growth story.

The big question now is: Will this spark a wider change in how India builds and uses AI? If anything, this signals that AI in India now seems to be within reach for many.


AI Firms Say Analysing Use Cases and Robust Data Strategies Key for AI
Explore the AI landscape in India: From emerging startups to challenges companies face. Uncover the trends, myths, and the crucial need for AI training.

WIDGET: questionnaire | CAMPAIGN: Simple Questionnaire 

Must have tools for startups - Recommended by StartupTalky

Read more