AI is transforming the world at an unprecedented pace, yet the divide between those who can access cutting-edge AI tools and those who cannot remains wide. In order to solve India’s longstanding problems of computing accessibility and bridge this gap, the Indian Institute of Technology Madras (IIT Madras) and IITM Pravartak Technologies Foundation have partnered with Ziroh Labs, a California-based innovation-driven deep-tech startup, to establish a Centre of Artificial Intelligence Research (CoAIR). This collaborative initiative aims to make high-performance AI accessible even in environments where powerful computing infrastructure is unavailable or unaffordable.
The CoAIR is designed to focus on building useful and viable Artificial Intelligence solutions that emphasize CPU and edge device inferencing. Rather than depending on GPUs (Graphics Processing Units), which are expensive and often hard to obtain, the Centre will work on developing AI models that can run efficiently on widely available CPUs. These CPUs are commonly found in cloud data centers, smartphones, and edge devices, and do not require specialized hardware or sustained internet connectivity. This approach is especially relevant for India, where a significant portion of the population resides in areas with limited technological infrastructure.
As part of this transformative initiative, Ziroh Labs launched the first version of “Kompact AI” on April 9 at IIT Madras. This new platform is seen as a precursor to harnessing Artificial Intelligence models to solve accessibility challenges, particularly in rural and underprivileged regions. It is specifically built to enable the deployment and development of simple yet efficient Artificial Intelligence models onto CPUs, removing the need for complex GPU-based systems. The platform’s ability to perform without reliance on the internet or external communication channels makes it particularly powerful in off-grid or remote settings.
Since it employs various algorithmic and implementation-level optimizations, it can execute models on the low end of hardware availability without sacrificing too much performance. This efficiency, paired with a lack of dependency on external support like the internet, makes it revolutionary in its potential impact. IIT Madras plans to utilize this platform under the CoAIR to create compact AI models tailored to specific applications, focusing on sectors like healthcare, agriculture, education, and public welfare.
These models, once created and deployed, will not only enhance execution efficiency but also dramatically improve accessibility and applicability of high-performance AI in real-world conditions. By doing so, the initiative aims to democratize AI, enabling developers and innovators across the globe to build, train, and infer using CPUs—hardware that is far more common and affordable than GPUs. In doing so, it helps circumvent a major barrier that has prevented AI from reaching underserved communities.
In a statement issued by IIT Madras, it was highlighted that Kompact AI allows for the development of Artificial Intelligence solutions without violating any data privacy or data residency regulations. This means that data can remain on local systems, protecting user privacy while ensuring compliance with international and regional regulations—something that’s increasingly critical in sectors such as healthcare, defense, and finance.
During the launch event, the Ziroh Labs team demonstrated Kompact AI and showcased its unique capabilities. The demonstration proved that the platform could redefine AI development over CPUs—delivering performance that was previously thought possible only through GPUs. What makes Kompact AI stand apart is its ability to support AI model optimization, such as for DeepSeek, Qwen, and Llama, making them viable for affordable CPU-based environments.
AI has often been viewed as a technology reserved for the elite—how can platforms like Kompact AI ensure that it reaches every corner of society, especially the underserved and remote communities?
This is one of the core missions of the CoAIR and Kompact AI initiative. AI cannot remain confined to high-end laboratories and urban centers if it is to solve the world’s pressing challenges. Solutions like Kompact AI are critical for scaling AI applications across domains where cost, connectivity, and complexity are major barriers. Through such platforms, developers in small towns, students in rural schools, and healthcare workers in villages can all gain access to the power of Artificial Intelligence without needing to invest in expensive infrastructure.
The launch was a prestigious event attended by fellow students, IIT Madras staff, scientists, and an illustrious group of dignitaries. These included Prof. V. Kamakoti, Director of IIT Madras; Dr. Whitfield Diffie, Royal Society Fellow and Turing Award winner; Mr. Scott McNealy, Co-Founder of Sun Microsystems; Dr. William J. Raduchel, former CTO of AOL Time Warner and Senior Strategy Officer at Sun Microsystems; Prof. S. Sadagopan, Former Director of IIIT-Bangalore; and Mr. Raj Singh, Founder of APIGee and Fiberlane. Their presence emphasized the global significance of this initiative and the technological revolution it signifies.
Kompact AI’s key features include the ability to easily design and deploy AI models directly onto CPUs, eliminating the need for costly GPUs. It can optimize large language and inference models, while also functioning offline—making it an ideal solution for deployment in off-site or remote locations such as border regions, disaster-hit zones, isolated villages, and mobile healthcare units.
Addressing the audience during the launch, Prof. V. Kamakoti explained the philosophy behind the development of Kompact AI. He said, “Nature has taught us that one can effectively acquire knowledge and subsequently infer in only a limited set of domains. Efforts to get everything in the universe are not sustainable and will ultimately fail.” This statement reflects the CoAIR’s strategic focus on domain-specific AI development, rather than an overambitious pursuit of general-purpose intelligence.
Ultimately, the establishment of CoAIR and the launch of Kompact AI underscore India’s determination to lead the world in accessible, ethical, and inclusive AI innovation. By focusing on low-cost, privacy-respecting, offline-compatible, and CPU-based models, this collaboration is set to transform the AI landscape not only in India but across the world. As AI continues to evolve, platforms like Kompact AI will play a central role in ensuring that the technology remains equitable, relevant, and available to all.
ALSO READ