ARTICLE AD BOX
New Delhi: World leaders usually exchange diplomatic symbols when they give gifts to one another. A painting, a cultural artifact, maybe something homemade. The gift Prime Minister Narendra Modi received from UAE President Mohamed bin Zayed Al Nahyan, however, was quite different: an AI chip during his recent visit to the UAE.
The action could seem like a showmanship performance. The reality is that it represented something far more significant – the official launch of an ambitious AI infrastructure partnership between India and the UAE which might define the future course of India’s AI ecosystem.
The chip, which was sent to Modi, was created by Cerebras Systems, a privately held US semiconductor startup that’s one of the more watchful firms in the AI hardware fray. The chip will be used in an AI supercomputing project called “Condor Galaxy India”, which is being developed in collaboration with Abu Dhabi-based G42.
The moment was, in some ways, a mirror of the point at which the energy industry or telecom has become part of geopolitical strategy.
A transition from Software to Compute.
In the past two years, the discussion on AI has mainly focused on Chatbots, Generative AI, and large language models. However, there is another less obvious and less popular one underneath: compute infrastructure.
AI models need to be trained using tremendous computational resources, dedicated hardware, and extensive data infrastructures. In countries where such compute capacity is not available, reliance is often on cloud services outside their borders and on external infrastructure. Although India is one of the world’s biggest software ecosystems, the country still imports a significant amount of the high-end AI computing capacity.
This is one of the reasons why the agreement between India and the UAE is important.
As part of the partnership, 64 Cerebras CS-3 systems will be installed in India to form one of the world’s biggest AI compute clusters. The system, once operational, is projected to deliver sovereign artificial intelligence compute infrastructure in India, enabling startups, researchers, and enterprises to access cutting-edge computing capabilities without relying 100% on cloud ecosystems outside the country.
The scale is not the only impressive thing; it is the timing. The world is now in a race to secure access to the technology before AI becomes even more centralized among a few companies and countries.
What Makes Cerebras Different
Unlike traditional chip makers, which manufacture several small chips on silicon wafers, Cerebras Systems took a different approach.
Its flagship Wafer-Scale Engine (WSE) is a single processor carved from a silicon wafer. This new WSE-3 chip packs over 4 trillion transistors and almost one million AI-optimised cores into a single silicon chip.
The ability to produce wafer-scale computing systems has long been deemed commercially infeasible because the production process can introduce manufacturing defects in chips large enough to render the entire wafer useless. To work around this, Cerebras added additional cores to the architecture, which the system is supposed to bypass if they are faulty.
The result is a processor especially optimized for AI workloads, especially large model training and inference. According to the company, the architecture can offer a dramatic increase in performance over traditional GPU clusters, with much of the data movement within the chip eliminated.
Cerebras is also playing on turf dominated by giants such as NVIDIA. That’s a noteworthy increase. It was created in 2015 by Andrew Feldman, Gary Lauterbach, Michael James, Sean Lie and Jean-Philippe Fricker, who were primarily interested in developing solutions to semiconductor engineering challenges that many in the industry had long left behind.
It has now become the hub of a state-level AI partnership between two governments.
Why is India interested in AI?
AI’s agenda has gained momentum in India over the last year, particularly in the realm of sovereign capabilities and the establishment of domestic infrastructure. Policymakers have come to view AI compute as a strategic national resource and not just an asset to the private sector.
The proposed 8 exaflop system is likely to be much more powerful than India’s current top-notch AI supercomputers such as AIRAWAT and PARAM Siddhi-AI at C-DAC Pune.
Access is not only an issue of scale, it’s a practical problem, too.
One challenge for AI startups is the cost of training large models on overseas cloud services. Changing to a domestic compute infrastructure could save money, lower latency, and provide more control over sensitive datasets.
The scenario applications are not limited to chatbots. Such facilities are expected to be used in fields such as: Drug discovery, Genomics, Disaster management, Satellite imaging, Climate modelling and smart energy systems.
Another key consideration is data sovereignty. The partnership could mitigate regulatory and security risks associated with cross-border data transfer by providing compute and data infrastructure in India.
A more comprehensive India-UAE Technology Partnership.
The collaboration on AI also reflects the changing dynamics of India-UAE relations, moving beyond trade and energy cooperation.
The two countries have been steadily enhancing cooperation on digital infrastructure, clean energy, fintech, and emerging technologies over the last few years. Companies such as G42 in the UAE have been making their mark as significant players in AI and technology investments worldwide.
Such collaborations also benefit India by providing access to infrastructure and capital, especially as AI development becomes increasingly compute-intensive.
Projects such as Condor Galaxy India are still far from achieving their full potential. Timings for deployment, commercial availability, and longevity are still a question mark.
However, the Cerberus chip’s symbolism in hands of Modi was hard to miss. The gift was more than technology; in a world where compute power is becoming as strategically important as nature, it was golden. It underscored the growing significance of AI infrastructure in diplomacy, economic strategy, and national competitiveness.
English (US) ·