Diplomatic gifts are usually symbolic. A ceremonial sword. A piece of local art. A handcrafted item that ends up in a display cabinet somewhere in South Block, admired briefly and then forgotten.

What Sheikh Mohamed bin Zayed Al Nahyan handed Prime Minister Narendra Modi during their bilateral meeting in Abu Dhabi on May 15 was none of those things. It was a Cerebras CS-3 — one of the most powerful AI processors ever built — and it was accompanied by something considerably more consequential than a handshake. It was the formal activation of a partnership that will give India one of the largest AI computing clusters on the planet.

WHAT WAS ACTUALLY ANNOUNCED

The physical chip handed to PM Modi marked the formal execution of Condor Galaxy India — an ambitious 8-exaflop AI supercomputing partnership between India and Abu Dhabi-based tech major G42. Under this agreement, 64 Cerebras CS-3 systems will be deployed to build one of the largest AI compute clusters in India.

To understand what 8 exaflops actually means: the proposed capacity is almost 19 times the combined peak computing power of India's two flagship AI supercomputers — AIRAWAT and PARAM Siddhi-AI — both currently hosted at C-DAC Pune. This is not an incremental upgrade. It is a generational leap. The project was first announced during Al Nahyan's visit to India in January 2026, followed by a term sheet signed between G42 and India's Centre for Development of Advanced Computing at the India AI Impact Summit 2026. Under the framework, G42 will be responsible for installation, deployment, operations, and maintenance of the system, in partnership with C-DAC.

THE CHIP ITSELF: WHY CEREBRAS IS DIFFERENT

The Cerebras CS-3 is not a conventional processor — it is an engineering philosophy made silicon.

Traditional GPUs are manufactured by printing hundreds of small chips onto a large circular silicon wafer, which is then cut into individual chips for commercial use. Cerebras Systems builds a single massive processor using an entire silicon wafer — an architecture it calls the Wafer-Scale Engine. This chip houses over 4 trillion transistors and close to 1 million AI-optimised cores on a single piece of silicon.

By keeping the entire processing network on one continuous piece of silicon, Cerebras reduces data communication latency — delivering up to 20 times faster AI training and inference capabilities compared to conventional GPU clusters.

For decades, wafer-scale integration was considered technically unviable — a single speck of dust during manufacturing could destroy the entire chip. Cerebras solved this by building thousands of spare cores directly onto the silicon wafer, allowing the chip to automatically reroute data around any manufacturing flaws.

Founded in 2015 and publicly listed on the Nasdaq just days before this diplomatic exchange — raising $5.55 billion and briefly reaching a market capitalisation of $95 billion — Cerebras Systems has gone from selling individual hardware boxes to building giant cloud supercomputers.

WHAT THIS MEANS FOR INDIA

The Condor Galaxy India cluster is not just infrastructure. It is a statement about where India wants to be in the global AI hierarchy — and how quickly it intends to get there.

For India, the partnership brings sovereign AI compute onshore, reduces dependence on foreign cloud infrastructure, and gives startups and researchers access to frontier-scale computing power. The supercomputer facility will help train and run large AI models much faster than conventional infrastructure and offer greater control over data and latency.

The cluster will also offer affordable access to startups as well as public and private sector entities for research, application development, and commercial use — with planned applications spanning drug discovery, satellite data processing for disaster management, smart energy grid simulation, and scientific research across health, genomics, and geospatial analytics.

The data sovereignty dimension is particularly significant. Every AI query processed on domestic infrastructure is a query that stays within Indian law, Indian regulation, and Indian control. G42 CEO Mansoor Al Mansoori described the project as "delivering infrastructure that converts energy and compute into sovereign governed nation-scale intelligence."

THE BIGGER PICTURE

India's AI ambitions have long been constrained by one uncomfortable bottleneck: compute. The country produces exceptional AI talent — some of the finest researchers and engineers working on frontier models globally are Indian. What has consistently lagged is the hardware infrastructure to match that talent.

The Condor Galaxy India cluster begins to close that gap in a material way. Sixty-four CS-3 systems. Eight exaflops of computing power. Nineteen times the capacity India had last week.

The UAE President gave PM Modi a chip. What he actually gave India was a fighting chance at the AI era on its own terms.