UAE Gifts Cerebras AI Superchip to India, Boosting Tech Partnership

In a significant move underscoring deepening technological collaboration, UAE President Sheikh Mohamed bin Zayed Al Nahyan presented a Cerebras AI superchip to Indian Prime Minister Narendra Modi. This handover, occurring during Modi’s visit to the UAE, marked the formalization of the ‘Condor Galaxy India’ initiative, an ambitious 8-exaflop AI supercomputing partnership between the UAE and India.

The partnership involves the deployment of 64 Cerebras CS-3 systems to establish one of the largest AI compute clusters in India. This initiative is poised to bolster India’s domestic AI ambitions, enhance its sovereign AI compute capabilities, and reduce reliance on foreign cloud infrastructure. Startups and researchers in India will gain access to frontier-scale computing power, fostering innovation and development in the AI sector.

At the core of this collaboration are Cerebras’s novel wafer-scale AI chips. Unlike traditional GPUs that use multiple smaller chips, Cerebras utilizes an entire silicon wafer to build a single, massive processor. This Wafer-Scale Engine (WSE) architecture integrates billions of transistors and numerous AI-optimized cores on one piece of silicon, dramatically reducing data communication latency and enabling significantly faster AI training and inference.

The ‘Condor Galaxy India’ project was initially announced by Prime Minister Modi and President Al Nahyan in January of the same year. A subsequent term sheet was signed between Abu Dhabi-based tech firm G42 and India’s Centre for Development of Advanced Computing (C-DAC) to deploy the 8-exaflop supercomputer cluster. G42, in collaboration with C-DAC, will manage the installation, deployment, operations, and maintenance of the system.

Once operational, this supercomputer cluster will represent a substantial leap for India’s AI ecosystem. Its projected 8,000 petaflop capacity is nearly 19 times the combined compute capacity of India’s current flagship AI supercomputers, AIRAWAT and PARAM Siddhi-AI. This enhanced capability will facilitate faster training and execution of large AI models, provide greater control over data, and offer affordable access to cutting-edge computing resources for a wide range of entities.

The initiative is expected to drive joint research and development across critical sectors such as health and genomics, energy, and geospatial analytics. Furthermore, by hosting the infrastructure and data within India, the project addresses crucial aspects of data sovereignty, security, and national control, aligning with India’s strategic goals for its burgeoning AI landscape.