G42, Cerebras Unveils Brand New AI Supercomputer

Silicon Valley startup Cerebras Systems and G42, UAE-based technology holding group, have completed the first phase of the AI Supercomputer Condor Galaxy network, featuring nine interconnected supercomputers with a total capacity of four exaFLOPs and 54 million cores. 

This achievement initiates the next phase, Condor Galaxy 2 (CG-2), aiming to expand the network to eight exaFLOPs and 108 million cores. 

“G42’s strategic partnership with Cerebras Systems is paving the way toward our joint vision of creating the world’s largest and fastest AI supercomputers,” said Kiril Evtimov, Group CTO, G42. 

“We support organizations working with unique, massive, and diverse datasets that require extensive AI training capacity. These use cases span across different sectors including healthcare, energy, investments, and climate studies to name a few.” he added. 

“In partnership with G42, we are changing the worldwide inventory of compute and using our combined expertise to advance AI work in a powerful way, to quickly and efficiently train leading LLMs.”said Andrew Feldman, CEO and co-founder, Cerebras Systems.

The successful collaboration has resulted in the launch of models like Jais 13B and 30B bilingual models and the Med42 Clinical LLM. Dr. Andrew Jackson, EVP, chief AI officer at Core42, a G42 company, emphasised the platform's scale and ease of use, demonstrating accelerated AI solution development.

The Condor Galaxy, optimized for training large AI models, addresses significant compute, dataset, and specialised AI expertise requirements. Cerebras and G42's plan involves deploying nine interconnected supercomputers with a cumulative compute power of 36 exaFLOPs.

Condor Galaxy 2, optimised for LLMs and Generative AI, will deliver four exaFLOPs of 16-bit AI compute, supporting models with up to 600 billion parameters.