Cerebras Systems and G42 Collaborate on Condor Galaxy AI Supercomputer
Cerebras Systems, a Silicon Valley startup, and G42, a UAE-based technology holding group, have successfully completed the first phase of the AI Supercomputer Condor Galaxy network.
- Key Achievements in Phase 1:
- Nine interconnected supercomputers
- Total capacity of four exaFLOPs and 54 million cores
This accomplishment marks the initiation of the next phase, Condor Galaxy 2 (CG-2), with plans to expand the network:
- Condor Galaxy 2 Goals:
- Eight exaFLOPs
- 108 million cores
Kiril Evtimov, Group CTO at G42, expressed excitement about the partnership's vision of creating the world's largest and fastest AI supercomputers. The collaboration aims to support organizations in various sectors, such as healthcare, energy, investments, and climate studies, by providing extensive AI training capacity.
Andrew Feldman, CEO and co-founder of Cerebras Systems, highlighted the global impact of their combined expertise in advancing AI work. The collaboration has already led to the launch of models like Jais 13B, 30B bilingual models, and the Med42 Clinical LLM.
- Successful Outcomes:
- Launch of innovative models
- Accelerated AI solution development
Dr. Andrew Jackson, Chief AI Officer at Core42 (a G42 company), emphasized the platform's scale and ease of use, demonstrating accelerated AI solution development.
- Condor Galaxy Features:
- Optimized for training large AI models
- Addresses compute, dataset, and specialized AI expertise requirements
The long-term plan involves deploying nine interconnected supercomputers, providing a total compute power of 36 exaFLOPs. Condor Galaxy 2, optimized for Large Language Models (LLMs) and Generative AI, aims to deliver four exaFLOPs of 16-bit AI compute, supporting models with up to 600 billion parameters.