The initial phase of the Condor Galaxy 1 AI Supercomputer has been successfully concluded in collaboration with G42 from the UAE, positioning it as the sole AI Hardware startup to surpass the $100 million revenue mark. Users of Cerebras are actively sharing their Supercomputing 23 CS- 2 outcomes, showcasing the exceptional processing speed enabled by wafer-scale technology. However, this enterprise currently faces challenges.
Cerebras recently announced a groundbreaking partnership with G42 (Group42), a prominent artificial intelligence and cloud computing firm, marking a significant milestone in the AI industry. The collaboration, unveiled just four months ago, will result in the creation of one of the most extensive artificial supercomputers globally, featuring 256 CS- 2 silicon level nodes delivering an impressive 36 Exaflops of artificial performance.
Following the successful deployment of an initial data center, Cerebras has swiftly moved on to the next phase of development. Both companies are capitalizing on the lucrative $70 billion market (projected for 2028) for Large Language Model services, especially amidst the ongoing scarcity of NVIDIA H100 units. This scarcity presents a unique opportunity for Cerebras to thrive. Additionally, Cerebras recently introduced the Jais30B, the most extensive Arabic Language Model, in collaboration with Core42 utilizing the CS-2 platform. This innovative solution aims to streamline the creation of large AI models by eliminating the need for decomposition and distribution.
Cerebras, in conjunction with Argonne National Labs, disclosed a remarkable 130-fold increase in throughput over the NVIDIAA100, alongside notable advancements with the Condor Galaxy project. The CS-2 wafer-scale engine powers applications like the Monte Carlo atom modeling, demonstrating its exceptional capabilities. Furthermore, the CS- 2 client, Supercomputer core Kaust, has been honored with the prestigious Gordon Bell Award, with the official announcement scheduled during SC ‘23.
Key Takeaways
After witnessing numerous AI hardware startups emerge, stagnate, and fade away over the past eight years, it is truly gratifying to witness the success of Cerebras and its acquisition of a substantial customer base, including commitments exceeding $800M. Undoubtedly, many aspiring players in the industry will look towards G42 and wafer-scale technology to shape the future of AI. Building and sustaining momentum is crucial in this dynamic landscape.
