The one AI {Hardware} startup to comprehend income exceeding $100M has completed the primary part of Condor Galaxy 1 AI Supercomputer with companion G42 of the UAE. Different Cerebras clients are sharing their CS-2 outcomes at Supercomputing ‘23, constructing momentum for the inventor of wafer-scale computing. This firm is on a tear.
4 brief months in the past, Cerebras introduced essentially the most vital deal any AI startup has been capable of assemble with companion G42 (Group42), a synthetic intelligence and cloud computing firm. The eventual 256 CS-2 wafer-scale nodes with 36 Exaflops of AI efficiency might be one of many world’s largest AI supercomputers, if not the biggest.
Cerebras has now completed the primary knowledge middle implementation and began on the second. These two firms are shifting quick to capitalize on the $70B (2028) gold rush to face up Massive Language Mannequin companies to researchers and enterprises, particularly whereas the provision of NVIDIA H100 stays tough to acquire, creating a chance for Cerebras. As well as, Cerebras has not too long ago introduced it has launched the biggest Arabic Language Mannequin, the Jais30B with Core42 utilizing the CS-2, a platform designed to make the event of large AI fashions accessible by eliminating the necessity to decompose and distribute the issue.
Along with Cerebras’s progress on Condor Galaxy, the corporate and Argonne Nationwide Labs have introduced a 130-fold speedup over the NVIDIA A100 on a Monte Carlo particle simulation software utilizing a CS-2 wafer-scale engine. One other CS-2 buyer, supercomputer middle Kaust, is a finalist for the celebrated Gordon Bell Award, to be introduced on Tuesday at SC’23.
Conclusions
After eight years of watching AI {hardware} startups come, fade, and go, it’s extremely satisfying to see one (alas, just one) succeed and land some dozen clients, certainly one of them at a dedication degree exceeding maybe $800M. Others will certainly observe G42 and undertake the wafer-scale know-how to construct the following technology of AI. Momentum issues.
Disclosures: This text expresses the creator’s opinions and shouldn’t be taken as recommendation to buy from or put money into the businesses talked about. Cambrian AI Analysis is lucky to have many, if not most, semiconductor companies as our purchasers, together with Blaize, BrainChip, Cadence Design, Cerebras, D-Matrix, Eliyan, Esperanto, FuriosaAI, Graphcore, GML, IBM, Intel, Mythic, NVIDIA, Qualcomm Applied sciences, Si-5, SiMa.ai, Synopsys, Ventana Microsystems and Tenstorrent. We have now no funding positions in any of the businesses talked about on this article and don’t plan to provoke any within the close to future. For extra data, please go to our web site at