Cerebras Systems announced on Tuesday a new version of its server hardware called the CS-4, featuring its dinner-plate-sized chips designed to accelerate AI chatbot queries. According to reports from Business Standard, the new system is a server rack powered by three large chips that the company designs, which Cerebras said translates into better performance. The hardware is based around the company's Nexus server architecture, which includes pluggable modules that house the chips. As per the latest market data, Cerebras has been recognized as a leading AI platform with $250M+ in funding and 501-1,000 employees, positioning it among the top AI infrastructure companies in the current market landscape.
Cerebras has achieved a significant performance milestone with its CS-4 system delivering up to 30 times faster processing than GPU-based solutions. As reported by Business Standard, the company's chips already demonstrated 21.6 petabytes per second (PB/s) of memory bandwidth, which was 1,000x faster than Nvidia's or AMD's best GPUs. The new machine is available in the third quarter and the chips are fabricated with the TSMC 5-nanometer manufacturing process. The server rack includes a chip called WSE-3 Turbo and new networking components that Chief Technology Officer Sean Lie said would speed data movement between the chips.
The CS-4 system delivers up to 10x more throughput per watt compared to previous generations, representing a significant improvement in energy efficiency. According to Business Standard, Cerebras' chips gain a speed advantage because they are large enough to avoid the energy and slowdown of moving data from one chip to another. The company also designed the new system to be easier to set up with 50% fewer components, which Chief Technology Officer Sean Lie said at a media briefing in San Francisco would speed data center construction. The company plans another generation of the chip and server in 2027.
CEO Andrew Feldman stated that the company's engineering plans are focused on speeding the amount of data its future chips and systems can crunch. As reported by Business Standard, Feldman said at the briefing, "We're going to get four times as fast between now and the end of the year, end of 2027, and we're going to get 20 times more throughput." The company's engineering focus targets the portion of AI called inference, the computing process of generating an answer in a chatbot such as Anthropic's Claude. Cerebras competes directly with Nvidia in the AI hardware and chip market, targeting the inference portion of AI computing processes.
According to Business Standard, last week, Cerebras reported an adjusted loss of $6.9 million on sales of $180.1 million. The company competes with Nvidia in the AI hardware and chip market, targeting the inference portion of AI computing processes.
