US Markets
Home›US Markets›Equities›Cerebras unveils CS-4 rack-scale AI system aimed at Nv…
Cerebras unveils CS-4 rack-scale AI system aimed at Nvidia
Cerebras says the CS-4 delivers 30x more tokens per second per user than GPUs, while shares slid more than 35% to $218 as of midday Tuesday after weaker-than-expected results.
Cerebras has unveiled its newest rack-scale AI system, the Cerebras CS-4, positioning it as an alternative to Nvidia for inference workloads. The company claims the system delivers 30x the tokens per second per user versus graphics processing units, and says it uses three wafer-scale WSE-3 Turbo processors, built with 4 trillion transistors.
The CS-4 is designed for AI inference rather than training, relying on static random-access memory instead of dynamic random-access memory. Cerebras said SRAM is faster but more complex and expensive than DRAM, and it also argued that its single product design shortens data travel compared with multi-chip approaches.
Beyond hardware, the company runs its own AI cloud services on the chips it rents out to customers. Cerebras also went public in May at $185 per share and began trading at $350, but the stock has since fallen sharply, with shares down more than 35% to $218 as of midday Tuesday.
The stock decline also followed its Q2 earnings, which included a loss per share of -$2.98, compared with a profit of $1.91 in the same quarter last year. While Cerebras offered better-than-anticipated Q3 guidance, Yahoo Finance reports it was not enough to impress investors.