Cerebras Systems, a Sunnyvale, California-based AI chipmaker said in San Francisco on August 18 that it introduced the CS-4 server rack for AI inference, the process of generating answers in chatbots.

The new system uses three WSE-3 Turbo chips and new networking components.

Cerebras said CS-4 is the first product based on its Nexus architecture, a modular design for compute, power, and input/output to make upgrades and deployment easier.

Cerebras said CS-4 follows its CS-3 system and uses 50% fewer components.

It also adds programmable input/output and direct wafer links to connect wafers within and across racks with lower latency.

The system uses chips made with TSMC’s 5-nanometer process and will be available in the third quarter.

Cerebras competes with Nvidia in inference hardware for workloads such as chatbot response generation.

The launch comes after Cerebras last week reported an adjusted loss of $6.9 million on sales of $180.1 million.

Earlier initial public offering disclosures said 87% of the company’s revenue came from G42.

🔗 Source: Reuters

Recent Cerebras developments