Cerebras Systems
Builder of the wafer-scale CS-3 AI system and WSE-3 chip, delivering up to 15x faster AI inference and training than GPU-based systems.
Cerebras Systems is a US AI compute company founded in 2015 and headquartered in Sunnyvale, California, that designs wafer-scale processors and systems purpose-built for ultra-fast AI training and inference. Its flagship Wafer-Scale Engine (WSE-3) is manufactured on TSMC's 5nm process and is the largest chip ever built, packing 4 trillion transistors and 900,000 AI-optimized cores onto a single 46,225 mm2 wafer with 44 GB of on-chip SRAM and 21 PB/s of SRAM bandwidth, delivering up to 125 FP16 PetaFLOPS. The WSE-3 is roughly 58x larger than leading GPUs, eliminating the inter-chip communication bottlenecks that constrain GPU clusters. The chip powers the CS-3 system, a 15U water-cooled unit that can train models up to 24 trillion parameters and, paired with external MemoryX memory and SwarmX interconnect, scales to clusters of up to 2,048 systems.
Cerebras also operates a cloud inference platform offering drop-in, OpenAI-API-compatible access to open frontier models such as GLM, Qwen, Llama, and GPT-OSS at world-record token speeds (over 2,000 tokens/second on some models), plus dedicated private-cloud endpoints and on-premises deployment. Customers and partners include OpenAI, AWS, Meta, GSK, Mayo Clinic, Notion, AlphaSense, Cognition, and Argonne National Laboratory, spanning agentic coding, real-time voice, enterprise search, drug discovery, and scientific computing. By delivering high-throughput, low-latency AI compute, Cerebras supplies the underlying inference and training infrastructure that powers AI-native applications including perception, reasoning, and autonomy workloads.
Primary type & automation activities this supplier delivers:
Contact Cerebras Systems
WEBSITE
https://www.cerebras.ai/HEADQUARTERS
United States
Company Facts
Founded
2015
Primary Role
Component Supplier
Company Size
-
Primary Region
North America
Annual Sales
-
Funding Stage
-
Funding Total
-
Keywords
- Cerebras Systems
- wafer-scale AI compute
- AI chips
- wafer-scale engine
- AI compute
- AI inference
- AI training
- CS-3
- WSE-3
- AI accelerator
- data center
- Component Supplier
- Cloud & Data
- Compute & Chips
- Education & Research
