Cerebras Systems
Cerebras builds wafer-scale AI computing systems and software intended to remove the memory, communication and distributed-computing bottlenecks that slow AI training and inference. It sells on-premise AI supercomputers and provides pay-as-you-go/cloud inference so organizations can build and deploy fast AI applications.
At a Glance
- AI-native startups and developers
- Enterprise software and agentic-AI companies
- Cloud and hyperscaler platforms
- Healthcare, pharma and genomics
- +4 more
AI Tools by Cerebras Systems
(1)
Cerebras
Wafer Scale AI Inference Cloud
Discussions
No discussions yet
Be the first to start a discussion about Cerebras Systems
Latest News
OpenAI GPT-5.3-Codex-Spark powered by Cerebras announced in research preview, running at over 1,000 tokens/second.
OpenAI and Cerebras announced a multi-year agreement to deploy 750MW of Cerebras wafer-scale systems for high-speed inference.
Cerebras raised $1.1 billion in Series G at an $8.1 billion post-money valuation.
Cerebras partnered with Carahsoft and joined the U.S. DoD Tradewinds Solutions Marketplace.
Products & Services
Cerebras’s first wafer-scale AI processor, designed to put compute, memory and bandwidth on one very large chip.
The first Cerebras data-center system built around WSE-1.
Second-generation wafer-scale processor and system for large-scale AI training and scientific computing.
Third-generation wafer-scale processor and AI supercomputer for training and inference.
Market Position
Cerebras positions its wafer-scale architecture as a specialized alternative to GPU-based AI infrastructure, emphasizing substantially higher inference speed, simpler single-device training and lower latency. Its competitive set includes NVIDIA and AMD accelerators, Google TPU, AWS Trainium/Inferentia and other AI-cloud providers; its differentiation is the very large single wafer-scale processor and vertically integrated systems/software stack.
Leadership
Founders
Andrew Feldman
Co-founded SeaMicro with Gary Lauterbach; became Cerebras CEO and led the company from its founding.
Gary Lauterbach
SeaMicro co-founder and computer-architecture technologist; Cerebras co-founder and CTO, emeritus.
Michael James
Former SeaMicro colleague specializing in software and compilers; Cerebras Chief Architect, Advanced Technologies and co-founder.
Sean Lie
Former SeaMicro server-architecture engineer; Cerebras CTO and co-founder.
Jean-Philippe Fricker
Former SeaMicro colleague and hardware engineer; Cerebras Chief System Architect and co-founder.
Executive Team
Andrew Feldman
Chief Executive Officer and Co-Founder
Previously co-founded SeaMicro; leads Cerebras.
Sean Lie
Chief Technology Officer and Co-Founder
Former SeaMicro architecture engineer and founding Cerebras technologist.
Board of Directors
Founding Story
After AMD acquired SeaMicro in 2012, the five founders set out to tackle the mismatch between conventional processors designed for graphics and the requirements of deep learning. They founded Cerebras to make wafer-scale computing commercially viable, despite the feasibility being uncertain at the time.
Business Model
Revenue Model
Sells on-premise AI systems and related software/services, dedicated training and inference capacity, and API/cloud inference on a pay-as-you-go or enterprise contracted basis.
Pricing Tiers
GPT OSS 120B: $0.35/M input and $0.75/M output tokens; Qwen 3.8 27B: $0.99/M input and $1.49/M output tokens.
Production capacity, higher rate limits, enhanced latency options, dedicated account team, custom weights, fine-tuning and training by agreement.
Target Markets
- AI-native startups and developers
- Enterprise software and agentic-AI companies
- Cloud and hyperscaler platforms
- Healthcare, pharma and genomics
- Energy and scientific research
- Federal, state and local government and defense
- Real-time LLM and generative-AI inference
- Training and fine-tuning large language models
- Agentic AI and coding agents
- Drug discovery, genomics and medical research
- Molecular dynamics and scientific/HPC simulation
- Energy and climate modeling
- Meta Llama API
- OpenAI
- Perplexity
- Mistral