CS-3 System
Provider 
Founded 2016
Sells To —
Pricing Model —
Ownership —
Cerebras Systems offers CS-3 System
ProductOperating Status: Currently Offered
Wafer-scale AI supercomputer in mini-fridge form with cluster-scale power.
Wafer-scale AI supercomputer in mini-fridge form with cluster-scale power.
The CS-3 system deploys your own AI & HPC supercomputer with highest levels of performance, privacy, and control. At its heart is the innovative engine block wafer packaging with direct power delivery and uniform closed internal water cooling (all supplies redundant and hot-swappable).
900,000 AI-optimized cores, 214 Petabits/s interconnect bandwidth, 44GB on-chip SRAM (250× more), 21PB/s memory bandwidth (2,625× more), 1.2Tb/s system I/O. Purpose-built for large-scale AI workloads with better performance, lower power, smaller footprint, and fraction of the cost of GPU-based systems.
| Attribute | CS-3 System | DGX, HGX | |
|---|---|---|---|
| Provider | |  | |
| Founded | 2016 | 1993 | |
| Sells To | — | Consumers, Enterprise | |
| Pricing Model | — | Licensing, Product Sales, Subscription | |
| Ownership | — | Public |
Provider
Founded 2016
Sells To —
Pricing Model —
Ownership —
Provider
Founded 1993
Sells To Consumers, Enterprise
Pricing Model Licensing, Product Sales, Subscription
Ownership Public
Dedicated physical GPU servers without virtualization overhead for maximum AI performance.
PowerEdge XE9680 is Dell's 6U GPU-dense server for AI/ML supporting up to eight NVIDIA, AMD, or Intel accelerators.
Dedicated physical GPU servers without virtualization overhead for maximum AI performance.
Rack-mounted AI inference system integrating proven compute IP with photonic interconnect and large, high-bandwidth memory.
Rack-mounted AI inference system integrating proven compute IP with photonic interconnect and large, high-bandwidth memory.
IBM Power Systems are enterprise servers built on POWER processors for mission-critical Linux, AIX, and IBM i workloads.
AMD Instinct is a family of data-center accelerators for AI and high-performance computing.
Tenstorrent scale-out server family for production AI training and inference.
RDU is SambaNova's Reconfigurable Dataflow Unit, the company's AI chip family for inference and training at scale.
Trainium is AWS's purpose-built AI accelerator chip family for training and inference of large generative models.
IBM Power Systems are enterprise servers built on POWER processors for mission-critical Linux, AIX, and IBM i workloads.
Intel Gaudi is a data-center AI accelerator product family.
AMD Instinct is a family of data-center accelerators for AI and high-performance computing.
Tenstorrent scale-out server family for production AI training and inference.
PowerEdge XE9680 is Dell's 6U GPU-dense server for AI/ML supporting up to eight NVIDIA, AMD, or Intel accelerators.
RDU is SambaNova's Reconfigurable Dataflow Unit, the company's AI chip family for inference and training at scale.
TPU is Google's custom AI accelerator ASIC family delivered through Google Cloud TPU Pods for ML training and inference.
Trainium is AWS's purpose-built AI accelerator chip family for training and inference of large generative models.
Groq LPU is a custom inference chip with on-chip SRAM and deterministic scheduling delivering low-latency AI inference for LLMs.
SambaNova DataScale is a data-center AI computing platform.