Product overview
Purpose-built accelerator for enterprise AI workloads
Designed for dense inference, model adaptation, and private AI infrastructure where predictable availability, localized supply, and software compatibility are critical.
Architecture
MUSA
Memory & Performance
High-bandwidth memory profile for AI inference and training.
VRAM capacity80
Memory typeHBM3e
Memory bandwidth1600
Interconnect typeMTLink
Interconnect speed784
Architecture
Compute architecture and software execution model.
ArchitectureMUSA
Compute units-
Power & Thermal
Data center integration requirements.
Thermal design power450
CoolingПассивное
Form factorOAM
Pixel Rate1024
Texture Rate1024
Benchmarks
Peak theoretical compute for common AI precisions.
FP64
16
FP32
32
FP16
32
TF32
256
BF16 Tensor
512
FP8 Tensor
1024
INT8 Tensor
1024
Compatibility
Interfaces, frameworks, and deployment environment.
PCIe interfacePCIe 5.0x16
Video encoding16x1080p30fps
Video decoding128x1080p30
Physical Dimensions
Card dimensions for server platforms.
Slots-
Length- mm
Height- mm
Width- mm
Pricing
On request
Volume pricing available for cluster deployments and pilot batches.
Documentation