VasEdge
A localized, immersion-cooled AI computing unit that brings data-center-grade performance into any normal room, without the noise, heat, or infrastructure requirements of traditional servers.
Engineered for precision.
| Available Capacity | 16U |
| Power Consumption | 12kW |
| Immersion Coolant | Dielectric Coolant |
| Equipment Noise | ≤55dB |
| Dimensions (L×W×H) | 990 × 670 × 1109 mm |
| Equipment Weight | 190kg / 360kg (full load) |
| Power Supply | 400V |
| Operating Temp | -10°C to 40°C |
| Cooling Capacity | 10kW |
| Redundancy | Active + Standby Pump, Dual Fans |
| Monitoring | Liquid Level, Flow, Pressure, Temp (Modbus-TCP) |
| Dry Cooler Power | 1kW |
| Dry Cooler Dimensions | 1080 × 362 × 720 mm |
| Dry Cooler Weight | 70kg |
Flexible configuration.
The VasEdge supports BYD modular servers with an innovative block-style architecture. CPU, GPU, and storage modules can be freely combined, reducing unnecessary initial hardware investment while enabling easy maintenance and upgrades.
GPU Module
3U / 4U12 PCIe Slots, supporting 8 dual-width GPUs and 4 NICs
CPU Module
1U2× Intel EagleStream CPU AI-specific Motherboards
Storage Module
1U6× 2.5-inch Drives
Hybrid Modules
VariableFlexible configuration for mixed workloads
Ready-to-deploy AI packages.
DeepSeek 32B Standard Edition
Optimized for standard inference workloads with excellent cost-performance ratio.
DeepSeek 70B Pro Edition
Enhanced capability for complex reasoning and enterprise-grade AI applications.
Full-Powered Edition (671B)
Maximum performance configuration supporting the complete DeepSeek 671B model.
Custom NVIDIA Configs
L40S / H200 / RTX 5090 / A100 — tailored to your specific GPU requirements.
Plug-and-play simplicity.
The VasEdge is designed for rapid deployment in any environment. Connect to 400V power and an external dry cooler, and go live within 4 hours.