DTCT-AIDC Liquid-cooled Modular Micro Edge Intelligent Computing All-in-one Machine
DTCT-AIDC modular micro edge intelligent computing all-in-one machine (liquid cooling)
Dtct-Aidc Modular Micro Edge Intelligent Computing All-In-One Machine (Liquid Cooling)
PRODUCT OVERVIEW
PRODUCT OVERVIEW
The most silent, energy-efficient, and powerful edge computing solution, tailored for high-end sensitive scenarios.
PRODUCT FEATURES
PRODUCT FEATURES
01
Ultra-Quiet Operation
Liquid cooling cuts fan noise, ideal for noise-sensitive spaces.

02
Ultra-High Compute Density
Up to 60kW per cabinet, delivering over 2x computing power of air-cooled units in the same footprint.

03
Ultra-Low Energy Consumption
PUE<1.1, over 30% less energy than air cooling, greatly cutting long-term operating costs.

04
High-Density Mini Integration
All-in-one integration of compute, storage, network, liquid cooling and monitoring, compact size for flexible deployment.

PRODUCT ADVANTAGES
PRODUCT ADVANTAGES
Construction Period
Space Occupation
Equipment Integration
Compatibility
Energy Saving
Operation
After-sales Maintenance
Modular Data All-in-One
Within 7 Days

Modular rack deployment, highly integrated to save room space

One-stop supply, pre-integrated for unified management

Factory pre-integration & pre-testing with great compatibility

Built-in rack air conditioner cools only cabinet devices, saving over 50% energy vs traditional solutions

Fully enclosed cabinet for high cleanliness and stable operation

Unified multi-level maintenance; 7×24 unattended operation to cut O&M costs
Traditional Computer Room
After-sales Maintenance
Subsystems including UPS, air conditioner and cabinets require separate spaces
Multi-supplier, separate construction and independent management
Complex commissioning & poor compatibility for integrated multi-vendor products
PUE: 2.0-3.0 with low cooling efficiency
Dust accumulation risks unstable operation and shortens equipment lifespan
Multi-vendor maintenance leads to ambiguous responsibility and slow response for related faults
PRODUCT SPECIFICATIONS
PRODUCT ADVANTAGES
| Core Configuration Items | Liquid-cooled Pro Model | Liquid-cooled Max Model | Liquid-cooled Ultra Model | ||||||||
| Infrastructure Base | Cabinet Cluster Specification | 2-cabinet Unit | 3-cabinet Unit | 4-cabinet Unit | |||||||
| Green Variable-frequency Refrigeration System | 6kW Rear Door Air Conditioner, 30kW CDU | 6kW Rear Door Air Conditioner*2, 30kW CDU*2 | 6kW Rear Door Air Conditioner*3, 30kW CDU*3 | ||||||||
| Power Supply & Distribution System | 40KVA/36KW Rack-mounted UPS | 40KVA/36KW Rack-mounted UPS*2 | 40KVA/36KW Rack-mounted UPS*3 | ||||||||
| Battery Backup System | Valve-regulated Lead-acid Battery or Lithium Battery Optional | ||||||||||
| In-cabinet Power & Environment Monitoring System | Basic in-cabinet temperature & humidity monitoring + IT equipment monitoring + O&M Intelligent Agent | ||||||||||
| HMI Touch Screen | 10-inch Embedded Touch Screen | ||||||||||
| Standard General Accessories | Cabinet lighting, emergency fan, blue background lighting assembly | ||||||||||
| IoT Communication Module | Mobile APP + VPDN IoT communication + Cluster-level remote management | ||||||||||
| Core IT Hardware | Core Computing Parameters | 32-card cluster, ≥64GB per card, total video memory computing power ≥9.4 PFLOPS | 40-card cluster, ≥64GB per card, total video memory computing power ≥11.7 PFLOPS | 48-card cluster, ≥64GB per card, total video memory computing power ≥14.1 PFLOPS | 56-card cluster, ≥64GB per card, total video memory computing power ≥16.4 PFLOPS | 64-card cluster, ≥64GB per card, total video memory computing power ≥18.8 PFLOPS | 72-card cluster, ≥64GB per card, total video memory computing power ≥21.1 PFLOPS | 80-card cluster, ≥64GB per card, total video memory computing power ≥23.5 PFLOPS | 88-card cluster, ≥64GB per card, total video memory computing power ≥25.9 PFLOPS | 96-card cluster, ≥64GB per card, total video memory computing power ≥28.2 PFLOPS | 104-card cluster, ≥64GB per card, total video memory computing power ≥30.6 PFLOPS |
| General Computing Server System | Optional | Optional | Optional | Optional | Optional | Optional | Optional | Optional | Optional | Optional | |
| Storage System | Local SSD Storage: 1.92TB×4 | Local SSD Storage: 1.92TB×5 | Local SSD Storage: 1.92TB×6 | Local SSD Storage: 1.92TB×7 | Local SSD Storage: 1.92TB×8 | Local SSD Storage: 1.92TB×9 | Local SSD Storage: 1.92TB×10 | Local SSD Storage: 1.92TB×11 | Local SSD Storage: 1.92TB×12 | Local SSD Storage: 1.92TB×13 | |
| Expandable | Expandable | Expandable | Expandable | Expandable | Expandable | Expandable | Expandable | Expandable | Expandable | ||
| Security Protection System | Enterprise-grade Firewall | ||||||||||
| Network System | 25GbE Management Switch*1 + 100GbE Computing Switch*1 | 25GbE Management Switch*1 + 100GbE Storage Switch*1 + 200Gb Computing Switch*1 | 25GbE Management Switch*1 + 100GbE Storage Switch*1 + 400Gb RoCE Computing Switch*1 | ||||||||
| Platform Software Layer | AI Computing Platform | Basic training & inference base, unified resource pool management, supporting single model deployment, lightweight fine-tuning and basic multi-tenant isolation | Parallel training & inference, supporting resource pooling scheduling, concurrent job execution and department-level operation support | Enterprise-level training-integration stage, stably supporting parallel development of multiple projects, model service release and continuous delivery | High-concurrency inference and incremental fine-tuning closed loop, supporting continuous business optimization, model iteration and unified service governance | Enterprise core computing platform, supporting distributed training, disaster recovery monitoring, unified O&M and service-level governance | Provide high-throughput training & inference, elastic scheduling and cost optimization for multiple business lines, realizing cross-team shared computing base | Capable of serving large models and MoE deployment, supporting massive concurrent inference, high-availability operation and unified SLA management | Support stable service of super-large models, multi-tenant QoS and key business continuity, entering group-level resource supply phase | Support unified orchestration of super-large parameter clusters, cross-domain resource scheduling and multi-cluster collaboration to meet regional or group-level computing coordination | Build flagship computing base, supporting integrated training & inference of super-large models, heterogeneous cluster collaboration and external export of platform capabilities |
| AI Agent Platform | Supports single-agent orchestration, basic Agent Skill access and preliminary adaptation of OpenClaw & Harness | Supports workflow orchestration, Skill plugin management and OpenClaw toolchain linkage to satisfy automated multi-task processing | Supports multi-agent collaboration, Harness evaluation regression and knowledge enhancement process to improve capability of complex task decomposition | Supports orchestration engine, Agent Skill access, process execution governance and human feedback closed loop for production system implementation | Supports orchestration base, full-link evaluation, gray release, audit tracking and permission governance to ensure controllable agent production | Supports long-chain execution of multi-agent, complex plan decomposition, cross-system collaboration and audit closed loop for organizational reuse | Deeply adapts orchestration engine, OpenClaw execution kernel and Harness quality control system to form agent factory capability | Supports orchestration governance, Skill permission sandbox, complex tool calling and domain agent combined orchestration to improve autonomy of complex business | Compatible with full-stack capabilities of agent ecosystem, supporting continuous evaluation & optimization, cross-system collaborative execution and closed-loop autonomous operation | Build enterprise-grade agent operating system covering orchestration base to autonomous evolution, supporting platform operation and continuous self-optimization | |
| Model Pre-installation Layer | Baseline Pre-installed Models (Continuous Update) | GLM5.1、Qwen3.6-35B、DeepSeek-32B | GLM5.1、Qwen3.6-35B、DeepSeek-32B | GLM5.1、Qwen3.6-35B、DeepSeek-70B | GLM5.1、Qwen3.6-35B、DeepSeek-70B | GLM5.1、Qwen3.5-122B-A10B、DeepSeek-70B | GLM5.1、Qwen3.5-122B-A10B、DeepSeek-70B | GLM5.1、Qwen3.5-122B-A10B、DeepSeek-671B | GLM5.1、Qwen3.5-122B-A10B、DeepSeek-671B | GLM5.1、Qwen3.5-397B-A17B、DeepSeek-671B | GLM5.1、Qwen3.5-397B-A17B、DeepSeek-671B |
| Scenario Application Layer | Baseline Pre-installed Scenarios | Industrial Intelligent Quality Inspection, Knowledge Base Q&A, Intelligent Customer Service | Document Review, Work Order Sorting, O&M Q&A | Contract Review, Investment Research Assistant, Defect Detection | Bill Audit, Public Opinion Analysis, Production Scheduling Optimization | Policy Interpretation, Predictive Maintenance, Risk Control Assistant | Supply Chain Collaboration, Intelligent Audit, Teaching & Research Generation | Digital Twin Analysis, Financial Compliance, Smart Teaching | Government Hotline, Cross-factory Quality Inspection, Compliance Audit | Urban Governance Analysis, Full-link Risk Control, Research Assistance | Regional Knowledge Hub, Emergency Command, Operation Dashboard |
| Industry Solution Support | Exclusive full-industry and cross-industry customized solutions | ||||||||||
| ▋ Six Major Functions | ● Fusion Analysis ● Comprehensive Monitoring ● Topology Management ● Unified Management ● Asset Management ● Auxiliary Management |
|
| ▋ Five Major Advantages | Centralized Control | Centralized management of multiple all-in-one machines |
| Diversified Monitoring | Multiple monitoring modes including large-screen mode, list mode and 3D mode | |
| Real-time Monitoring | Full-screen dynamic display of real-time computer room monitoring data and alarm data | |
| Green Product Design | Supports computer room energy consumption management and PUE display | |
| Systematic Management | Supports access control and video management; flexible multi-level alarm configuration for alarm linkage | |
| ▋ Touch Screen | Dynamic environment monitoring data is synchronously displayed on the all-in-one machine touch screen Allows users to view device data intuitively Administrators can judge the computer room operating status immediately via the touch screen |
|