Menu
Menu
QCT AI POD Solution
Software-defined architecture with cluster-level scalability for AI demands
QCT AI POD overview
QCT AI POD is a software-defined solution for cluster-scale AI infrastructure, featuring pre-integrated hardware and software, built-in monitoring, management and deployment tools as well as a ready-to-use AI development environment to accelerate AI adoption.
Key Benefits
Pre-configured Cluster Solution
Pre-validated Total Solution
Open Platform with Flexibility
Elevate Your AI Journey with QCT Software Service
QCT offers comprehensive software services that enable streamlined deployment, efficient management, advanced monitoring, and ready-to-use AI development.
Users can choose from the development environments tailored to their specific needs, ensuring flexibility and productivity across diverse workloads.
Bare-Metal Package
Cloud-Native Package
Converged Package
High-performance computing on
dedicated hardware with
traditional workload scheduling
Best suited for:
- Universities & Research Labs run scientific simulations such as CFD, molecular, and physics workloads.
- Enterprises support legacy HPC workloads that require bare-metal performance.
IT & Administrator
QCT AI POD supports simplified cluster deployment and management efforts with real-time system monitoring.
Simplified Cluster Management
System Monitoring & Alerting
Rapid Deployment and Provisioning
Account Management
and 2FA
and 2FA
Developer
QCT AI POD provides Cloud-Native and Bare-Metal environment with a web-based workspace, streamlining resource management and boosting development efficiency.
User Workspace
Provide multiple user interface options based on user development preferences.
Development Tools
Provide essential tools for code compilation, optimization, and execution
Orchestration
Integrate an advanced job scheduler to maximize resource utilization.
Experience QCT’s best-fit system for AI workload
Software-defined architecture with cluster-level scalability for AI demands,
offering multiple portfolio options with different scales tailored to the application scenario.
offering multiple portfolio options with different scales tailored to the application scenario.
Generative AI / Agentic AI
Computer Vision
LLM Fine-tuning & training
AI Factory
High-end multi-GPU nodes with high-speed networking deliver the throughput needed for Generative AI with RAG, ensuring efficient data transfer and supporting real-time summarization and low-latency interactions.
8 PCIe GPU SKU
Solution Positioning:
High-performance system for diverse AI and HPC applications
High-performance system for diverse AI and HPC applications
Networking
Infiniband Switch
(Cross-Compute Node Communication)
Ethernet Switch
(Network for Storage and Management)
(OOB Management)
Utility Node
Login Node/Admin &
Deployment Node
K8s Node
Compute Node
(Flexible in GPU option to support diverse workload types)
Storage Node
High Performance File Storage Node
Capacity storage node (optional)
8 GPU SXM Air-cooled SKU
Solution Positioning:
Purpose-built system for LLM fine-tuning and advanced deep learning training
Purpose-built system for LLM fine-tuning and advanced deep learning training
Networking
Infiniband Switch
(Network for Compute, Storage & Management)
Ethernet Switch
(Management and Storage Network)
(OOB Management)
Utility Node
Login Node/Admin &
Deployment Node
K8s Node
Compute Node
(HGX B300 with air-cooling)
Storage Node
High Performance File Storage Node
Capacity storage node (optional)
8 GPU SXM with Liquid Cooling SKU
Solution Positioning:
Enhanced efficiency for LLM fine-tuning and deep learning training at scale
Enhanced efficiency for LLM fine-tuning and deep learning training at scale
Networking
Infiniband Switch
(Cross-Compute Node Communication)
Ethernet Switch
(Network for Storage and Management)
(OOB Management)
Utility Node
Login Node/Admin &
Deployment Node
K8s Node
Compute Node
(HGX B300 with liquid-cooling)
Storage Node
High Performance File Storage Node
Capacity storage node (optional)
Cooling System
# OOB Management
8 GPU SXM with Liquid Cooling SKU
Solution Positioning:
Enhanced efficiency for LLM fine-tuning and deep learning training at scale
Enhanced efficiency for LLM fine-tuning and deep learning training at scale
Networking
Infiniband Switch
(Cross-Compute Node Communication)
Note: It can also be selected as an ethernet switch: NVIDIA Spectrum-4 SN5610
Ethernet Switch
(Network for Storage and Management)
(OOB Management)
Utility Node
Login Node/Admin &
Deployment Node
K8s Node
Compute Node
(Rubin NVL8 with liquid-cooling)
Storage Node
High Performance File Storage Node
Capacity storage node (optional)
Cooling System
CDU solutions can be further discussed based on customer requirements and data center conditions
Generative AI with RAG
QCT AI POD Engagement Process
QCT AI POD delivers a comprehensive approach to accelerate HPC and AI workload deployment, helping shorten the journey from
months to days while empowering partners to unlock greater agility, performance, and operational efficiency
QCT DevCloud Program
Join QCT DevCloud Program to experience latest solutions and systems.
QCT DevCloud offers a comprehensive HPC & AI environment. Users can experience QCT AI POD solutions and latest technology for remotely testing applications on various QCT hardware platforms.