Huawei will Launch Intelligent Computing O&M Solution at MWC 2026 for Cluster High Availability and Business Stability

[Barcelona, Spain, March 1, 2026] Currently, AI computing power construction is entering the fast lane, and various AI applications are accelerating into core production stages of various industries. This requires the operation of computing power infrastructure to be as efficient, stable, and reliable as network infrastructure. To achieve this goal, Huawei will launch MindOps Intelligent Computing O&M Solution during MWC 2026.

Enabling Compute-Storage-Network Collaborative O&M, Visibility and Manageability

By building a 7-layer digital twin of AIDC and collaborating with the professional large model in the field of intelligent computing O&M (EDNS 2.0), Huawei release MindOps, a new generation of intelligent computing O&M solution that integrates network and computing power infrastructure. Through the intelligent and efficient O&M of compute-storage-network collaboration, it ensures the efficient and stable operation of training and inference services. Through 7-layer visibility, risk prediction perception, and real-time proactive fault repair, it realizes the efficient and long-term stable operation of the computing power platform, increasing the availability of computing clusters from the industry average of 90% to 99.9%.

Intelligent Computing 7-Layer Visibility and Manageability

Three Core Key Capabilities: ·

Equipment Health Self-check: Shifting from reactive response to proactive risk elimination

Based on risk perception algorithms, periodic health checks are performed on key components such as liquid cooling, CDU, and optical modules. Before a failure occurs, the system can realize automatic risk warnings, supporting O&M personnel to quickly complete risk elimination and ensuring the long-term stable operation of the intelligent computing center.

Intelligent Computing 7-Layer Observability: Building "second-level" running status perception capability

Based on digital twin technology, 7-layer observability is achieved from L1 data center infrastructure, L2 computing cluster infrastructure, RoCE network, and collective communication to AI platforms, models, and applications, achieving "second-level" perception of the system's running status.

Automated Fault Diagnosis and Switchover: Pre-set with AI professional large models, creating 99.9% high availability

Addressing common problems and challenges such as slow cards, slow networks, and model performance degradation in typical scenarios, MindOps, empowered by the professional large model in the intelligent computing O&M field (EDNS 2.0), realizes risk perception and prediction, shortens precise system fault demarcation and location to minute-level, and achieves 99.9% computing power high availability through timely automatic switchover.

The intelligent computing O&M solution MindOps, by incorporating digital twins and the professional field large model EDNS 2.0, achieves observability and automated troubleshooting, redefining the paradigm of intelligent computing O&M, ensuring the long-term stable operation and ultimate performance of computing power platforms, and letting AI release new quality productive forces.