Huawei Upgrades Stellar AI Fabric Solution to Build Efficient AI Computing Production Networks

[Shanghai, China, September 17, 2026] At HUAWEI CONNECT 2026, Huawei hosted the AI DC Innovation Summit under the theme "Leading AI DC Innovation, Shaping the Agentic World." The summit gathered over 600 global industry leaders and researchers to evaluate infrastructure design and sustainability requirements for agentic AI workloads. During the event, Huawei unveiled its upgraded Stellar AI Fabric Solution. The solution creates always-on data center networks with full computing power, helping enterprises build efficient production networks for dedicated AI compute clusters.

Keynote by Erick Zhang, President of Data Center Network Domain, Huawei Data Communication Product Line

Speaking at the summit, Erick Zhang noted that AI is shifting enterprise IT into an era of intelligent productivity, turning traditional data centers into computing hubs. In this agentic era, the network fabric dictates token throughput and overall compute yield.

To address these demands, Huawei upgraded its Stellar AI Fabric Solution. The release integrates the Rock-Solid Architecture 2.0, StarryWing Digital Map 2.0, Hyper-Converged Fabric (HCF) Architecture, and Network Packet Load Balancing (NPLB). Huawei also upgraded its NetMaster network AI agent. NetMaster automates root-cause analysis for 95% of network faults and isolates issues within minutes. These capabilities maintain consistent production uptime across high-density AI clusters.

On the hardware front, Huawei launched the CloudEngine SF9300 series UBG switches worldwide. Built on a two-layer multi-plane architecture, this platform cuts network buildout costs by 30%. Simplified UB forwarding lowers end-to-end latency by 40%. To ensure link stability, link-layer retransmission (LLR) executes microsecond-level recovery during packet errors, preventing data loss during transient link flapping.

Huawei also introduced the CloudEngine XH9300 series, featuring proprietary 100T and 51.2T Near-Packaged Optics (NPO) switches. The platform uses a centralized light source and proprietary 3.2T optical engine (OE) to reduce interconnect power draw by 40%. By eliminating the optical DSP (oDSP) stage, the NPO architecture lowers forwarding latency by 26%. In addition, a pluggable snap-in design for the OE increases hardware servicing efficiency tenfold.

Launch of the Technical Specification for Network High Availability of Financial Data Center

The summit also marked the release of a new financial industry standard. The Beijing FinTech Industry Alliance, in collaboration with Huawei and 17 industry contributors, issued the Technical Specification for Network High Availability of Financial Data Center.

The document establishes high-availability requirements and test methodologies across two categories and five network types, spanning both intra-DC and cross-DC environments. The framework evaluates availability across three core dimensions: architectural redundancy, degraded-state and fault recovery, and proactive fault prevention. This benchmark addresses operational continuity demands for financial institutions, establishes standard technical guidelines, and supports concurrent advancements in networking and AI.

Case Study: Institute of High Energy Physics, Chinese Academy of Sciences

Qi Fazhi, Director of the Computing Center at the Institute of High Energy Physics (IHEP), Chinese Academy of Sciences, presented deployment results achieved with Huawei. IHEP implemented the Huawei Stellar AI Fabric Solution to support large-scale scientific workloads.

A network-wide load balancing algorithm lifted computing efficiency by more than 10%. This approach eliminates traffic imbalances during distributed model training and clears bottlenecks across high-volume telemetry pipelines. As a result, researchers can process deep-space signals faster and with greater accuracy.

Huawei continues to collaborate with global enterprises and research partners to advance data center network engineering. By expanding R&D in intelligent networking and driving generational architectural shifts, Huawei aims to unlock greater operational and economic value for enterprise infrastructure.