Huawei Unveils Stellar AI Fabric for Data Center Networks
Huawei launched its upgraded Stellar AI Fabric at HUAWEI CONNECT 2026, claiming 30% cost savings and higher uptime for AI computing clusters.
Key points
- Huawei claims its new switch architecture cuts data center buildout costs by 30% and reduces latency by 40%.
- An automated AI agent identifies the root cause of 95% of network faults within minutes to maintain uptime.
- A new industry standard for financial data centers was released by Huawei and 17 partners to define network availability baselines.
Huawei has upgraded its Stellar AI Fabric solution to address the growing demand for stable and efficient networks in artificial intelligence data centers. The announcement, made during the HUAWEI CONNECT 2026 summit in Shanghai, targets production environments where network reliability directly impacts compute output. The company stated that the new architecture aims to maximize token throughput by reducing bottlenecks in data transmission.
According to a report distributed via PR Newswire, the upgraded system integrates several new components designed to automate fault detection and balance network loads. Huawei claims these additions will help maintain consistent production uptime, a critical factor for organizations running large-scale AI models that require continuous data flow.
Hardware cuts costs and latency
The core of the upgrade involves two new switch families. The CloudEngine SF9300 uses a two-layer architecture that Huawei says reduces buildout costs by 30 percent. It also simplifies data forwarding to lower end-to-end latency by 40 percent. The system includes a feature for microsecond-level link recovery, which prevents packet loss during temporary connection drops.
The CloudEngine XH9300 focuses on power efficiency and speed. By removing a specific optical processing stage, the design cuts forwarding latency by 26 percent. It also lowers interconnect power consumption by 40 percent. Maintenance is faster due to pluggable optical engines, but these proprietary components may limit compatibility with third-party equipment.
Automated fault analysis speeds recovery
A new network AI agent automates the analysis of network faults. Huawei states that this tool can identify the root cause of 95 percent of issues within minutes. This automation is intended to reduce downtime and manual intervention, allowing the network to maintain consistent performance during heavy workloads.
Industry standards and research results
Alongside the hardware launch, Huawei collaborated with 17 partners to release a technical specification for financial data centers. This standard defines baselines for network availability and fault recovery in both single and multi-site environments. In a separate scientific deployment, the Institute of High Energy Physics reported that using the Stellar AI Fabric increased computing efficiency by more than 10 percent.
