ZTE, a Chinese telecom equipment maker, unveiled its OEX supernode at the “ZTE Full-Stack Intelligent Computing Ecosystem Forum” during the 2026 World Artificial Intelligence Conference on July 18.

The company said the system uses what it calls an orthogonal backplane-free architecture with a “three-zero cable” design, along with standardized protocols and unified interfaces.

ZTE said this makes the product compatible with China’s broad graphics processing unit ecosystem.

The company also said the supernode supports different multi-chip combinations and links high-speed GPU interconnects within a machine, with lossless switching between machines, to improve cluster efficiency and shorten deployment time.

The launch comes as export controls from the US have constrained access to advanced Nvidia GPUs in China, adding uncertainty to AI infrastructure supply.

ZTE is entering a segment where Huawei has also introduced a domestic AI supernode cluster that does not rely on Nvidia’s advanced chips.

🔗 Source: Security Times