ZTE's first 'flagship GPU server' for China has NVIDIA H800 AI GPUs, works around US sanctions

ZTE has just unveiled its very first "flagship GPU server" for the Chinese market, powered by Intel Xeon Scalable processors and NVIDIA H800 AI GPUs. Check it out:

2

China hasn't stopped with its pursuits in artificial intelligence (AI) with multiple Chinese companies making their own "AI solutions" because of it. ZTE is known for its smartphones, networking, enterprise solutions, and more, but now the Chinese company has jumped into the AI industry with the introduction of the ZTE R6900 G5 flagship GPU server.

The new ZTE R6900 G5 flagship GPU server packs an undisclosed Intel Xeon Scalable processor (we don't know how many, or which models, clock speeds, etc) as well as 8 x NVIDIA H800 NVLink GPU modules or 8 x OCP OAM 8-GPU modules. Networking-wise, the new ZTE R6900 G5 server supports IB/RoCE networking between individual notes with up to 4.8TB/sec of non-blocking bandwidth.

ZTE is using GPU+CPU dual liquid cooling technology, which should significantly reduce the fan speed and power consumption of the R6900 G5 flagship GPU server by more than 1000W. This means the R6900 G5 flagship GPU server is a great choice for companies in China who want to run large-scale training applications for AI, with more performance and lower power consumption (and less heat).