(Yicai) Dec. 12 -- ZTE will build a new generation of smart computing equipment to provide the huge computing power needed to support large language model training and reasoning, the Chinese telecoms equipment supplier told Yicai.

ZTE will develop computing equipment, such as high-performance artificial intelligence servers, switches, data processing units, AI accelerators and AI processors, to cater to the latest advances in artificial intelligence, the Beijing-based company said.

What LLMs need most is a large amount of computing power, ZTE said. By studying modeling and technologies that make equipment lighter, ZTE will be able to greatly slash the cost of LLM reasoning by making their training and application less expensive and more energy efficient.

ZTE, which commands over one-third of China’s wireless market, is applying its AI technologies across network management, office and operation systems. The firm is also building an efficient digital base for three main areas, namely computing cloud and communication networks, as well as individual, household and industry users, it added.

ZTE is aggregating upstream and downstream businesses in the industrial chain to form an industrial ecology. It released its Digital Nebula System, which uses AI technologies to help businesses build their own digital platforms, last year and version 2.0 was launched in April. 

Editors: Shi Yi, Kim Taylor