(JW Insights) Apr 11 -- Chinese telecoms equipment provider ZTE said it will debut a GPU server that can be used to train OpenAI's chatbot ChatGPT by December, reported Yicai Global citing online news outlet the Paper on April 7.

ZTE will launch devices that support large language model training, such as artificial intelligence servers, high-performance switches, and GPUs by year-end, said Xu Ziyang, president of the Shenzhen-based firm, who was talking during an earnings briefing on April 6.

ZTE has already started investing in the GPU server project by leveraging its advantages to improve the research and development efficiency of itself and that of its clients, Xu said. "In addition to strong computing power, the bandwidth should also be sufficient." In terms of software, the firm will integrate these capabilities into its digital information solutions, he added.

ZTE will also independently develop an AI accelerator chip to reduce processing costs, Xu said, adding that the firm intends to work with partners to integrate large language models with key production capacity.

ZTE has already teamed up with Baidu which started testing its ChatGPT-like Ernie Bot last month with select members of the public. The pair said that ZTE’s servers will support Ernie Bot to advance AI applications.

Last year, ZTE recorded a nearly 19% increase in net profit to about RMB8.1 billion ($1.2 billion) from 2021, according to its earnings report. Its revenue climbed by 7% to around RMB123 billion ($17.9 billion), said the Yicai Global report.