(Feb 13): Chinese artificial intelligence (AI) model developer Zhipu is planning a second listing in Shanghai following its recent US$558 million ($705.31 million) initial public offering (IPO) in Hong Kong, a sign of confidence that the country’s generative AI rally still has legs.
The company formally known as Knowledge Atlas JSC Ltd has hired Guotai Haitong and CICC to advise its listing on Shanghai’s Star Market, according to a document filed on Wednesday. Its stock jumped as much as 23.4% on Friday to a record HK$496, extending its gain since its Jan 8 debut to about 320%.
Zhipu was part of a rush in Hong Kong listings by Chinese technology companies, including OpenAI challenger MiniMax Group Inc and chip designer Montage Technology Co, which contributed to a record January in the city. Companies that listed in Shanghai, such as those of Moore Threads Technology Co and Metax Integrated Circuits Shanghai Co posted triple-digit first-day gains.
But Zhipu is the first of the group to seek the less frequent path of going public in Hong Kong and later venturing into a second listing on the mainland, the opposite route being far more popular. The unusual move may help the company tap into new investment pools and take advantage of mainland shares trading at a higher valuation to Hong Kong counterparts. Shanghai’s chip-heavy Star 50 Index is up more than 9% this year, compared with an almost 3% decline in the Hang Seng Technology Index.
“We continue to observe a valuation premium for A-shares over H-shares,” said Chelsey Tam, a senior equity analyst at Morningstar. “Listing on the mainland also allows companies to broaden their investor base and improve access to domestic capital.”
See also: OpenAI debuts first model using chips from Nvidia rival Cerebras
Should the deal go ahead, the company will be counting on investor excitement over its latest large-language model, GLM-5, which surpassed a competitor from US-listed Moonshot AI to clinch the top place among open-source models on ranking site Artificial Analysis.
“That shows Chinese AI models continue to innovate on efficiency improvements in both compute and memory, and using local chips given limited access to Nvidia chips. This is an important positive direction for China AI given skyrocketing memory prices and expensive compute,” said Jefferies analysts including Edison Lee in a note.
Uploaded by Tham Yek Lee
