Chinese-listed shares have declined by nearly $2.5 trillion in market capitalization as of October 28 from their peak on May 13, according to calculations by the magazine. This drop is largely explained by growing investor doubts about the profitability of companies in the AI sector and by the prolonged stagnation of consumption in China.
- The stock of Unitree, the robotic manufacturer and one of the world leaders in the humanoid-robot market, has fallen more than a third from its first day of trading in August.
- CXMT, the largest Chinese DRAM memory chip maker, has seen its shares shed more than a quarter from their August mid-peak, i.e., about $170 billion.
- MetaX and Moore Threads, two companies specializing in the design of graphics processors used in data centers, have lost half their value since July.
Despite last week’s intervention by the “national team” — state-linked funds and brokers tasked with carrying out massive stock purchases to stabilize market prices — investors in mainland China continued to sell their shares in Hong Kong: HK$20 billion ($2.5 billion) since the start of the month, a monthly record since July 2021, according to the Financial Times .
Results expected this week could prolong the decline.
- Tomorrow, Thursday the 29th, two Chinese companies that manufacture equipment used notably in AI-dedicated data centers are set to publish their financial results.
- These include Zhongji Innolight, one of the world’s leading manufacturers of optical modules, essential for linking servers in data centers, and Victory Giant Technology, a printed circuit board maker based in Huizhou, in Guangdong Province. The shares of both companies have fallen by more than half from their records this year.
- While margins at Chinese chip foundries, such as SMIC and Hua Hong, are rising, memory manufacturers like CXMT are facing higher costs on their most advanced products, which eats into their profits.
The performance of these tech companies largely hinges on massive investments in artificial intelligence, whose models require computing hardware and infrastructure to train and operate. Yet, in both China and the United States, AI companies are still struggling to find a profitable business model.