China offers investors a strange but very clear setup at the moment. The economy appears weak, shoppers are not spending as aggressively, and retail sales in April grew at the slowest pace since the economy reopened after COVID-19.
However, stock trading is not really about malls, restaurants, or hospital names. It’s about artificial intelligence, semiconductors, hard technology, software, cloud capacity, and companies close to Beijing’s push for self-sufficiency.
Investors continue to buy into China’s AI supply chain while the broader economy remains uneven
Modern Alpha Manager’s Liqian Ren believes the technology’s growth story will continue as well. While she said that many companies in the AI ecosystem continue to make good profits, she clearly cautioned that such companies do not have enough scale to transform the entire economy. “It is very uneven,” she stressed.
While many hardware manufacturers trade their A-tier shares on mainland China exchanges, not Hong Kong, this is important given that mainland stocks have outperformed this year.
China’s CSI 300 index, which includes large companies trading in Shanghai and Shenzhen, is up nearly 5% this year, while Hong Kong’s Hang Seng Index is almost unchanged.
Large private companies are not accessible to equity investors. Private companies like ByteDance and Huawei are not publicly listed. However, several Chinese chip producers, AI model developers, and high-tech component makers have recently gone public.
Leonid Mironov’s fund has Tencent Holdings (0700.HK, TCEHY) and Alibaba Group (BABA, 9988.HK) as its largest positions. It also owns hardware names such as Anji Microelectronics (688019.SS) in Shanghai.
Leonid said investors still miss the amount of political support that helped small and medium-sized companies make money. “I think people don’t really see or appreciate how fundamentally beneficial this policy has been to the bottom line of these small and mid-cap names,” he said.
However, he doesn’t buy every typical AI story. Leonid said he is still waiting for Zhipu and MiniMax because he wants clearer proof that customers will stay and that the business model can survive. Morgan Stanley (MS) takes the other side. The bank has an overweight over Zhipu, MiniMax and Alibaba. It also has an overweight rating on Cambricon Technologies (688256.SS) with a price target of 2,000 yuan, or about $294.
DeepSeek cuts prices for V4 Pro and pits China’s AI cost trading against OpenAI and Anthropic
Finally, an aspect that plays an important role in telling the AI story in China is pricing. DeepSeek, a Hangzhou-based startup, has maintained a 75% discount on the V4 Pro for one month after launching the V4 series. The V4 Series includes both the V4 Pro and the lightweight V4 Flash.
In doing so, DeepSeek puts itself in the middle of the global competition on cost. According to Artificial Analysis, a third-party benchmarking company, the V4 Pro ranks #1 globally for intelligence per dollar cost.
In other words, this rating is based not only on intelligence, but also on the amount of output buyers receive from the model. The latter factor is particularly important because powerful computing is limiting, while running large AI models is expensive.
The official API price for DeepSeek’s V4 Pro model ranges from $0.0036 per million cached input tokens and $0.87 per million output tokens.
According to synthetic analysis, the cost of running the IQ indicator on this model is approximately $268. Meanwhile, the cost of doing the same thing on OpenAI’s GPT-5.5 models and Anthropic’s Claude Opus 4.7 models would be about 12 and 19 times higher, respectively.
This is suitable for all software developers, exchange houses, trading houses and developers of artificial intelligence tools. The output cost may be a bit of an issue considering the cost that will be added from the tokens. Third-party testing is important because not all AI companies use the same prices or results for their AI models.
DeepSeek is not the only Chinese name on the list of costs per intelligence unit. On the list are the M2.7 model from MiniMax and the MiMo V2.5 Pro from Xiaomi.
