Withdraw from the seven giants in the US stock market and bet on Chinese AI! Is Wall Street betting on a rise in the east and a fall in the west?

Withdraw from the seven giants in the US stock market and bet on Chinese AI! Is Wall Street betting on a rise in the east and a fall in the west?

As the global artificial intelligence industry reaches a crossroads, the AI development patterns of the United States and China are undergoing subtle and profound structural reversals.

Recently, Wall Street investment research institutions have made heavy statements, the price difference between China and the United States is as high as two orders of magnitude, and the product delays of Silicon Valley technology giants are all pointing to one fact: The "computing stick" violent aesthetics that large American models have long relied on is facing a severe test of the bottom line of economics.

China's AI ecosystem, which focuses on localization of supply chains and ultimate cost-effectiveness, seems to be becoming a new oasis for global capital to re-anchor.

Quantum Strategy’s rare slogan: withdraw from AI, except China On Monday local time, Quantum Strategy, a well-known Wall Street research institution, dropped a bombshell in its latest report to clients. The agency made it clear that after conducting a new study on the AI industry, it decided to adjust its asset allocation—withdrawing funds from U.S.

technology leaders and turning strongly optimistic about Chinese stocks related to AI applications. Quantum Strategies said it is becoming more bullish on Chinese stocks, especially in the field of artificial intelligence. Currently, it is moving funds out of leading U.S.

technology companies while becoming more active on Chinese stocks related to AI applications.

In its latest portfolio recommendations, Quantum Strategy stated that its current strategy is to "withdraw from AI (excluding China) and the seven US technology giants", while recommending that investors do more in industries that benefit from AI deployment and "pay special attention to China." Changes in related trends may have begun to take shape in the U.S.

stock market recently. In the past June alone, as investors began to examine the huge AI capital expenditure bill, the market value of the "Big Seven" in the US stock market evaporated by approximately US$2.3 trillion in just one month, and the Big Seven index fell 10% that month.

At the same time, according to an industry report, hedge funds began to reduce their exposure to the Big Seven U.S. stocks in May and buy Chinese companies listed in the United States. The most popular U.S. stock ADRs held by hedge funds include Alibaba, Pinduoduo and Baidu.

The shift underscores the growing appeal of Chinese technology stocks, which have become attractive to foreign investors as signs of China's growing influence in developing new technologies emerge. One of the main reasons that has caused many Wall Street to waver is undoubtedly the "economic account" behind the Sino-US model that cannot be ignored.

On the AI commercialization path that is a matter of survival, Chinese models are implementing a "dimensionality reduction attack" on their Western counterparts with a price advantage of one to two orders of magnitude. For example, OpenAI’s GPT-5.5 costs approximately $5 per million input tokens and $30 per million output tokens.

The cost of DeepSeek's open source weighted V4-Pro model to run the same tasks is only about US$0.44 and US$0.87 - and its more lightweight Flash version is as low as about US$0.14 and US$0.28, which is about 30 times cheaper than the former in terms of output, and more than 100 times cheaper in the cheapest version.

Another Chinese open source model, Zhipu AI’s GLM-5.2, is priced at US$1.40/US$4.40 – still several times cheaper than the flagship model in the United States. And the output quality of these Chinese models is not bad. In the standard SWE-bench programming test, several of the models were within 1 point of the top Western models.

In a practical test that included 18 real programming tasks, GLM-5.2 cost about $2.74 to complete the task, while GPT-5.5 required $16.10, which is only about one-sixth of the latter.

Local closed loop from software to hardware: Macquarie outlines new territory for China’s AI chips The United States has previously tried to choke China's AI by restricting Nvidia's high-end GPU exports to China, but this move has instead accelerated the formation of China's local AI chip supply chain.

Macquarie also pointed out in a recent report that China has been committed to building its own semiconductor industry for many years, and now is the time to buy.

Macquarie analysts said in the report: "Considering the development of AI, China's domestic large language model players, and the token economy, we believe that the best time to invest in Chinese AI chip manufacturers has arrived." The agency also mentioned, "The Chinese government's support for domestic AI chip companies has enhanced the growth visibility of leading domestic companies." Currently, although the U.S.

government has begun to allow the sale of some less advanced Nvidia chips to China, China's willingness to buy these chips is waning. Macquarie rated five relevant Chinese AI chip companies for the first time in the report. One of its favorite targets is Cambrian - giving it an Outperform rating.

Macquarie analysts said: "We believe Cambrian has shifted its core customer base from government smart computing clusters to leading domestic cloud service providers and large language model developers.

These customers provide a more balanced sales mix, healthy margins and solid cash flow." Macquarie's target price for Cambrian is $2,060, which is more than 50% higher than Monday's closing price.

Among the AI chip stocks listed in Hong Kong, China, Macquarie is more optimistic about Biren Technology and gave a target price of HK$140, which is more than double the closing price on Monday.

Analysts at the agency said: "We are optimistic about Biren Technology's general graphics processing unit (GPGPU) product portfolio that favors high computing power, chip interconnection and large-scale computing clusters.

In addition, the company's focus on the domestic supply chain may soon bear fruit, thus assisting the launch of its new products." Other favored companies include Hong Kong-listed Tianshu Zhixin and A-share listed Muxi Shares. Macquarie's relatively unfavorable stock among this group is Haiguang Information, due to concerns about its loss of market share.

Macquarie analysts said: "Haiguang Information occupies a favorable position in China's CPU and AI chip fields, but we attribute a large part of its success to AMD's technology transfer, and we believe that the development of AI intelligence brings limited upside to the sector." They rated Haiguang as "underperform the market." In any case, in the second

half of this Sino-US AI competition, the competition seems to be quietly shifting from "blind stacking of parameter scales" to "extreme squeezing of cost per unit of computing power." The American AI giants are being dragged into the quagmire by their high capital expenditures and inefficient token consumption; while Chinese AI is showing amazing

capabilities at the global commercial application level by virtue of open source dividends, cost-effective advantages without competitive pressure, and the closed-loop local chip supply chain. As some industry analysts have said, this battle of "efficiency duel" may be the beginning of the change of dominance in the global AI industry. (

#Stocks #Nvidia #AMD #AI #Semiconductors