# Mainland Investors Bought Hong Kong Artificial Intelligence Stocks Through August and Sold Financials

The rotation used a market decline to add exposure to model developers such as MiniMax, at a moment when Western investors are questioning returns on artificial intelligence spending.

- Published: 2026-09-03T05:19:30.989Z
- Canonical: https://polylog.news/2026-09-03/mainland-investors-bought-hong-kong-artificial-intelligence
- Publisher: Polylog (Global desk)
- Section: markets
- Sources: [South China Morning Post](https://www.scmp.com/business/china-business/article/3366174/mainland-chinese-investors-buy-hong-kong-tech-stocks-ai-pivot-sell-financials), [Financial Times](https://www.ft.com/content/115c886f-23e4-4a8d-9656-5d7fc9480803?syn-25a6b1a6=1), [The Hindu](https://www.thehindu.com/sci-tech/technology/new-york-mayor-zohran-mamdani-announces-ban-on-ai-use-for-students-until-high-school/article71422209.ece</source)

Investors on the Chinese mainland rotated out of banks, insurers and traditional industrial companies and into Hong Kong listed technology names tied to artificial intelligence during August, [the South China Morning Post reported](https://www.scmp.com/business/china-business/article/3366174/mainland-chinese-investors-buy-hong-kong-tech-stocks-ai-pivot-sell-financials). The model developer MiniMax led the purchases. The buying came during a market decline, which means these flows were adding to positions as prices fell rather than following momentum.

That is close to the opposite of the mood in Western markets, where the debate has moved to whether the capital being invested in artificial intelligence will earn a return. The Financial Times ran a piece this week on [how the industry might repair its public standing](https://www.ft.com/content/115c886f-23e4-4a8d-9656-5d7fc9480803?syn-25a6b1a6=1), a discussion that only becomes necessary when public tolerance is already thin. In New York, Mayor Zohran Mamdani [announced a ban on artificial intelligence use by students below high school age](https://www.thehindu.com/sci-tech/technology/new-york-mayor-zohran-mamdani-announces-ban-on-ai-use-for-students-until-high-school/article71422209.ece), permitting limited pilot programmes only for students aged roughly 14 and above.

The divergence has a structural cause. Chinese investors buying MiniMax are buying a domestic substitution story, in which export controls on advanced chips force a home-grown technology stack that captures a protected market. Western investors buying the same theme are buying an open competition in which returns depend on winning against well-funded rivals. The two groups are not valuing the same asset even when the label matches.

Neither position is safe. The mainland bet depends on Chinese firms delivering competitive models on hardware that remains a generation behind the best available. The Western position depends on enterprise customers converting pilots into recurring spending. What both share is concentration, and concentrated index exposure to a single theme is what makes a correlated drawdown possible when either group changes its mind.

## What this means

Southbound buying gives Hong Kong listed technology stocks a source of buying support from within China that is insensitive to United States sentiment, which partially decouples the two markets and shrinks the effect of foreign selling on Chinese artificial intelligence valuations. Mainland retail and institutional investors carry the risk if domestic model developers cannot close the hardware gap, while Hong Kong listed banks and insurers lose the passive support those flows previously provided. Public policy moves against artificial intelligence in schools and workplaces slow the consumer adoption that some revenue forecasts assume.

## What to watch

- Monthly southbound net purchase totals, which show whether the August rotation into technology continues or reverses.
- Whether Chinese model developers publish benchmark results on domestically produced accelerators, the evidence that the substitution story is working.
- Enterprise software companies' disclosed revenue from artificial intelligence products in the next reporting season, the test of whether spending converts into income.