Southbound Money Chased Nine Hong Kong Stocks, but Sold Tencent and Alibaba
Mainland investors increased their holdings in nine Hong Kong stocks by more than 15% during the week ending July 24. Yet they simultaneously sold Tencent and Alibaba heavily.
That split is the important part of the story. Southbound capital was not making a simple bet on a rising Hong Kong technology market. It was rotating from the largest internet platforms toward semiconductors, newer AI listings, and selected smaller companies.
The rotation occurred as Hong Kong’s major indexes advanced. The Hang Seng Index gained 1.63% from July 20 through July 24. The Hang Seng China Enterprises Index rose 1.65%, while the Hang Seng Tech Index added only 0.14%.
Southbound Stock Connect recorded a combined net inflow of HK$2.978 billion during the week, according to Data Bao figures republished in a market flow analysis. That modest market-wide inflow concealed much larger moves between individual companies.
Hua Hong Semiconductor received HK$3.318 billion in net buying. NetEase, Meituan, and Xiaomi also attracted capital. Tencent and Alibaba moved in the opposite direction, with reported net sales of HK$6.408 billion and HK$5.186 billion.
This was not an indiscriminate technology rally. It was a week when mainland investors appeared willing to fund particular AI and semiconductor narratives while reducing exposure to established internet leaders.
Southbound investors made large moves beneath a modest headline inflow
The week’s HK$2.978 billion net inflow understates how aggressively investors rotated between individual technology stocks.
Stock Connect allows eligible mainland investors to trade designated Hong Kong securities through the Shanghai and Shenzhen exchanges. Southbound trading refers to money moving from mainland China into Hong Kong-listed securities.
The program has become a major source of liquidity for the Hong Kong market. Its influence now extends well beyond traditional state-owned companies and financial stocks.
Tencent generated the highest combined Southbound buying and selling turnover among the week’s active names. Its total reached HK$43.64 billion, despite the company ending the week with HK$6.408 billion in net selling.
Zhipu followed with HK$39.754 billion in combined turnover. Semiconductor Manufacturing International Corporation and Alibaba each exceeded HK$30 billion.
These figures describe activity, not conviction. A stock can generate enormous turnover while investors reduce their aggregate position. Tencent provided the clearest example.
Hua Hong Semiconductor showed the opposite pattern. The chip foundry led reported net purchases with HK$3.318 billion. Its shares gained 8% during the week, placing it among the stronger technology performers.
NetEase received HK$1.409 billion in net buying even though its shares fell 7.45%. That combination suggests some investors used the decline to build exposure. It does not prove that the stock had reached a bottom.
Meituan attracted HK$1.031 billion, while Xiaomi recorded HK$126 million in net purchases. Those figures show that mainland investors did not abandon internet and consumer technology companies as a group.
Instead, they treated each platform differently. NetEase and Meituan found buyers, Xiaomi attracted a smaller inflow, and Tencent and Alibaba experienced substantial selling.
The nine stocks with holdings growth above 15% formed another layer of this rotation. Dajin Heavy Industry led with a 128.14% weekly increase in Southbound shareholdings. Xunce, Befar Group, and NetEase followed with increases of 32.21%, 31.6%, and 27.21%.
A percentage increase can look dramatic when the starting position is small. Dajin’s 128.14% increase therefore requires more context than Tencent’s multibillion-dollar net sale.
The result is a market carrying two messages at once. Southbound investors remained net buyers overall, but their strongest decisions concerned where not to keep capital.
Chips and new AI listings are challenging the old internet trade
Mainland capital increasingly treats Hong Kong as a market for semiconductor and AI exposure, not merely a home for established internet platforms.
Hong Kong’s listed technology sector has broadened rapidly. It now includes internet services, electric vehicles, chipmakers, biotechnology companies, robotics developers, and recently listed AI model companies.
HKEX said Southbound average daily turnover reached HK$123.1 billion during the first half of 2026. That represented a 10.9% increase from the comparable period.
The exchange also reported that Hong Kong cash-market average daily turnover reached HK$283 billion. Technology and AI-related stocks helped support the activity, alongside new listings and elevated Stock Connect participation.
This broader context matters because the July rotation did not occur in a quiet or shrinking channel. It happened after Southbound trading had already reached a historically high level.
During 2025, Southbound average daily turnover more than doubled from HK$48.2 billion to HK$121.1 billion. It accounted for 23% of Hong Kong cash-equity turnover by the fourth quarter.
The Stock Connect review also counted 564 eligible Hong Kong stocks and 23 eligible exchange-traded funds at the end of 2025. This expanded menu gives mainland investors more ways to express specific industry views.
New AI listings have become particularly important. Zhipu’s HK$39.754 billion in weekly Southbound turnover placed it just behind Tencent, a company with a much longer public-market history.
That comparison does not mean Zhipu has matched Tencent’s scale, earnings, or business maturity. It shows that a newer AI company can already compete for investor attention and trading liquidity.
HKEX has designed benchmarks around the same shift. Its Tech 100 Index tracks eligible companies spanning AI, information technology, internet services, electric vehicles, biotechnology, and robotics.
The exchange reported that Southbound turnover averaged HK$121.5 billion through May. This represented about 20% of Hong Kong cash-market turnover. Zhipu and other recent technology listings entered the index soon after becoming eligible.
The technology index framework reflects an important change in market structure. Mainland investors no longer need to use Tencent or Alibaba as broad substitutes for Chinese technology exposure.
They can buy a foundry when they favor domestic chip demand. They can trade an AI model developer when interest shifts toward generative AI. They can select software, hardware, or platform companies separately.
That choice creates pressure for the largest internet groups. Their liquidity remains unmatched, but their shares must now compete with more focused technology narratives.
A portfolio manager seeking AI exposure can compare Tencent’s diversified business with Zhipu’s concentrated AI identity. An investor expecting stronger chip demand can choose Hua Hong Semiconductor or SMIC directly.
This fragmentation helps explain how Tencent could lead weekly turnover and still record the largest net sale. High liquidity made it easy to use the stock as a source of funds for other trades.
Hua Hong became the clearest semiconductor winner
Hua Hong’s HK$3.318 billion net inflow suggests that investors favored direct chip exposure over a general technology basket.
Hua Hong operates semiconductor foundries, which manufacture chips designed by other companies. Foundry demand connects the company to automotive electronics, industrial devices, consumer products, and computing infrastructure.
The stock gained 8% during the July 20 to July 24 period. That rise accompanied the week’s largest reported Southbound net purchase.
SMIC also generated more than HK$30 billion in combined Southbound turnover. Together, the two foundries gave mainland investors liquid ways to position around domestic semiconductor manufacturing.
The foundry trade rests partly on expectations for AI-related computing demand. Data Bao cited Counterpoint Research figures indicating that the broader Foundry 2.0 market grew 23% year over year during the first quarter.
That estimate covered more than traditional wafer fabrication. Foundry 2.0 includes related manufacturing services such as advanced packaging and testing, which makes the category broader than a conventional foundry revenue measure.
Investors should therefore avoid treating the cited market growth as Hua Hong’s company-specific growth. It describes an industry category, not the foundry’s audited results.
Still, the mechanism behind the trade is understandable. AI accelerators and custom AI chips require manufacturing capacity, packaging, memory interfaces, and supporting components.
Demand does not benefit every supplier equally. Process technology, customer mix, utilization, production yields, and capital expenditure determine whether industry growth reaches a company’s earnings.
The week’s performance also showed differences inside the chip supply chain. Montage Technology gained 30.43%, leading the active-stock group. The company had forecast first-half net income attributable to shareholders between RMB1.9 billion and RMB2.1 billion.
Montage attributed its growth partly to increased shipments of DDR5 register clock driver chips. These components help manage signals between processors and memory modules in servers.
The company also cited rising revenue from interconnect products, including PCIe retimers. A retimer rebuilds degraded high-speed signals, helping data move reliably across server hardware.
According to the company’s forecast, first-half interconnect revenue reached approximately RMB3.111 billion, an increase of about 26.4%. Second-quarter revenue reportedly rose 19.5% from the previous quarter.
Those are company projections rather than completed audited results. They still provided investors with a concrete operating narrative that the broader AI infrastructure theme often lacks.
By comparison, printed-circuit-board-related shares weakened. Kingboard Laminates fell 10.15%, while Kingboard Holdings declined 9.26%.
That divergence is a warning against describing the week as a universal AI hardware rally. Investors rewarded some businesses associated with chips and server connectivity while reducing exposure elsewhere in the electronics chain.
Hua Hong’s inflow matters because it was large enough to exceed the market’s total net Southbound buying. Other stocks collectively offset a substantial portion of that purchase.
This is the central reversal in the week’s data. Mainland capital remained a net supporter of Hong Kong equities, yet the aggregate flow existed alongside sharp internal selling.
Tencent and Alibaba became funding sources for the rotation
The largest internet platforms remained highly liquid, but that liquidity made them natural sources of capital for newer technology trades.
Tencent’s HK$6.408 billion net sale was almost twice the market’s total Southbound net inflow. Alibaba’s HK$5.186 billion net sale added another major outflow from established platform companies.
These figures do not establish a lasting bearish view. Weekly flows can reflect portfolio rebalancing, profit-taking, derivatives activity, index adjustments, or short-term reactions to news.
They do establish that investors were willing to reduce two widely held technology leaders while buying other names. That is more informative than the market’s positive index performance alone.
Tencent’s HK$43.64 billion in combined trading illustrates its role as a liquidity anchor. Large investors can adjust exposure without searching for a willing market in a thinly traded security.
Alibaba performed a similar function. It remained among the small group of companies with more than HK$30 billion in weekly Southbound turnover.
The two companies also carry complex operating exposures. Tencent combines gaming, advertising, payments, cloud services, social media, and investment holdings. Alibaba spans commerce, cloud computing, logistics, and local services.
Diversification can reduce dependence on a single business. It can also make the stocks less precise tools for investors pursuing one theme, such as semiconductor manufacturing or foundation models.
Zhipu presents the opposite proposition. Its appeal is more tightly linked to enthusiasm for generative AI and Chinese foundation-model development.
That concentration offers a clearer narrative, but it also creates higher execution risk. A younger public company has less market history and fewer mature business lines to absorb setbacks.
The contrast is therefore not old technology versus new technology. It is diversified cash-generating platforms versus more concentrated exposure to current investment themes.
NetEase complicates any claim that investors rejected internet companies. Its Southbound holdings increased 27.21%, and it received HK$1.409 billion in net purchases.
The stock fell 7.45% during the same week. Buyers were increasing exposure into weakness rather than following upward momentum.
Meituan also attracted more than HK$1 billion in net buying. Xiaomi received a smaller positive flow.
These differences suggest that investors evaluated valuation, earnings expectations, and company-specific catalysts. Sector labels alone do not explain the results.
Tencent and Alibaba could regain buying quickly if earnings, cloud growth, AI monetization, or shareholder returns exceed expectations. Weekly selling does not erase their strategic positions.
However, the expanding list of eligible technology stocks changes their competitive environment in the capital market. Investors have more alternatives whenever confidence in a platform weakens.
That change raises the standard for internet giants. Being large, liquid, and familiar no longer guarantees that Southbound inflows will treat them as default technology holdings.
A 15% holdings increase can exaggerate conviction
The reported percentage gains are useful signals, but they can overstate the importance of changes that began from small positions.
Dajin Heavy Industry recorded a 128.14% increase in Southbound holdings, the largest percentage move in the group. The number is striking, but it does not disclose the starting position in the headline.
A position can more than double while remaining small in absolute value. Conversely, a minor percentage change in Tencent can represent billions of Hong Kong dollars.
That denominator problem applies to all lists based on weekly percentage growth. Investors should compare the share increase, market value, turnover, and percentage of outstanding shares.
The shareholding database provides daily Southbound positions held through the Central Clearing and Settlement System. It also shows those holdings as a percentage of issued shares.
HKEX cautions that the issued-share figure used in its percentage calculation might not immediately reflect every corporate action. The database is an important primary record, but it is not a complete explanation of investor intent.
Shareholding data also cannot identify why each investor traded. Stock Connect combines decisions from institutions and individuals using both the Shanghai and Shenzhen channels.
The data does not reveal a single unified “mainland investor” strategy. It aggregates many portfolios with different horizons, mandates, and risk limits.
Company-specific announcements can also create temporary jumps. Dajin disclosed that a subsidiary had signed contracts with a Greek shipowner covering three confirmed bulk carriers and one option vessel.
The company placed the combined value of all four vessels at approximately RMB2.1 billion. The three confirmed vessels represented about RMB1.575 billion, while the option vessel represented roughly RMB525 million.
Dajin said the contracts could support future operating performance. That remains a company expectation until deliveries, revenue recognition, costs, and customer payments become visible.
Xunce, whose Southbound holdings rose 32.21%, gained 33.92% in the market during the week. The company also announced a strategic cooperation memorandum involving tokenization services.
A memorandum signals planned cooperation, not guaranteed commercial revenue. Investors still need details about contracts, customers, implementation schedules, and recognized sales.
Befar Group’s holdings rose 31.6%. NetEase followed at 27.21%, but NetEase’s larger scale makes the absolute significance of its move potentially different.
Ownership concentration adds another dimension. Among the stocks with holdings growth above 15%, Jingwei TianDi reportedly had Southbound holdings equal to 58.07% of its Hong Kong shares.
PegBio and Befar Group followed at 10.06% and 9.64%. A high ownership share can show sustained mainland interest, but it can also increase sensitivity to future Southbound selling.
Liquidity deserves equal attention. A concentrated position in a smaller stock can be harder to unwind than a similarly valued position in Tencent.
None of these limitations invalidates the weekly data. They change the question investors should ask.
The useful question is not simply which holdings rose fastest. It is whether buying was large enough, durable enough, and supported by operating results.
The wider market makes selective buying easier to sustain
Record Stock Connect activity gives sector rotations more weight, but it does not guarantee that one week’s winners will keep attracting capital.
Hong Kong’s cash market entered the second half of 2026 with unusually high participation. Average daily turnover reached HK$289.5 billion during the second quarter, a quarterly record.
The first-half average of HK$283 billion was 17.8% above the comparable 2025 figure. It was also more than twice the full-year average recorded in 2024, according to the first-half update.
Southbound average daily turnover reached HK$123.1 billion in the first half. That slightly exceeded the HK$121.1 billion record established during 2025.
This matters because active capital can rotate across more companies without leaving the market. Selling Tencent does not necessarily mean withdrawing money from Hong Kong.
An investor can sell a diversified platform and purchase a foundry, AI developer, server-component supplier, or technology-focused exchange-traded fund within the same channel.
The expansion of eligible securities reinforces that behavior. HKEX maintains separate lists for Shanghai and Shenzhen Southbound trading, and it updated those lists on July 20.
Eligibility determines access, but it does not provide an endorsement. A stock’s inclusion means qualifying investors can trade it through the program, not that the exchange expects it to perform well.
The program’s scale also creates feedback effects. More Southbound participation can improve liquidity and visibility, which can attract additional trading.
The reverse can happen when positions unwind. Companies with high Southbound ownership can face sharper pressure if sentiment changes across mainland portfolios.
The July 20 to July 24 index results demonstrate this balance. The Hang Seng Index and Hang Seng China Enterprises Index rose more than 1.6%, but the technology index barely advanced.
Underneath that small technology gain, Montage and Zhipu rallied while NetEase and printed-circuit-board companies declined. Tencent and Alibaba experienced heavy net selling despite high turnover.
The broad index therefore hid a more decisive argument about technology leadership. Investors were separating companies with direct AI infrastructure or fresh catalysts from businesses carrying more mature expectations.
Three signals will show whether that judgment lasts.
First, watch subsequent Southbound shareholding records. Continued increases in Hua Hong, Zhipu, and smaller technology names would support the rotation thesis. A rapid reversal would identify it as a short-lived trade.
Second, track earnings and management forecasts. Semiconductor suppliers need higher utilization, revenue, and margins to justify enthusiasm linked to AI demand. Internet platforms need measurable returns from cloud and AI spending.
Third, compare net buying with price performance. Rising holdings during falling prices can indicate accumulation, but persistent declines can also signal that buyers underestimated fundamental risk.
The eligible securities list will matter as new technology companies enter the channel. Each addition creates another competitor for mainland capital.
Investors should resist turning the nine-stock list into a recommendation. It is a snapshot of positioning during one active week, not a forecast.
The stronger conclusion concerns market structure. Southbound capital now has enough scale and choice to distinguish between several versions of Chinese technology exposure.
That makes future flow data more revealing, but also more difficult to summarize. A positive total can coexist with sharp selling in the largest companies.
The next meaningful signal will not be another broad statement that mainland money entered Hong Kong. It will be whether chip and AI buying survives earnings scrutiny while Tencent and Alibaba regain demand.
Readers following this rotation should compare weekly flows with daily holdings, company disclosures, and actual operating results. Which companies keep their new shareholders after the immediate catalyst fades?



