Southbound Stock Connect's 104 Headline Masks a Sharp Tech Rotation
Southbound Stock Connect ended August 2026 with the 104 figure in focus, but the headline number needs translation. Mainland investors recorded net purchases of HK$10.379 billion, equal to 103.79 hundred-million Hong Kong dollars.
That total was far below July's HK$62.869 billion, according to August fund data. The monthly slowdown came with a sharper internal rotation. Investors bought information technology shares while withdrawing money from healthcare and financial companies.
This was not a broad retreat from Hong Kong stocks. It was a selective repositioning toward technology names, including MiniMax, Alibaba, Tencent, Zhipu, and YOFC. Meanwhile, investors sold heavily into strength across healthcare and banking shares.
The central tension is therefore technology concentration versus market-wide caution. Southbound investors kept buying companies tied to artificial intelligence, cloud computing, digital platforms, and optical infrastructure. Yet their total net buying fell by more than four-fifths from July.
That combination matters more than the 104 headline alone. It suggests mainland capital still wants Hong Kong technology exposure, but investors have become less willing to support the wider market indiscriminately.
What the 104 Figure Actually Measures
The August total shows continued net buying, but it also marks a major loss of momentum from July.
Cailian Press published the monthly review at 6:41 p.m. China Standard Time on August 31, 2026. The report cited Wind data covering the final Hong Kong trading session of August.
Southbound trading refers to purchases of eligible Hong Kong securities by mainland Chinese investors through the Shanghai and Shenzhen Stock Connect channels. The Stock Connect program links the Hong Kong, Shanghai, and Shenzhen markets through shared trading and clearing arrangements.
The widely circulated 104 figure is not HK$104 billion. Chinese market reports commonly express values in units of one hundred million. The precise monthly total was HK$10.379 billion, which the original headline rounded to nearly 104 hundred-million Hong Kong dollars.
That distinction is essential for English-language readers. Treating 104 as billions would overstate the flow by approximately ten times.
August remained positive overall, but the comparison with July changes its meaning. Net buying fell from HK$62.869 billion in July to HK$10.379 billion in August. That represents a decrease of roughly 83.5%.
More August sessions also ended with southbound net selling. Cailian Press connected that caution with the Hang Seng Index struggling below 26,000, although the relationship should not be read as proven causation.
The final session contributed HK$2.139 billion of net buying. Shanghai Connect supplied HK$2.058 billion, while Shenzhen Connect added HK$81 million.
Those figures show how uneven the daily contribution was. One session accounted for more than one-fifth of the month's reported net inflow. That concentration makes the monthly total sensitive to a small number of strong trading days.
Individual sessions were volatile throughout the second half of August. On August 19, southbound investors sold a net HK$10.621 billion, the largest daily outflow since April 9. They sold another HK$10.412 billion on August 20.
The direction then reversed. Net purchases reached HK$11.567 billion on August 24, before shifting back to a HK$6.605 billion outflow on August 25. Smaller inflows followed later in the week.
On August 27, for example, southbound trading produced a net inflow of HK$452.67 million. Alibaba, YOFC, and SMIC led purchases that day, according to the published daily trading breakdown.
This sequence makes the monthly result less straightforward than a single positive number implies. Investors did not steadily accumulate Hong Kong stocks throughout August. They alternated between large withdrawals and concentrated buying.
The 104 result is therefore best read as the residue of a volatile month. It records where net flows ended, not how calmly they arrived there.
Technology Buying Survived the Monthly Slowdown
Southbound investors reduced their overall commitment while directing substantially more capital toward information technology.
Information technology received HK$21.759 billion of net inflows during August. That amount was more than twice the entire market's net southbound purchase for the month.
The apparent contradiction is possible because large inflows into technology were offset by withdrawals elsewhere. Sector totals reveal a rotation that the aggregate 104 figure cannot show.
MiniMax was the largest reported individual beneficiary. Southbound investors bought a net HK$10.052 billion of its shares during the month. The stock gained 51.34% in August.
MiniMax develops generative artificial intelligence models and applications. Its presence at the top of the flow list gave the month a clear AI component, but price appreciation and net buying should not be treated as evidence of business performance.
Alibaba ranked second with HK$7.857 billion of net purchases. Its shares fell 2.39% during August, producing a very different price-and-flow combination from MiniMax.
That divergence is important. Investors bought MiniMax during a strong rally and bought Alibaba during a monthly decline. The same technology allocation therefore included both momentum exposure and apparent accumulation into weakness.
Alibaba's position also carried a separate financing dimension. On August 23, the company launched an HK$80 billion share placement and said the proceeds would support full-stack AI capabilities. The planned uses included chips, infrastructure, model development, and deployment, according to the reported AI financing plan.
The placement increased the number of shares available and created potential dilution for existing investors. At the same time, it offered a direct route for financing Alibaba's expanding AI investment program.
Southbound investors' monthly purchases do not resolve that tradeoff. They show demand for the shares during a period when the company was asking public markets to finance a larger AI commitment.
Tencent received HK$6.716 billion of net buying while its shares declined 4.67% in August. As with Alibaba, the pattern suggests investors accumulated exposure despite weak monthly price performance.
Zhipu attracted HK$2.952 billion. Its shares rose 21.01% during the month, although the final five-day position data indicated a short-term reduction of 400,000 shares.
YOFC, formally Yangtze Optical Fibre and Cable, received HK$3.196 billion of net buying. Its shares gained 72.30% in August, and the reported five-day data showed accelerating inflows near month-end.
YOFC broadens the technology story beyond software models. Optical fiber and cable suppliers provide physical connectivity for telecom networks and data infrastructure. Investors were buying both AI developers and parts of the infrastructure chain surrounding digital demand.
The mix does not establish one unified AI investment strategy. MiniMax and Zhipu are model developers, Alibaba and Tencent operate diversified internet platforms, and YOFC supplies communications infrastructure.
What connects them is access. Hong Kong listings let mainland investors assemble a technology portfolio that differs from the selection available in domestic exchanges.
Stock Connect has expanded that menu over time. Eligible securities include qualifying large-cap, mid-cap, and selected small-cap constituents, along with eligible exchange-traded funds.
ETF access has also become more important. Average daily southbound ETF turnover reached HK$5.8 billion during the first seven months of 2026. That was 48.7% above the 2025 full-year average, according to HKEX ETF data.
This wider access helps explain why southbound flows can move between individual companies and diversified technology products. It also means a technology inflow does not necessarily represent conviction in every underlying company.
The more defensible conclusion is narrower. Mainland investors preserved a strong appetite for Hong Kong technology exposure even as their overall net buying slowed sharply.
Healthcare and Banks Became the Other Side of the Trade
August's technology inflow was financed partly by substantial selling in healthcare and financial stocks.
Healthcare recorded HK$15.420 billion of net outflows during August. Financial companies lost another HK$12.978 billion.
Together, those withdrawals exceeded the information technology sector's reported inflow. They explain how technology could attract HK$21.759 billion while the full market added only HK$10.379 billion.
The direction was also counterintuitive because healthcare and financial shares showed periods of strong market performance. Southbound investors sold into that strength instead of treating higher prices as a reason to add exposure.
This creates the article's primary opponent: concentrated technology buying versus profit-taking or de-risking in healthcare and banking.
The final trading day illustrated the financial side. China Construction Bank recorded HK$1.030 billion of net selling, while Industrial and Commercial Bank of China lost HK$579 million.
Both bank shares rose that day. China Construction Bank gained 3.15%, and ICBC rose 1.68%. The reported five-day figures also showed continued reductions in their share counts.
That combination does not reveal why every investor sold. It does show that net selling occurred alongside rising prices, which is consistent with investors using strength to reduce exposure.
Healthcare displayed a similar pattern earlier in the month. On August 20, CSPC Pharmaceutical gained 9.61%, while southbound investors sold a net HK$1.092 billion of the stock.
Innovent Biologics and Akeso also experienced notable selling on different August sessions. These moves came despite continued investor interest in China's biotechnology sector.
A sector outflow should not be interpreted as a verdict on medical innovation. Flow data captures executed purchases and sales, not the complete reasoning behind them.
Investors may rebalance because of valuation, portfolio concentration, liquidity needs, short-term performance, or expectations about future catalysts. Public flow totals cannot separate those motives.
Financial stocks carry another complication. Banks often attract investors seeking dividends and lower valuation multiples, but they also serve as liquid sources of funds when traders want to rotate quickly.
Selling a bank to buy a technology company does not necessarily mean investors expect the bank's earnings to deteriorate. It can mean the perceived upside elsewhere has become more urgent.
The August pattern therefore reflects relative preference. Technology received incremental capital, while healthcare and banks supplied some of the liquidity for that choice.
That relative framing is more useful than calling one side a winner and the other a loser. Monthly flows can reverse quickly, especially when price moves change valuations.
The data also contains exceptions. Not every technology stock attracted money, and not every financial or healthcare stock was sold throughout the month.
Hua Hong Semiconductor lost HK$3.047 billion of southbound funds in August. Its shares declined 4.13%, and investors reduced their reported holdings by 14.18 million shares during the final five-day period.
Meituan recorded HK$2.018 billion of net selling and fell 14.70% for the month. However, its recent position data showed an increase of 11.72 million shares, indicating short-term buying after the broader monthly withdrawal.
These exceptions prevent a simplistic "buy all technology" reading. Investors distinguished between AI model companies, internet platforms, semiconductor producers, and local-services businesses.
The real rotation was selective. Capital favored particular AI, platform, and optical-network names while selling other technology shares and reducing broader exposure to healthcare and banks.
What the Numbers Do Not Prove
Southbound flows are evidence of trading behavior, not a complete measure of conviction, valuation, or future returns.
The first limitation is timing. Reported transaction flows describe purchases and sales executed during a defined period. They do not explain how long investors plan to hold those positions.
A large monthly purchase can represent long-term allocation, short-term momentum trading, ETF creation activity, or rebalancing around an index change. The published totals do not identify the strategy behind each order.
The second limitation concerns holdings data. Hong Kong cash-market trades generally settle on T+2, meaning settlement occurs two trading days after execution. The exchange describes the current settlement framework for exchange trades and clearing transactions.
Cailian Press accordingly noted that its five-day holding observations reflected positions available with a two-trading-day delay. Daily trade flow and reported shareholding changes can therefore refer to different effective dates.
That distinction helps explain apparent inconsistencies. A stock can show monthly net buying while its latest reported five-day holdings decline. The windows are different, and settlement timing creates an additional lag.
The third limitation is aggregation. "Southbound investors" are not one institution making a single portfolio decision. The label covers eligible mainland investors trading through both the Shanghai and Shenzhen channels.
One group can buy a stock while another sells it. Published net flow is the difference between those actions, not a unanimous market opinion.
The fourth limitation is the relationship between prices and flows. MiniMax rose 51.34% while attracting the month's largest individual net purchase. YOFC gained 72.30% alongside substantial buying.
Those pairings show that money followed strong price performance. They do not prove that the purchases caused the entire rally, nor do they establish that the valuations remained attractive after those gains.
Alibaba and Tencent present the opposite pattern. Both declined during August despite receiving billions of Hong Kong dollars in reported net buying.
That pattern could indicate bargain hunting, but the data alone cannot verify the motive. It could also reflect supply from other investor groups overwhelming southbound demand.
The fifth limitation is sector classification. Alibaba and Tencent generate revenue from several businesses, while YOFC's exposure differs significantly from an AI model developer.
Grouping all three under technology is useful for measuring broad allocation. It is less useful for judging their operating risks, capital needs, or competitive positions.
The 104 figure also provides no direct information about earnings quality. It cannot show whether AI spending will produce acceptable returns or whether product adoption will match investor expectations.
Alibaba's announced financing underscores this uncertainty. Management described a broad AI investment plan, but the commercial payoff depends on execution across chips, infrastructure, models, cloud services, and customer demand.
MiniMax and Zhipu carry different uncertainties. Public-market enthusiasm can increase expectations faster than revenue, margins, or user retention improve. The flow report did not provide those operating metrics.
The article's healthcare and bank outflows require similar restraint. Selling during a strong month can represent profit-taking rather than a deteriorating fundamental outlook.
Investors should therefore avoid turning a monthly flow chart into a complete investment thesis. The chart identifies where capital moved, which companies drew attention, and which sectors funded the rotation.
It does not answer whether those decisions will be profitable. It also does not substitute for company filings, financial statements, valuation work, or risk analysis.
Three Signals to Watch After August
September will test whether the technology rotation has durable support or merely captured a volatile month-end snapshot.
The first signal is whether southbound technology buying continues after the August slowdown. Information technology attracted HK$21.759 billion even though total net buying fell to HK$10.379 billion.
Continued inflows into MiniMax, Alibaba, Tencent, Zhipu, and infrastructure suppliers would strengthen the concentration thesis. A broad reversal across those names would weaken it.
Daily totals should be read alongside weekly and monthly figures. August showed that one strong session can offset several large outflow days, producing a positive final number without a stable trend.
September began with a net southbound purchase of HK$7.756 billion. Tencent and MiniMax led the reported buying, while Meituan recorded net selling.
That opening supports continuity, but one session cannot establish a new monthly pattern. The important test is whether buying persists through weaker market days rather than appearing only during rebounds.
The second signal is how technology companies convert capital access into operating evidence. Alibaba's proposed financing puts particular attention on AI infrastructure spending and its eventual contribution to cloud demand.
Investors should watch reported capital expenditure, cloud revenue growth, AI product adoption, and margins. Those measures can show whether higher spending supports a scalable business or mainly increases near-term costs.
MiniMax and Zhipu face a related test. Share-price performance and southbound demand have increased their visibility, but future financial disclosures must carry more analytical weight than fund-flow rankings.
For YOFC, the relevant evidence differs. Investors should watch orders, revenue, and capacity utilization connected with optical networks and data infrastructure, rather than treating every communications project as direct AI demand.
These company-level checks will reveal whether August's buyers selected businesses with improving fundamentals. Weak operating evidence would reduce the significance of even persistent fund inflows.
The third signal is whether money returns to healthcare and banks. Their combined August outflow was large enough to finance much of the technology allocation.
A renewed inflow into those sectors would suggest investors are broadening their exposure rather than abandoning the technology trade. Continued withdrawals would reinforce a narrower, more concentrated market structure.
Bank and healthcare price behavior also matters. If shares keep rising despite southbound selling, other investor groups are absorbing the supply. If prices weaken alongside continued outflows, the rotation is gaining broader market force.
This is why the 104 headline should remain a starting point, not the conclusion. August ended with positive southbound demand, but that demand was much weaker than July and highly concentrated by sector.
Readers tracking Hong Kong's technology market should keep three records together: daily Stock Connect flows, company operating disclosures, and sector-level price performance. Each answers a different question.
Flow data shows where mainland capital moved. Financial disclosures show whether the favored companies delivered. Prices reveal whether the wider market agreed with those choices.
The next meaningful development will not be another isolated daily ranking. It will be evidence that concentrated technology buying survived volatility, spread into fundamental results, or lost support as investors returned to other sectors.
For anyone following Chinese AI and digital infrastructure companies, the practical question is straightforward: does the next month confirm August's rotation, or expose the 104 figure as a temporary balance between a few crowded purchases and widespread selling?
