Alibaba Prices $10.2 Billion Hong Kong Share Sale at 8.4% Discount to Fund AI Spending; Hong Kong Shares Fall 8.5%

Alibaba Just Sold $10.2 Billion in New Stock. Its Own Investors Are Split on Whether That's Smart.

On Sunday, August 23, 2026, Alibaba priced 710 million new shares in Hong Kong at HK$112.70 each — 8.4% below Friday's closing price of HK$123^[2]. The sale raised about HK$80 billion, or US$10.2 billion, and in a filing to the U.S. Securities and Exchange Commission the company said every dollar of net proceeds would go toward what it calls its "full stack" AI push: chips, data centers, and the models that run on top of them^[1]. The deal is due to close August 26^[1].

The next morning, Alibaba's Hong Kong-listed shares fell as much as 10% and closed down about 8.5%^[3][4]. Its U.S.-listed shares, which weren't part of the sale, fell a smaller 3.4% in premarket trading^[3][4]. Both moves are real, and both numbers are correct. They're also being read as opposite stories, depending on who's telling it.

That split isn't really about the arithmetic. Reported demand for the new shares hit about US$28 billion, nearly three times what Alibaba sold, including roughly US$6 billion from long-only funds and sovereign wealth investors who tend to hold for years, not days^[3]. So the same event produced both a sharp one-day stock drop and a deeply oversubscribed order book. Figuring out which of those facts matters more is the actual argument.

Why a Company Sells Stock It Doesn't Have to Repay

A share placement works differently from a company selling something it already owns. Alibaba created brand-new shares out of nothing and sold them directly to big investors, all in about a day^[1]. To make that happen fast, the price has to be set below where the stock is already trading — that's the discount, and it's less a judgment on the company than the toll for speed and size^[2]. Existing shareholders now own a smaller slice of Alibaba than they did before: the new shares equal about 3.8% of the company's share count last quarter, working out to roughly 3.7% dilution^[5].

Why raise money this way instead of borrowing it? Alibaba's own numbers make the case. In the June quarter, free cash flow — money left over after running the business and paying for new equipment — was negative RMB44.7 billion, about US$6.58 billion, while capital spending was up 75% year over year^[3][6]. Debt comes with fixed payments due on a schedule, whether or not the AI bet pays off on time. Equity doesn't. Alibaba is choosing not to lock itself into payments it might not be able to make if the buildout runs long.

There's a second force behind the "full stack" language, and it isn't just corporate branding. U.S. export controls limit China's access to the most advanced foreign AI chips^[7]. A U.S. company raising the same $10 billion would mostly be handing it to Nvidia for chips. Alibaba is partly trying to build the chips itself. That's a more expensive, more uncertain bet than just buying compute off the shelf — and it's a bet the company says the money is funding^[1][7].

The Same Earnings Report, Read Two Ways

Days before the sale, on August 18, Alibaba reported June-quarter net income of RMB10,444 million (US$1,539 million) — down 75% from a year earlier — with its operating margin cut from 14% to 6%^[6]. In the same report, revenue from AI Cloud and Compute Services rose 45% to RMB48,437 million (US$7,139 million), and AI-related product revenue has now posted triple-digit growth for twelve straight quarters^[6].

Skeptics point to the first set of numbers. Investor Michael Burry, known for his bet against the U.S. housing market before 2008, said publicly he "cannot bless" the share issuance^[3]. Analyst Nicholas Mugalli of World Trade Securities called the raise the "ultimate white flag" and said the market reaction was "pure relief for mainland shortsellers"^[4]. Their argument is that a business generating real cash wouldn't need to sell stock at a discount to fund its own growth — and that negative free cash flow, alongside a 75% profit drop, is the tell.

Alibaba and the funds that bought into the placement point to the second set of numbers. Their case is that cloud computing is a market where customers pick a platform and stay, so whoever has the chips and data centers ready when demand shows up keeps that revenue for years^[6]. Winston Ma, an adjunct professor at NYU School of Law and a former head of North America for China's sovereign wealth fund, China Investment Corp, said the placement "proves that American and Chinese tech giants are operating off the exact same strategic playbook"^[5]. On this view, raising money now, while the order book is strong, beats raising it later in a downturn^[3].

Two Stock Prices, One Deal

Part of why this story reads so differently depending on the outlet comes down to a technical detail: Alibaba trades in two places, as ordinary shares in Hong Kong and as American depositary shares (ADRs) in New York. The new stock was priced off the Hong Kong market and sold only to investors outside the United States^[5]. That means the Hong Kong shares absorbed the full 8.5% drop, while the ADRs fell about 3.4%^[3][4] — a smaller move because the ADRs weren't diluted the same way.

That split also produced two different, equally correct discount figures circulating in the press. Measured against Friday's Hong Kong close of HK$123, the discount was 8.4%^[2]. Measured against the prior U.S. close of the ADRs, it was 3.6%^[5]. Neither number is wrong. They're just measuring the same sale against two different starting points, and a story that quotes one without saying so can make the deal look bigger or smaller than the other version implies.

That's borne out in how outlets actually covered it. Reuters paired the price drop with the word "sharp" a paragraph before the actual 8.4% figure, and switched between the two reference prices across its own stories without flagging the change^[2]. Bloomberg led with "record" — this is being called the largest-ever Hong Kong primary share sale and the third-largest in the world this year, after Alphabet and Intel — putting dilution lower in the story^[3]. Benzinga built its headline around Mugalli's "white flag" quote without matching space for an opposing view, and led with the smaller ADR decline even though the Hong Kong drop was more than twice as large^[4]. The South China Morning Post, which is owned by Alibaba, reported the 10% opening drop but framed the raise as fuel for global expansion — a conflict of interest the paper doesn't always flag in the story itself^[6].

What Happens Next Isn't Knowable Yet

Alibaba now has roughly $10.2 billion in cash it never has to repay, funding a plan the company says it has already spent about half of — a three-year, RMB380 billion pledge to AI and cloud infrastructure^[2][3]. Its existing shareholders own about 3.7% less of the company than they did last week^[5]. Both of those are settled facts. What isn't settled is which one turns out to matter more.

That answer depends on something neither side can prove yet: whether Alibaba's cloud revenue keeps growing faster than the money being poured into it. Cloud and AI compute revenue is already up 45% a year^[6]. Free cash flow is already deeply negative^[3][6]. Whether the first trend outruns the second is a question that will take several more earnings reports to answer — not this week's stock price.