Finance News · Monday, 24 August 2026

01 · Briefing · what happened

Alibaba's profit fell 75%, its spending rose 75%, and it just asked for $10.2bn more

Finance News 2 min 14 sources

The market marked the shares down to almost exactly the price of the new stock. Nobody can yet tell whether the AI money is being built into something or burned - and that is the whole argument.

$10.2bn

Alibaba's share sale

710 million new shares, all of it for AI [1][2]

-75% / +75%

profit down, spending up

in the same June quarter [1]

$4,617

gold an ounce

near May levels, from $4,000 in mid-July [12]

$93.45

Brent crude

down 1% as traders adjusted before the sanctions [6]

At a glance

Alibaba will issue 710 million new shares at HK$112.70, raising $10.2bn to spend on AI [1][2].
That is a discount to Friday's HK$123 close, and the shares fell as much as 10% to meet it [1][3].
The raise follows a 75% fall in quarterly profit and a 75% rise in spending on equipment [1].
Michael Burry sold his stake over the issue, saying he cannot bless share issuances [4].
Oil eased about 1% before Washington's Iran announcement, Brent to $93.45 [6][7].
Nvidia reports Wednesday, with revenue expected to nearly double to about $92bn [5].
The dollar held near multi-month lows; bitcoin rose almost 23% against it last week [9].
Gold traded at $4,617.49, up from about $4,000 in mid-July, on worry about the currency losing value [12].

Forces in play

AI spending High↑

Alibaba's equipment spending rose 75% in a quarter and it is raising $10.2bn more [1]

Doubt about the payoff Building↑

the shares fell to the offer price, and Burry sold arguing each pound invested earns less [1][4]

Money leaving the dollar Building↑

gold near $4,600, bitcoin up 23% in a week, the euro above $1.16 [9][12]

Trading before the news Easing↓

shares flat and oil down 1% as investors waited for Nvidia, Jackson Hole and the Iran package [5][6]

In play Alibaba — raising $10.2bn to fund AI after profits fell 75% Michael Burry — sold out over the share issue Scott Bessent — US Treasury Secretary, announcing Iran sanctions Monday Kevin Warsh — Fed chair, speaking at Jackson Hole on Friday Nvidia — reports Wednesday; the test of the AI spending case

How it unfolded

Mid-July gold trades around $4,000 an ounce [12]
Wednesday the Treasury says it will at least double long-bond buybacks [10]
Last week bitcoin gains almost 23% against the dollar, gold 5% [9]
Sunday Alibaba announces a $10.2bn placement of new shares [2]
Monday Alibaba falls to the offer price; oil eases before the sanctions [1][6]
Wednesday Nvidia reports; the placement closes [1][5]

Where this points

Nvidia's results on Wednesday are the nearest thing to a read on whether the AI spending is turning into revenue anywhere; watch the guidance rather than the quarter, because the quarter is already assumed.

Full briefing

Two numbers pointing opposite ways

Alibaba announced on Sunday it would issue 710 million new shares at HK$112.70 to investors outside the US, raising HK$80bn - about $10.2bn [1][2]. Friday’s closing price was HK$123 [3]. By Monday morning the shares had fallen as much as 10%, to HK$112.7 [1]. The market moved the price to the offer.

The raise landed days after Alibaba reported a 75% fall in June-quarter profit. Over the same stretch, its spending on buildings and equipment rose 75%, to 67.7 billion yuan [1]. Every yuan of that went out of the door as a cost. Whether it comes back as a business is the thing nobody can settle yet.

What Burry would not bless

Michael Burry, who bet against US housing before the 2008 crash, sold his Alibaba position and moved into JD.com, saying the shares had grown too expensive [4]. He named the raise directly: “I cannot bless share issuances” [4]. His argument was that Alibaba’s return on invested capital - the profit it earns on each pound it puts in - was likely to keep falling [4].

That is the disagreement in one line. Alibaba says it is buying the future. Burry says it is buying less and less with each pound.

A week that waits

Very little else moved. Asian shares were flat and S&P futures barely budged [5]. Oil slipped about 1% as traders adjusted before Washington’s announcement, Brent to $93.45 and US crude to $86.14 [6][7]. Scott Bessent, who unveils the Iran measures on Monday afternoon, wrote that “at dawn begins an economic D-Day” [8]. Commerzbank’s Barbara Lambrecht expects the package to aim at buyers of Iranian oil, “the most important among them is China” [7].

Nvidia reports on Wednesday, with analysts looking for revenue to nearly double to around $92bn [5]. Fed chair Kevin Warsh speaks at Jackson Hole on Friday [5].

Out of the dollar

The dollar sat near multi-month lows [9]. The euro held above $1.16, the yen stayed stronger than 159, and the Australian and New Zealand dollars traded just short of three-month highs [9]. Bitcoin rose nearly 23% against the dollar last week; gold rose 5% [9].

The trigger was at home. On Wednesday the Treasury said it would at least double its planned buybacks of long-term government debt [10]. The aim was to calm a sell-off that had pushed the 30-year borrowing rate to its highest since 2007 [11]. It calmed the bond market briefly and unsettled everything else. ING’s strategists put gold’s run down to revived worry about the currency losing value [12]. Spot gold traded at $4,617.49, back near May levels after starting mid-July around $4,000 [12].

Elsewhere: Samsung Electronics fell 5.2% after saying shareholder returns could reach 110 trillion won this year [13]. Shein opened books on a Hong Kong listing, seeking up to $1.77bn at a valuation of $26.8bn [14].

02 · Lesson · why it matters

The two kinds of spending that look identical on the day

Money that builds and money that bleeds leave the account the same way, and only what comes back later tells them apart.

How it works

Money goes out of the door
The accounts record a cost, either way
A cost that builds something and one that does not look identical
Only what comes back later tells them apart
So the market prices the doubt in the meantime

The twist

Spending that is building something and spending that is bleeding look exactly the same on the day it leaves - the difference is made later, which is why the argument is never about the number.

Where you've seen this

A student

three years of no wages is either tuition or lost income, and nobody knows which yet

A farmer

a field left fallow is a year of nothing, or next year's yield

A government

borrowing for a railway and borrowing for salaries appear in the same column

A sports club

a rebuilding season is a plan, or a bad season with a nicer name

The catch

The delay is exactly what makes it a good excuse. Anything can be called an investment while the payoff is still in the future, which is why watchers fall back on a cruder question: is this spender earning more or less on each pound than they used to?

Full lesson

A company reports two numbers, pointing opposite ways

Alibaba’s profit for the June quarter fell by three quarters. Its spending on buildings and equipment rose by three quarters. Days later it asked investors for another $10.2bn, all of it earmarked for artificial intelligence.

Read that as a collapse and it fits. Read it as a company building hard and it fits equally well. Both readings use the same two numbers.

That is not a failure of reporting. It is a fact about what money looks like on the day it leaves.

The account has one column

When a company spends, the money goes out. That is all the ledger records at the moment of the transaction. It cannot mark one payment “this becomes a data centre that earns for a decade” and another “this was wasted”.

Accountants have a rough fix. Money spent on lasting things gets spread over the years it is expected to serve, instead of landing all at once. But that spreading is a forecast about the future, made by the people doing the spending. It does not settle the question. It just states an opinion in a tidier format.

The question - was that a cost or an investment? - is not answerable on the day. It is only answerable by what the thing goes on to earn.

Which is why the argument is never about the number

Nobody disputed Alibaba’s figures. Michael Burry, who made his name betting against American housing before the 2008 crash, sold his stake anyway. His stated reason was not that the spending was too large. It was that the company was earning less and less on each pound it put in.

That is the only honest way to argue about spending whose payoff has not arrived. You cannot look at what is going out. You look at the record of what this same spender has got back before.

Notice that this is a claim about history, not about the future. It is what you reach for precisely because the future is unavailable.

The delay is the loophole

Here is the uncomfortable part. The gap between spending and payoff is real, and it is also the perfect hiding place.

Anything can be called an investment while the return is still ahead of it. A struggling company and a company building the next decade say the same sentence, in the same tone, with the same conviction. For a while, neither can be contradicted. The word “investment” does work that no evidence has yet done.

Every one of us uses this. The unread books. The gym membership. The course. The year spent on something that might come to nothing. We are not lying when we call these investments. We just do not know yet, and the not-knowing is comfortable enough that we rarely go back and check.

Who else is standing in this gap

A student earns nothing for three years. That is either tuition or lost income, and the same person will describe it both ways depending on how the following decade goes.

A farmer leaves a field unplanted. That is a year of no harvest, or it is next year’s yield, and the soil decides.

A government borrows. If the money built a railway, the debt bought something; if it paid this year’s salaries, the debt bought this year. Both appear in the same line of the same accounts, and both are defended with the same word.

The people affected are rarely the ones spending. The Alibaba shareholder who woke up on Monday to find the shares marked down to the price of the new stock did not choose the spending. Neither did the employee whose department is now a cost line, nor the citizen who will service the borrowing.

What this leaves us holding

The gap between spending and knowing is not a flaw somebody could fix with better accounts. It is the shape of the thing. Anything that takes time to build takes time to judge, and during that time the two possibilities are genuinely indistinguishable.

So the confident readings on both sides today - the collapse, the visionary build - are made in the same fog, from the same two numbers. Nobody making them will know for years. Including the ones who sound the most certain, and including us, reading it from outside.

03 · Lab · your turn

Cost or Investment

Judge which companies were building and which were bleeding, first from the accounts alone and then from what each earned on every pound invested.

04 · Hope · carry this

Alibaba can raise ten billion dollars in a weekend because enough strangers still believe a Chinese company will honour a share certificate. That much trust is ordinary now, and it was not always.