01 The Biggest Share Placement in Hong Kong History
On 26 August, Alibaba formally closed a HK$80 billion share placement: 710 million new shares at HK$112.70 each, raising net proceeds of HK$79.7 billion. The book was reportedly covered almost three times in under an hour, with total demand reaching roughly US$28 billion. Sovereign wealth funds from the Middle East and long-only institutions in Europe were among the key buyers; together, sovereign and long-term investors took more than 40 per cent of the deal.
This is Alibaba’s first Hong Kong placement since its 2019 secondary listing, and the largest new-share placement ever completed in Hong Kong’s equity market.
Alibaba has also been unusually explicit about where the money will go. Around 60 per cent, or HK$47.9 billion, is earmarked for expanding global computing infrastructure, while the remaining 40 per cent, or HK$31.9 billion, will go towards building hyperscale AI data centres. In short, every dollar is being pushed into AI.
So is Alibaba short of cash? On the face of it, no. As of 30 June, the company held roughly RMB474.5 billion in cash and other liquid investments. Quarterly revenue came in at RMB268.9 billion, up 9 per cent year on year, and operating cash flow remained positive at RMB22.9 billion.
The problem is not solvency. It is the sheer cost of staying in the AI race.
Quarterly capital expenditure has jumped from RMB38.6 billion a year earlier to RMB67.6 billion, up 75 per cent year on year and 151 per cent quarter on quarter. Free cash flow, which was already negative RMB18.8 billion a year ago, deteriorated further to negative RMB44.6 billion. Losses in Alibaba’s AI laboratory and application segment widened from RMB3.2 billion to RMB13.8 billion.
Back in February 2025, Alibaba unveiled a three-year plan to invest at least RMB380 billion in AI and cloud infrastructure — more than it had spent in the area over the previous decade combined. By the end of the June quarter, it had already deployed about RMB190 billion, meaning the programme was more than halfway through.
At the current pace, Alibaba is spending close to RMB200 billion a year. Even a large cash pile begins to look less comfortable when the burn rate starts to resemble the market capitalisation of a mid-sized listed company every quarter.
Which raises the obvious question: if the company has cash, why not spend its own money, or raise debt instead?
The answer is straightforward. Alibaba wants to buy computing power without tying itself to fixed repayment obligations.
GPU clusters are typically depreciated over three to five years. Data centres are assets with a life measured in decades. These are long-duration investments in a field where the technology changes quickly. Match them with debt, and you create a rigid repayment burden. If the technology stack shifts or utilisation disappoints, the debt remains even if the economics do not.
Equity financing is different. Existing shareholders are diluted — the placement amounts to about 3.57 per cent of the enlarged share capital — but there is no fixed maturity and no refinancing cliff. The risk stays in the equity rather than migrating into a credit problem.
Nomura has made a related point: the global surge in AI capital expenditure has already driven a sharp increase in tech bond issuance, while credit spreads have widened and order books have become less exuberant. Debt is no longer as attractive, or as cheap, as it once was.
As investor Guo Tao put it, this placement is not a sign that Alibaba’s cash flow has collapsed. It is an active capital allocation decision made in the middle of an AI arms race.
In other words, Alibaba is not out of money. It simply does not want to gamble its own balance sheet alone.
On the earnings call, chief executive Eddie Wu argued that full-stack AI platform services are, by nature, a capital-intensive business, but that AI computing investment can pay back within three years.
That is management’s case. The market, however, is asking harder questions.
02 Management is buying while Michael Burry is heading for the exit
On 24 August, the day the placement was priced, Alibaba’s Hong Kong shares closed at HK$112.50, down 8.54 per cent, after falling more than 10 per cent intraday and briefly trading below the placement price. Roughly HK$200 billion of market value disappeared in a single session.
Then Joe Tsai stepped in.
He bought 720,000 shares at roughly HK$112 apiece, spending about HK$80 million. Eddie Wu followed with 350,000 shares at around HK$111.6, for about HK$40 million. Both men bought at levels even below the placement price of HK$112.70.
Jack Ma has also been adding to his Hong Kong-listed holdings, with cumulative purchases reportedly exceeding HK$600 million in recent days.
Taken together, Ma, Tsai and Wu have bought more than HK$800 million worth of stock at a moment of peak market scepticism. Management is sending a simple signal: we still believe in the AI strategy, and we are willing to use personal capital to say so.
But on the very same day, one prominent investor went the other way.
Michael Burry, whose trades inspired The Big Short, wrote on Substack on 23 August that he had sold out of Alibaba and rotated into JD.com.
His objection was blunt: he does not agree with the decision to issue new shares. In his view, the placement is not a one-off event but the opening chapter of a multi-year AI capex cycle that will drag further on return on invested capital. He went further still, saying he would not be interested in Alibaba again unless the share price fell by half.
That view is not based on rhetoric alone. The valuation contrast between Alibaba and JD is stark.
Alibaba trades on about 25 times trailing earnings, while its free-cash-flow yield is negative 4.2 per cent. Put simply, investors are paying up for a company that is currently burning cash.
JD looks very different: around 17.9 times trailing earnings, only 8.3 times forward earnings, a positive free-cash-flow yield of 10.7 per cent, and a dividend yield of 3.3 per cent. Its core retail business continues to post record margins, whereas Alibaba’s latest quarterly net profit fell by roughly 75 per cent year on year.
At a basic level, Burry has made a classic value rotation: sell the expensive cash-consuming asset and buy the cheaper cash-generating one.
Of course, not everyone is convinced by JD either. Critics argue that it remains highly exposed to electronics demand and could feel the pain if government trade-in subsidies fade.
Still, the contrast on 24 August could hardly have been sharper. Alibaba insiders were buying. Burry was selling. And southbound funds were net buyers to the tune of HK$11.567 billion, with internet heavyweights among the main beneficiaries.
Management, Burry and mainland capital are reading three very different versions of the same story.
03 Everyone is spending — and spending heavily
Alibaba is not burning cash in isolation.
Tencent’s second-quarter capital expenditure reached RMB52.8 billion, up 176 per cent year on year and 65 per cent quarter on quarter, well ahead of expectations. Its free cash flow also turned negative for the first time since listing. Net cash fell from RMB146.8 billion at the end of March to RMB58.1 billion by the end of June — a 60 per cent drop in just three months. Pony Ma has framed this as the birth of an “AI-enabled new Tencent”, while Bank of America has called it Tencent’s “Midway moment”: the turning point has arrived, but the winner is not yet clear.
Between them, Alibaba and Tencent spent more than RMB120 billion on capital expenditure in a single quarter.
The US giants are spending on an even larger scale. Microsoft, Alphabet, Amazon and Meta are projected to spend more than US$750 billion on AI capital expenditure in 2026, nearly double the 2025 level.
Alphabet’s second-quarter capex reached US$44.9 billion, up 100 per cent year on year. Free cash flow fell to negative US$5.855 billion — its first negative quarterly free cash flow since listing in 2004. After earnings, the shares dropped 7 per cent, wiping out more than US$280 billion in market value in a day, while five-year CDS widened to a record 67 basis points.
CDS, or credit default swaps, are a rough market gauge of default risk. The higher the spread, the more investors worry about whether a company’s debts might become harder to service. When Alphabet’s CDS hits a record high, the market is effectively asking whether even one of the world’s greatest cash machines can spend this aggressively without damaging its financial profile.
Microsoft, by contrast, looks much steadier. Free cash flow remains solid, CDS trades below 20 basis points, and the share price reaction has been far milder.
The contrast matters. Both companies are spending heavily, but the market sees Alphabet as spending too fast and Microsoft as spending in a way it can clearly absorb.
Look a little further ahead, and another force comes into view. Anthropic could go public as soon as October, with some investors floating a valuation of US$2 trillion. If that happens, it would overtake SpaceX’s US$1.77 trillion and become the largest IPO valuation in commercial history. By choosing to place shares now, Alibaba may be trying to lock in long-term capital before US AI listings start siphoning off global investor attention.
The whole world is spending on AI. Alibaba is simply one of the most visible participants.
04 The real question is not how much is being spent, but when it pays back
The contrast between Alphabet and Microsoft points to the market’s new obsession. Investors no longer want to know only who is spending more. They want to know when, and whether, that spending will earn a return.
Several analysts argue that the market’s framework for pricing the AI boom is shifting. Trillions in capex are eating into free cash flow, and valuation is moving away from distant narrative premiums towards demonstrated cash returns.
In plain English, the market used to buy the story. Now it wants to inspect the ledger.
Three deeper forces are driving that shift.
First, rates. Ten-year US Treasury yields are holding around 4.7 per cent, while the 30-year has moved above 5.2 per cent. When discount rates rise, distant cash flows become less valuable. Investors grow less tolerant of prolonged cash burn.
Second, the industry cycle. AI is moving from the infrastructure build-out phase into the phase of commercial validation. It is no longer enough to show that you can deploy computing power. The market wants to know whether downstream customers will pay, what the unit economics look like, and how quickly invested capital comes back.
Third, the investor base. Long-only allocators — pension funds, insurers and other patient capital — are becoming increasingly important at the margin. These investors are not paying for slogans. They care about stable free cash flow, sustainable buy-backs and the capacity to keep returning capital.
The market is no longer content to pay in advance for a grand future. It is starting to cost out the AI arms race line by line.
Alibaba stands right at the centre of that transition.
On one side of the ledger, the business does have evidence that AI is generating revenue. AI cloud and computing services brought in RMB48.4 billion, up 45 per cent year on year, the fastest growth in 22 quarters. Revenue from AI-related products has now delivered triple-digit growth for a twelfth consecutive quarter. The cloud division’s margin improved to 12 per cent, while adjusted EBITA rose 133 per cent.
On the other side, losses in the AI laboratory and applications segment widened from RMB3.2 billion to RMB13.8 billion. Free cash flow has been negative for two straight quarters, with first-half net outflows above RMB62 billion.
Eddie Wu says AI computing investments can pay back within three years, and that the cycle could shorten to two and a half years — or even two — as product margins improve and in-house chips replace more third-party hardware.
Can Alibaba deliver that? Three tests matter.
First, can AI-related revenue continue to outgrow the broader group? At the moment, cloud is growing at 45 per cent against overall revenue growth of 9 per cent, so the foundation is there.
Second, can unit computing costs come down, helped by proprietary chips and scale effects?
Third, can the cloud business keep expanding margins as it grows? The current 12 per cent margin is an encouraging early sign.
If Alibaba passes all three tests, this HK$80 billion placement will look disciplined and farsighted. If it fails, the same deal will be remembered very differently.
Which brings us back to the starting point. The placement is done. Sovereign funds rushed in. Management bought shares. Southbound money added exposure. Yet Burry walked away, and the stock fell 8.54 per cent on pricing day.
One camp says: we believe in Alibaba’s AI future.
Another says: come back when the price is half this level.
That divide may be the most important signal of all. The market has started to do the arithmetic.
Some investors are leaning in. Others are walking away. Time will decide who is right.