Alibaba Profit Falls 75% as AI Capex Swamps 9% Sales Growth
Alibaba's quarterly revenue rose 9% on AI-fueled cloud demand, but profit dropped 75% and adjusted earnings missed estimates as capital spending on AI infrastructure ballooned.
Alibaba reported on Thursday that quarterly revenue rose 9% on strong AI-driven cloud demand, but profit fell 75% and adjusted profit missed analyst estimates because of heavy capital expenditure on AI infrastructure.
Alibaba Group Holding Ltd. (BABA) told investors on Thursday that its quarterly revenue grew 9%, helped by AI-driven demand in its cloud division, and that profit fell 75% as the company poured cash into the data centres and chips needed to serve that demand. Adjusted profit — the figure that strips out one-off items and is the number most analysts model — came in below estimates, and the company attributed the shortfall to heavy capital expenditure.
That is the whole tension of the quarter in one sentence: the top line is being pulled up by artificial intelligence, and the bottom line is being pushed down by the cost of supplying it. Which of those two forces the market chooses to weight determines whether a 75% profit decline reads as an investment cycle or as the beginning of durable margin erosion.
The market shrugged, then bought the dip
Shares did not behave like a stock that had just disclosed a three-quarters drop in profit. As of the last trade at 18:49 GMT on Thursday, 20 August 2026, BABA was quoted at 129.89 in our licensed market feed — up 0.77% from the prior close of 128.90. The feed does not specify the listing currency for the quote, so the level is best read as a percentage move rather than an absolute price.
More revealing than the net change is the day's path. The intraday range ran from a low of 121.88 to a high of 130.24, meaning the stock recovered roughly 6.6% from its worst level of the session to where it was trading late in the day — an illustrative calculation from the range figures, not a company-reported number. In other words, the first reaction was to sell the profit line, and the second reaction was to buy the revenue line.
It did so against a weak tape. The S&P 500 tracker (SPY) was down 0.68% at $763.82, the Nasdaq 100 proxy (QQQ) was off 0.89% at $709.68, and the Dow 30 fund (DIA) was the worst of the three at $528.20, down 1.14%. All three sat at or near the bottom of their daily ranges. A Chinese internet name closing green on a day when U.S. large-cap benchmarks were broadly lower is a signal in itself about how the AI spending narrative is being received.
Why capex is doing the damage
Capital expenditure hits reported profit through two channels, and it is worth separating them because they have different half-lives.
- Depreciation. Servers, accelerators and data centre fit-out are capitalised and then written down over their useful lives. Every dollar of AI hardware bought this year becomes a recurring charge against operating profit for several years afterwards, whether or not the revenue arrives on schedule.
- Operating drag. Power, cooling, network capacity and the staff to run the estate hit the income statement immediately, and they are incurred ahead of the utilisation that eventually pays for them.
That combination is what turns a 9% revenue increase into a 75% profit decline. The revenue growth is real, but the cost base was built for a bigger number than the one that has arrived so far. Alibaba's own framing — that the adjusted profit miss was due to heavy capital expenditure rather than weak demand — is the standard defence of an investment cycle, and it is a defensible one. It is also unfalsifiable in any single quarter, which is precisely why the disclosure discipline around capex and cloud growth matters more from here than the headline earnings figure.
What separates an investment cycle from margin erosion
The distinction is not philosophical; it shows up in a handful of observable lines over the next several reporting periods.
The distinction is not philosophical; it shows up in a handful of observable lines over the next several reporting periods.
First, does cloud revenue growth accelerate faster than the capex that precedes it? Spending that runs ahead of demand for one or two quarters is a build-out. Spending that runs ahead of demand indefinitely is a structural cost problem. Alibaba has told investors the cloud business is growing on strong AI demand — the test is whether that growth rate keeps widening its gap over group revenue growth of 9%.
Second, does the e-commerce core keep funding the build? Alibaba's marketplaces have historically been the cash engine that pays for everything else. If the group's consolidated growth stays in single digits while capex rises, the internal subsidy gets harder to sustain without either slowing the AI programme or leaning on the balance sheet.
Third, does depreciation stabilise as a share of cloud revenue? That is the cleanest single read on whether the infrastructure is being absorbed. A rising depreciation charge on flat utilisation is margin erosion by another name.
The same trade-off, across the whole sector
Alibaba is not making an unusual choice. Every large operator of AI compute is currently reporting the same shape of result: demand commentary that is stronger than the prior year, capital plans that are larger than the prior year, and near-term earnings that are worse for it. The market has, for the most part, been willing to pay for the demand and forgive the earnings — which is roughly what Thursday's intraday recovery in BABA looked like in miniature.
The risk in that bargain is that it is priced on faith in utilisation. Data centre capacity that is committed today has to be filled by customers who have not yet signed, at prices that have not yet been set, in a market where the cost of the underlying accelerators is still moving. Investors who accept a 75% profit decline as the price of admission are, in effect, underwriting that filling.
The revenue and profit figures were reported Thursday, as detailed by BNN Bloomberg. What to watch next is not the next headline profit number — that is likely to stay pressured while the build continues — but the trajectory of cloud growth against the capex line, and whether management gives investors a clearer end point for the spending.
Key facts
- Ticker and last trade: BABA — 129.89, +0.77%, as of 18:49 GMT on 20 Aug 2026 (feed currency unspecified)
- Profit change: Down 75% after ramping up AI infrastructure spending
- Revenue growth: Up 9% in the quarter, reported Thursday
- Adjusted profit: Missed analyst estimates, attributed to heavy capital expenditure
Frequently asked questions
Why did Alibaba's profit fall 75% if revenue grew?
The decline was driven by heavy capital expenditure on AI infrastructure. Building data centre capacity creates immediate operating costs for power, cooling and staffing, plus recurring depreciation charges on the hardware. Those costs are incurred before the revenue they are meant to support arrives, so a 9% revenue increase was not enough to offset them in the quarter.
Did Alibaba miss analyst expectations?
Yes on the earnings side. The company missed estimates for adjusted profit — the measure that strips out one-off items and is the figure most analysts model — and pointed to heavy capital expenditure as the reason. Revenue grew 9% in the quarter, with the cloud business benefiting from strong artificial intelligence demand.
How did the stock react to the results?
BABA held up better than the profit line suggested. As of the last trade at 18:49 GMT on 20 August 2026 it was quoted at 129.89, up 0.77% from the prior close of 128.90, having traded as low as 121.88 earlier in the session. That gain came on a day when major U.S. benchmarks were lower.
What was driving the cloud growth?
Alibaba attributed cloud growth to strong artificial intelligence demand. AI workloads require far more compute per customer than traditional hosting, which lifts revenue per account but also requires the operator to install expensive accelerators, networking and power capacity ahead of that demand — the source of the current earnings pressure.
Is this an investment cycle or a permanent margin problem?
That cannot be settled from one quarter. The distinguishing tests are whether cloud revenue growth accelerates faster than capital spending, whether depreciation stabilises as a share of cloud revenue, and whether the e-commerce core keeps generating enough cash to fund the build without leaning on the balance sheet.
How does Alibaba's situation compare with other AI infrastructure builders?
The pattern is common across large operators of AI compute: stronger demand commentary, larger capital plans and weaker near-term earnings as a result. Markets have generally been willing to look past the compressed profits and pay for the demand, which effectively means investors are underwriting future utilisation of capacity being committed today.
