Alibaba bets $100b cloud-ai jackpot while profits crater 67%
Alibaba’s boardroom just served up a cocktail of panic and bravado: net profit nosedived 67% last quarter, the worst since early 2024, yet CEO Eddie Wu is promising to 5× revenue to $100 billion within five years—fueled almost entirely by cloud and AI.
The math that doesn’t add up
Revenue crawled up 2% to 284.8 billion yuan ($41.3 billion), missing分析师 consensus by a hair. U.S.-listed shares plunged 9.9% intraday, the steepest drop since April. Promotional subsidies meant to fend off JD.com and Meituan ate margins alive, while the core e-commerce engine sputters.
Now the company is pivoting hard: Agentic AI models, a proprietary Token Hub, and a capital-expenditure fire hose aimed at its cloud division. Analysts at Bloomberg Intelligence aren’t buying the spectacle. They note that even a surge in cloud demand won’t offset the margin pressure on Alibaba’s cash-cow retail and food-delivery units, which still bankroll every pipedream.
Why the cloud bubble could burst first
China’s hyperscalers—Alibaba, Baidu, Tencent—are locked in a GPU arms race against a backdrop of U.S. export bans. Nvidia’s latest Hopper chips can’t legally enter the mainland in volume, forcing Alibaba to lean on home-grown, lower-yield silicon or gray-market crates sold at 40% premiums. That drives up cost per training cycle and compresses gross margin faster than revenue can scale.
Meanwhile, enterprise clients remain tight-fisted. CIOs surveyed by Gartner last month ranked “regulatory uncertainty over AI output” as their top concern, delaying seven-figure cloud contracts. Alibaba’s own pipeline confirms the jitters: annualized cloud revenue growth decelerated to 3% excluding top-ups from sister subsidiaries.
The cold storage verdict
Investors wanted a turnaround story; they got a PowerPoint and a prayer. Until Alibaba proves it can monetize LLMs beyond internal cross-subsidies, that $100 billion target is vaporware drifting above a profit hole. The next earnings call, due in August, will show whether Wu’s AI chips are made of silicon—or just silicon theater.