Alibaba and tencent lose $66b in one day as ai dreams hit a wall
The promise was 100 billion dollars; the receipt was a 66 billion-dollar hole. In the span of a single trading session, China’s twin tech titans, Alibaba and Tencent, watched their combined market cap evaporate faster than any Lunar New Year fireworks. Investors, once giddy over OpenClaw-style agents, slammed the sell button when both companies failed to sketch a believable route from expensive silicon to actual profit.
Alibaba’s U.S.-listed shares plunged 7%, the steepest drop since October. Tencent, which had already shed $43 billion Thursday, went into free-fall again at the Hong Kong open, dragging the sector’s benchmark to a six-week low. The catalyst: quarterly calls that offered grand visions—Alibaba vowed to hit $100 billion in cloud-plus-AI revenue within five years—yet refused to detail who will pay, how much, and when.
Spending spree, revenue haze
Behind the rout lies a simple spreadsheet terror. Beijing’s champions are racing to build the same GPU empires that Meta and Amazon bankroll, but without the fat domestic cloud margins that let Silicon Valley absorb the burn. Alibaba’s net profit cratered 67% last quarter; marketing coupons for its new consumer bot, Wukong, ate what little cushion the e-commerce slowdown left.
Morgan Stanley cut Tencent’s price target 11%. Barclays did the same to Alibaba, scoffing that “the market now prices in perfection, and these numbers were anything but.” The firms insist they are planting seeds—Tencent harvests data from WeChat’s 1.3 billion thumbs, Alibaba commands logistics rails that snake into every third county—but traders wanted harvest forecasts, not soil science.
Openclaw euphoria meets cold storage
Only two weeks ago, Chinese users returned from holiday red-eyed from marathon sessions with OpenClaw, the local clone that books flights, ghostwrites memos and flatters your boss in the same chat bubble. Start-ups like MiniMax rocketed up the App Store, and Tencent’s stock hit a 52-week high. The momentum felt real until balance-sheet reality gate-crashed the party.
Now the sector stares at a paradox: consumers adore the product, yet no one can price it. Agents that summarise emails are cute; paying cloud bills for 50 million GPUs is not. Until either firm translates engagement into line items, every earnings call risks becoming a cliff.
The last time Chinese tech bled this fast, regulators were doing the punching. This time the enemy is inside the algorithm: an intelligence so artificial it still can’t invoice.