Zhipu shares are flying. The Chinese AI lab, listed in Hong Kong as Knowledge Atlas Technology, surged as much as 48 per cent on Monday before closing up about 33 per cent, as Wall Street banks bet that China is the clear winner from Washington’s clampdown on Anthropic.
The trigger was timing. Days after the US government forced Anthropic to pull its most powerful Claude models from all foreign users, Zhipu said it would release GLM-5.2, its newest and most capable model, as open-source software this week, with no usage restrictions. The company framed it pointedly.
“Cutting-edge intelligence should not belong to only a few, nor should it be withdrawn at any time,” it said. “It should be open, available, extensible and built to serve every developer.”
Why Wall Street is buying Zhipu
The analysts moved fast.
JPMorgan kept its overweight rating and lifted its target to HK$1,400 from HK$950, while downgrading Zhipu’s domestic rival MiniMax. Bank of America initiated coverage on both with buy ratings. Their shared thesis: as US frontier models get pricier and harder to access, China is positioned to capture the “value-for-money” segment with cheap, capable models.
The 💜 of EU tech
The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!
The market has already run with it. Zhipu’s stock has risen more than tenfold since its January IPO, giving it a market value around HK$489bn, roughly four times MiniMax’s. Both firms are now eyeing a second listing on Shanghai’s STAR Market.
The company also has pricing power to match the hype: it raised cloud API prices by 8 to 17 per cent alongside its previous model in April, its second increase this year.
Open versus closed, and a talent warning
The split is becoming the defining line in AI. US labs are under pressure to restrict access to their best models; Chinese firms are doing the opposite, leaning into open distribution and winning over cost-sensitive enterprises.
Early community testing suggests GLM-5.2, which carries a one-million-token context window, performs close to Anthropic’s Claude Opus 4.7 on coding and long-horizon agent tasks, though that is preliminary, not an independent benchmark.
There is a sharper edge to the story, too.
Z-Ben Advisors’ Peter Alexander estimates that around 40 per cent of US-based AI engineers were born in China, and that the new directive effectively bars many of them from the systems they helped build, a recipe, he warns, for “brain flight” toward Chinese labs like DeepSeek and Moonshot.
The surge itself is sentiment as much as substance, and analyst targets are bets, not outcomes. But the direction of the bet is unmistakable: every door the US closes, China is rushing to hold open.
