According to Bloomberg Business, Zhipu AI, officially known as Knowledge Atlas Technology JSC Ltd., began trading on the Hong Kong Stock Exchange on Thursday after a $558 million initial public offering. The company, founded by Tsinghua University researchers and backed by Alibaba and Tencent, sold 37.4 million shares at HK$116.20 ($14.9) each. Retail investor demand was massive, with shares oversubscribed by more than 1,159 times, and the stock jumped as much as 35% in the gray market before its official debut. Zhipu reported 2024 revenue of 312.4 million yuan and plans to use 70% of the IPO proceeds for R&D on its large AI models. Its listing is seen as a key test for China’s “AI tiger” startups, which aim to rival U.S. firms like OpenAI but face U.S. chip export restrictions.
The AI Gold Rush Is In Full Swing
Look, the frenzy around Zhipu’s IPO isn’t happening in a vacuum. It’s part of a huge, coordinated capital push into the entire Chinese AI supply chain. On the same day Zhipu debuted, GPU maker Shanghai Iluvatar CoreX also listed. Its direct rival, MiniMax, is set to go public Friday after a $619 million IPO. And just last week, chip designer Biren Technology had a blistering debut. So what’s going on here? Basically, the market is placing a massive, interconnected bet. Investors aren’t just buying Zhipu’s software story; they’re buying the whole domestic stack—from the silicon up. The narrative, as Bernstein analysts put it, is that “China‘s AI development is only months behind global leaders.” After years of playing catch-up, the market is screaming that it’s finally time.
Revenue vs. Ambition, The Hard Math
Here’s the thing that gives me pause. Let’s talk numbers. Zhipu is valued at about $6.6 billion post-IPO. Its 2024 revenue was roughly $43 million. That’s a staggering valuation multiple. For comparison, its market cap is actually lower than recent chipmaker listings like Biren and Moore Threads. That tells you where investor confidence is strongest right now: in the hardware layer. The bet on Zhipu is a longer-term, more speculative one on its ability to build a foundational model that can truly compete. The company says it’ll plow 70% of the IPO cash into R&D, which it absolutely must do. But with U.S. restrictions cutting off the most advanced Nvidia chips, that R&D is happening on a harder, more expensive playing field. Can they close the gap with OpenAI while building on potentially inferior hardware? That’s the billion-dollar question.
The State Is The Key Customer
Zhipu’s business model reveals a critical, and often overlooked, advantage for Chinese AI firms. They’re not just chasing the next ChatGPT clone for consumers. A huge part of their strategy is serving state-owned enterprises and government contracts. These clients prefer building custom, on-premise AI infrastructure rather than using public cloud services from the big tech giants. This is a moat. With backing from Alibaba, Tencent, and local government funds, Zhipu is perfectly positioned to be the “safe,” domestic supplier for these sensitive, large-scale projects. So while its current revenue is modest, its path to scaling might be more guaranteed and stable than a purely commercial Western startup. It’s a different kind of market, one where government relationships are as important as model benchmarks.
Bubble Talk Is Already Here
And you can’t ignore the bubble whispers. A Fidelity International director quoted in the report nailed it, pointing to the “ongoing debate around the risk of an AI-related market bubble.” When retail investors oversubscribe an offering by over 1,100 times, you have to wonder about the heat in the kitchen. This isn’t just about Zhipu’s technology; it’s about a macro narrative on Chinese tech self-sufficiency and a fear of missing out. One analyst thinks the shares could trade at nearly double the issue price. That’s euphoria. I think the real test won’t be the first-day pop—we’ve seen those across the sector—but where these stocks settle in six months. Will the revenue materialize to justify these valuations? Or is this a capital-intensive land grab where only a couple of players survive? The IPO is a milestone, but the marathon is just beginning.
