Zhipu is down 60% from its peak. More than three hundred billion Hong Kong dollars, gone in two trading days. So — that question we've kept asking? We have the answer now. If you're just joining, Greater China's AI listing rebound was already building before this week. Zhipu surged after its January Hong Kong IPO, then went hunting for a big share sale. Moonshot AI was reported to be prepping a Hong Kong listing inside six months after Kimi K3 caught the market's eye. All along, we were asking one thing: can private-market valuation ambition survive public scrutiny? This is AI IPO Watch. Today, the tape finally talks back — a $50 billion number nobody's filed, a CEO saying what the fundraise is actually for, and Zhipu's wreckage right in the middle of it. This one's from The Business Times:
MOONSHOT AI is preparing to begin discussions in August on a final round of fundraising before listing in Hong Kong, capitalising on the excitement around its latest model to raise capital at a valuation of as much as US$50 billion.
Moonshot AI, pre-IPO fundraising talks at up to fifty billion dollars, per The Business Times. Just to be clear, that's sourced to talks — no filed document, no priced round. It's a step-up, though — the last number around this name was north of thirty billion for the Hong Kong listing. So in a few weeks, the number being floated has jumped by about two-thirds. Right. And the one hard fact in here I'll actually credit is annual recurring revenue: three hundred million in June, up from two hundred million in April. That's a real gauge. We have ARR in the reporting; we don't have a filing for fifty billion. The Greater China listing rebound just got repriced upward on this. But here's what makes me twitch — Moonshot's floating fifty billion the same morning Zhipu's down sixty percent from its peak. Same venue, same category of paper. So now it's about timing: can this round close before institutional memory catches up to the tape? That's exactly it. If your comps are listed and cratering, staying private won't insulate you — the sentiment finds you anyway. Here's Joy Iyke at TipRanks:
Databricks CEO Ali Ghodsi has revealed the key reason behind the AI and cloud company’s latest fundraising effort. He made the revelation just days after Databricks announced a new funding round that valued the company at $188 billion. The comments highlight growing demand for AI computing infrastructure across major markets, increasing the need for additional investment.
Ali Ghodsi just said out loud that Databricks raised because they're running out of GPUs. You don't usually see that in a polished roadshow deck. It's a cost-and-capacity risk factor, and he's using it as a pitch. And it's on the record, CEO-sourced. Which is more than I can say for most of the numbers we've handled this week. This raise puts the private mark at $188 billion, up from $134 billion earlier in the year. So the price tag climbs while the stated reason for raising is: we can't get enough chips to run the thing. Right — and that changes how you read the whole raise. The money is going into operational scarcity while the business model is still maturing. Very different risk profiles, same valuation wrapper. It also sharpens the exit-pressure story. The candid version starts with the compute bill: 'the compute bill is eating us alive.' IPO readiness comes second. Honest, at least. Sid Times writes:
Jijia Vision is currently in talks to conduct an initial public offering on the Hong Kong stock exchange, potentially this year, as part of a wave of Chinese AI companies preparing for public listings. The Beijing-based firm, also referred to as GigaAI, is nearing the completion of a funding round that values the company at $3 billion as it moves closer to going public, according to founder and CEO Huang Guan.
GigaAI — Jijia Vision — now has its CEO on record saying they want to be the first world-model startup anywhere to ring the bell. Huang Guan said it to Bloomberg at the Shanghai AI conference, and he's tying it to a $3 billion round they're closing. The $3 billion is at least CEO-attributed and tied to a round that's nearly closed. That's a better-anchored number than a lot of what's crossed the desk this week. The 'first to go public in world models' line is the pitch, though. That's a category-of-one claim, and category-of-one claims are how you justify a multiple nobody can comp. And he wouldn't say whether they've hired bankers. So for now, the listing is still at the ambition stage. No mandate yet. Last week we said the VIE cleanup made a Hong Kong listing mechanical rather than aspirational. This confirms they're actively in talks this year. The plumbing's real. Right, but 'in talks, potentially this year' with no syndicate named is still a long way from a priced deal. The board clock decides that, not the funding round. Lei Jianping, writing in The Insight Asia:
Shares in Chinese artificial intelligence company Z.AI (Zhipu AI) (2513.HK) extended their steep decline on Monday, wiping more than HK$300 billion ($38.3 billion) from the company’s market value over two trading sessions as investors weighed the impact of expiring lock-up periods and rising competition in China’s fast-moving foundation model market.
On Monday, July 20, we were asking whether the market could absorb new supply once lock-ups expired. Z.AI just gave us the answer. Down 19.6% Monday, 28% Friday — 60% off the peak, HK$300 billion gone in two sessions. And the trigger's right there in the filing timeline. A HK$31.4 billion placement closed one week ago, and the lock-up expiry hit right behind it. Early investors got the green light to sell, and they sold. The placement priced at HK$1,588 a share. It closed Monday at HK$890.50. So the people who bought that paper are down 44% in a week — and we still don't know who they are, because the buyer identities were never disclosed. That $38 billion that just evaporated — that's bigger than the whole valuation of half the Chinese AI names queuing up in Hong Kong right now. It's the first real wreckage in the cycle, and it's tied to something completely mechanical: supply hitting a market that priced on narrative. Which is exactly why the Moonshot $50 billion number we hit earlier lands differently now. Z.AI is the listed comp, and it's cratering. You can't price pre-IPO paper off a story when the story just lost 60%. If AI IPO Watch helps you stay ahead of the market, take a moment to subscribe and leave a review wherever you’re listening. It really helps other investors and AI-watchers find the show.
Next, we're watching Moonshot's planned August pre-IPO fundraising discussions — the next checkpoint on its path toward a Hong Kong listing.
You’ll find links to every story we covered today in the show notes, so if something caught your ear, you can dig into the source material there. That’s AI IPO Watch for today. This is a Lantern Podcast.