Zhipu, the Beijing-based AI model developer that trades in Hong Kong as Knowledge Atlas Technology and markets its products internationally under the Z.ai brand, launched a $4 billion secondary share placement on Wednesday at a price of HK$1,588 to HK$1,698 per share – a discount of as much as 13% to Wednesday’s close, representing 19.8 million shares. The timing is not accidental. July 8 marks the expiration of the six-month lock-up period for the company’s cornerstone investors from its January IPO, meaning a large tranche of shares held by institutional backers who subscribed at HK$116.20 per share became freely tradable for the first time on the same day that the company announced the secondary offering. The stock has risen approximately 1,500% since listing. NEWSCENTRAL reads the share sale as a rational capital-raising exercise by a company that has generated exceptional paper wealth in a very short period and is now seeking to convert some of that market valuation into actual capital before conditions change – but the 13% placement discount on a stock that has risen 15 times in six months is worth examining carefully.
The commercial case for raising $4 billion is straightforward: Zhipu remains deeply unprofitable, with net losses that widened in 2025 as the company invested heavily in model development and compute infrastructure to remain competitive with DeepSeek, Moonshot AI, and the international frontier labs its founder has explicitly benchmarked against. Revenue in the trailing twelve months was approximately $101 million – a figure that implies a revenue multiple against the current market capitalization that cannot be sustained without a dramatic acceleration in commercial performance. The company’s GLM 5.2 model achieved rapid enterprise adoption after its June launch, landing within a percentage point of Anthropic’s Opus 4.8 on one agentic benchmark at roughly one-fifth the cost, and has been credited with capturing a growing share of enterprise workloads that were previously routed to American frontier models. That commercial momentum is real. Whether it is sufficient to justify a valuation of approximately $120 billion is the question the secondary offering forces into the open.
The pricing of the secondary offering at a meaningful discount to the market price reflects the institutional reality of placing $4 billion of shares in a short window. Cornerstone investors who bought at HK$116.20 in January are sitting on returns of more than 1,000% at the placement price floor of HK$1,588, making the discount from current trading levels operationally irrelevant to them. The question for incoming investors is different: they are being asked to pay HK$1,588 to HK$1,698 for a share in a company that generates $101 million in annual revenue and has widening losses, at a valuation that exceeds every Chinese technology company except the largest established platforms. Freddy Miller, Senior Analyst at NEWSCENTRAL, notes that the placement will reveal the depth of institutional conviction in the Chinese AI model story at current valuations: oversubscription at the floor price would confirm that global institutional investors have concluded the commercial trajectory justifies the multiple; a need to reprice below the range would suggest the opposite.
The broader context for Zhipu’s capital raise includes significant competitive capital formation at its peer group. Moonshot AI, which produces the Kimi models, is reportedly raising at least $1 billion before pursuing a public listing. DeepSeek, the most globally recognized Chinese AI model developer, reportedly closed more than $7 billion in funding. The capital race among Chinese AI developers is intensifying at the same moment that their models are gaining meaningful share of the enterprise workloads previously dominated by American providers. Zhipu’s decision to convert its extraordinary stock rally into capital now, before any deceleration in the Chinese AI investment cycle, is consistent with the management philosophy of any company that has watched similar rallies unwind in adjacent technology markets.
The timing of the placement also aligns with a specific competitive dynamic. The six-month lock-up expiration on July 8 was publicly known, and the secondary offering was widely anticipated. A company that did not raise capital on or shortly after lock-up expiration would have faced sustained selling pressure from cornerstone investors seeking to realize gains, potentially producing a disorderly decline in the absence of a managed placement. The $4 billion offering at a 13% discount is, from that perspective, an orderly management of a supply event that was going to occur regardless. To NEWS CENTRAL, the real test is what the company does with the capital – whether it accelerates the commercial trajectory that justifies the valuation, or primarily extends the runway of a deeply loss-making business at a multiple that assumes profitability will eventually follow.
Zhipu’s Commerce Department blacklisting – the company’s listed entity remains on the trade restriction list – has not visibly deterred institutional participation in its Hong Kong listing or in the secondary offering. The practical consequence of the blacklist has been primarily operational for U.S.-headquartered enterprises evaluating whether to use Z.ai’s models directly, rather than for international institutional investors participating in a Hong Kong capital markets transaction. NEWSCENTRAL considers that distinction increasingly important to track: the regulatory and market investment dimensions of the Chinese AI story are operating on different timescales, with capital markets responding to commercial momentum while regulatory restrictions address operational risk in an environment where both are simultaneously intensifying.