On July 17, Hong Kong-listed AI stocks continued their downward adjustment. Zhipu (2513.HK) saw its share price plunge sharply during the session, at one point dropping nearly 15% to HKD 1,316. The Hang Seng Tech Index opened slightly lower by 0.05%, with AI application stocks broadly under pressure. Since hitting an intraday high of HKD 2,222 on July 9, Zhipu has experienced a maximum drawdown of over 40% across six trading days. Measured from its all-time high of HKD 2,980, the cumulative drawdown exceeds 55%.
This round of declines is not an isolated event. Since July, Zhipu has faced the expiration of cornerstone shares, a massive placement, and extreme price volatility—plunging over 16% intraday on July 2, surging nearly 15% on July 8, and rallying more than 20% intraday after the placement announcement on July 9. Multiple single-day swings exceeding 10% in a single month reflect a rapidly widening divergence in market valuations for this leading AI stock.
Rapid Early Gains Lead to Valuation Overextension and Profit Taking
Since its IPO at HKD 116.2 in January 2026, Zhipu’s share price soared to a record high of HKD 2,980 in less than six months—a cumulative gain of more than 24 times. This surge was driven by concentrated market pricing around favorable domestic AI policy, expectations for AI Agent commercialization, and rising enterprise demand for AI applications.
However, after the rapid rally, the gap between valuation and fundamentals has come under renewed market scrutiny. Zhipu currently trades at a price-to-earnings ratio of -152.80, reflecting its ongoing loss-expansion phase. When Goldman Sachs initiated coverage on July 10, it assigned a "Neutral" rating and set a 12-month target price of HKD 1,880 based on discounted cash flow. This target represents a significant discount to the historical high, indicating that even under optimistic scenarios, investment banks believe previous valuations have fully—or even excessively—priced in growth expectations.
As the market shifts from a "storytelling phase" to a "performance phase," AI companies typically see their valuations recalibrated. Zhipu’s earlier rally was built on long-term narratives like domestic substitution, technological breakthroughs, and market share expansion. However, these stories must translate into verifiable financial results to support lofty valuations. The recent volatility essentially reflects the market’s tug-of-war over this transition.
HKD 31.4 Billion Placement: The Double-Edged Sword of Capital Raising and Share Dilution
Capital raising is another key variable affecting Zhipu’s share price. On July 9, Zhipu announced a placement of up to 19.78 million new H-shares at HKD 1,588 per share, aiming to raise approximately HKD 31.41 billion. The placement price represents a discount of about 12.99% to the July 8 closing price of HKD 1,825, with the new shares accounting for roughly 4.25% of the enlarged share capital. This marks the largest single placement fundraising by a Hong Kong tech company in 2026.
The fundraising comes amid a much faster-than-expected cash burn rate. As of June 30, 2026, Zhipu had already used over 93% of its IPO net proceeds. Capital expenditure for large model companies is accelerating—requiring massive investments in computing power procurement, GPU cluster construction, model training, and talent R&D.
The net proceeds of about HKD 31.375 billion from this round will be allocated to three main areas: foundational model R&D and computing infrastructure, commercialization and strategic investments, and optimizing capital structure and working capital. The company expects to utilize all proceeds by the end of 2027.
While capital raising itself is not inherently negative, the market’s concerns are intensifying on two fronts: first, the short-term dilution effect from new share issuance—per-share earnings are diluted and supply increases in the near term; second, the AI business model still requires ongoing validation—questions remain around investment levels, revenue generation, and the timeline to profitability.
A noteworthy detail is that this placement occurred just one week after the expiration of Zhipu’s cornerstone lock-up. In early July, the company marked six months since listing, freeing up 25.68 million cornerstone shares for trading. Launching a major placement immediately after the lock-up expiration sparked debate over "unlock and fundraise" timing. Although six institutions fully subscribed to all placement shares, the short-term increase in tradable shares still pressured the stock. On July 14, just one trading day after the placement was finalized, Zhipu’s intraday low fell to HKD 1,473—already below the placement price of HKD 1,588.
AI Foundation Model Competition Enters the "Knockout Stage"
Another source of valuation pressure is the evolving competitive landscape. China’s AI foundation model sector is shifting from a "technology race" to a "commercialization race." Zhipu faces competitors such as Alibaba’s Qwen, Baidu’s Wenxin, ByteDance’s Doubao, DeepSeek, and MiniMax.
The key competitive variables are changing. Previously, model parameter size and benchmark scores were the core market focus; now, commercialization capability has become the new battleground. Goldman Sachs divides China’s foundation model development into two phases: 2025 as the "DeepSeek Moment"—where the industry leverages MoE architecture to sharply reduce inference costs, with cost efficiency as the core advantage; and 2026 as the "Zhipu Moment"—when GLM 5.2 joins the global top tier in code, agents, and other high-value tracks, proving the global competitiveness of domestic models. After technological breakthroughs, the real test is converting these advantages into sustainable commercial revenue, which will determine long-term value.
Meanwhile, foundation model companies are shifting from traffic competition on the consumer side to value validation—after user base expansion, the next challenge is boosting willingness to pay and improving business models. The industry sees 2026 as the "deep water zone" for large-scale model deployment.
Changes in primary market valuation logic reflect the same trend. In 2026, institutional due diligence shifted from parameter counts and leaderboard rankings to monthly operating income, amortized computing costs, and paid user conversion rates. Projects unable to articulate a clear commercialization path are finding it difficult to secure new funding.
Valuation Logic Shifting from "Tech Leadership" to "Revenue Validation"
Zhipu’s future valuation reset hinges on answers to three core questions.
First, can AI models generate sustainable revenue? Current data show positive signals for Zhipu’s commercialization progress. In Q1 2026, API prices rose a cumulative 83%, yet call volume increased 400%, and MaaS platform ARR reached RMB 1.7 billion. The platform now serves over 4 million registered enterprises and users across more than 218 countries and regions. However, Goldman Sachs projects that API and subscription revenue for Chinese AI models will grow from RMB 35 billion in 2026 to RMB 879 billion in 2030. Achieving this scale means Zhipu must keep expanding market share amid fierce competition.
Second, can computing power costs be effectively reduced? The core challenge for foundation model companies is the race between revenue growth and rising computing costs. GPUs, servers, and cloud resources make up the largest fixed costs. JPMorgan raised Zhipu’s target price from HKD 2,000 to HKD 2,400, believing that the HKD 31.4 billion raised will effectively ease the company’s computing power bottleneck. However, the payback period and ROI for these investments remain key market concerns.
Third, can an ecosystem moat be established? Future competition will not be just about models, but about the integrated strength of "model + agent + enterprise application + developer ecosystem." In an internal letter, Zhipu CEO Tang Jie stated, "As the industry accelerates commercialization, we have decided to push upward," committing to strategic investment in next-generation AI model R&D. Whether this long-term strategy can win short-term capital market recognition remains highly uncertain.
Market Signal: The AI Rally Isn’t Over, but Pricing Rules Have Changed
The sharp adjustment in Zhipu’s share price signals not a disappearance of AI demand, but a shift in capital market expectations—AI companies must now prove their ability to commercialize.
From a broader perspective, Q2 2026 saw global AI concept stocks come under collective pressure. On July 17, amid global "AI fatigue," South Korean stocks fell 6.4%, Japanese stocks dropped 2.8%; in the US, TSMC’s earnings reignited concerns over AI spending, triggering a tech sell-off. Zhipu’s correction is part of a global AI sector revaluation, not an isolated case.
However, structural support remains in China’s AI market. Goldman Sachs has launched a tradable "China AI Value Chain" basket, forecasting that API and subscription revenue for Chinese AI models will grow about 25-fold over the next five years. OpenRouter reports that token consumption by Chinese AI models such as DeepSeek and Zhipu has already surpassed US competitors. The long-term drivers—growing enterprise AI demand, domestic AI infrastructure build-out, and expanding application scenarios—remain intact.
Zhipu’s future hinges on one core variable: whether technological leadership can be converted into sustainable commercial leadership. Capital markets are voting with their wallets—not denying AI’s value, but demanding a clearer commercialization path.
FAQ
Q: What were the direct causes of Zhipu’s (2513.HK) share price drop on July 17?
On that day, Zhipu’s shares fell nearly 15% intraday to HKD 1,316. Immediate triggers included profit taking after an extended rally, short-term supply pressure from the HKD 31.4 billion placement, and broader weakness across global AI stocks. The company also completed its English name change to "Z.AI" the same day, but this did not support the share price.
Q: Is the HKD 31.4 billion placement positive or negative for Zhipu’s share price?
The capital raise provides funding for R&D and computing expansion, which is necessary for long-term growth. However, the market is focused on short-term dilution (about 4.25%) and the placement discount (about 13%), and the placement came right after the cornerstone lock-up expired, raising concerns about increased share supply. On July 14, the share price had already fallen below the placement price of HKD 1,588.
Q: Why did Zhipu and MiniMax’s share price performances diverge?
After Zhipu’s 5.76% lock-up expiry in early July, its share price surged 13.66%, while MiniMax’s 48.9% lock-up expiry led to a 17.98% plunge. The difference stems from the lock-up ratios, fundraising structures, and differing market expectations for each company’s commercialization prospects. JPMorgan raised Zhipu’s target price to HKD 2,400, while cutting MiniMax’s to HKD 240.
Q: How is the competitive landscape in China’s AI foundation model sector changing?
Competition is shifting from a "model capability race" to a "commercialization capability race." In the primary market, valuation logic has moved from parameter counts and rankings to monthly operating income, amortized computing costs, and paid user conversion rates. Projects without a clear commercialization path are struggling to secure funding.
Q: How should Zhipu’s long-term investment value be assessed?
It depends on three core questions: whether its AI models can generate sustainable revenue, whether computing costs can be effectively reduced, and whether it can build an ecosystem moat. In Q1 2026, Zhipu’s API prices rose 83% cumulatively while call volume grew 400%, showing some pricing power. However, Goldman Sachs maintains a "Neutral" rating, and the market remains divided on valuation and profitability prospects.
