Short-video leader Kwai (01024.HK) released its fiscal fourth-quarter results for the year ending December 2025 after market close on March 25. The company's full-year revenue and adjusted net profit both rose year-on-year, and its profit figures met market expectations. Its AI business also delivered tangible results. Yet at yesterday's open, Kwai's share price plunged sharply — intraday losses exceeded 14% at one point, marking the largest single-day drop in nearly 11 months and wiping out over HK$10 billion in market value in a single day. Just one day earlier, global AI giant OpenAI announced the full retirement of its video-generation model Sora. With these two events unfolding back-to-back, is the market beginning to judge AI's "compute monetization" ability by an extremely demanding standard?
On March 25, OpenAI officially issued a statement that it would fully discontinue Sora across the board, covering the standalone app, API endpoints, and the ChatGPT-integrated video generation feature, leaving no operational foothold. Once hailed as a milestone in AI video, the product lasted only six months from launch to shutdown. Analysts point to an imbalance between cost and monetization as the root cause of its collapse. Research firm SemiAnalysis estimated Sora's daily operating costs at as much as $15 million, implying annualized burn exceeding $5.4 billion, while revenue was almost negligible. Even more damaging was extremely poor user retention: Andreessen Horowitz (a16z) reported a 30-day retention rate of only 1% for Sora and a 60-day retention rate that fell to zero. Against the backdrop of IPO preparations and tightening resources, OpenAI decisively cut loss-making businesses and redirected focus to faster-monetizing areas such as enterprise tools and core large models.
Regarding Kwai's stock plunge, some analysts note that although its in-house video-generation model can accelerate commercialization — generating quarterly revenue of RMB 340 million and reaching an annualized run-rate (ARR) above $300 million by January 2026 — and management has set a target to double revenue in 2026, the market remains worried about high AI investment and long-term profit pressure. That concern has prompted capital to abandon tech assets characterized by heavy up-front spending and slow payback.
Institutions and analysts weigh in densely: the AI industry says goodbye to wild growth
After the events unfolded, domestic and international brokerages and senior analysts spoke up in quick succession, and the tone was broadly cautious to pessimistic. Several major foreign investment banks swiftly cut Kwai's price targets: Jefferies lowered its target from HK$106 to HK$82, and CLSA cut its target from HK$85 to HK$68.
CICC (China International Capital Corporation) noted that Kwai's steep share decline essentially reflects the market's re-pricing of internet companies' AI investments. Capital is no longer willing to pay for vague concepts; the pace of earnings realization, cost-control ability, and the quality of revenue growth have become core evaluation criteria. Kwai's AI R&D spending continues to rise, squeezing short-term profit margins, and with 2026 revenue growth expectations revised down to below the industry average, the firm cut its earnings forecasts and target price.
Nomura downgraded Kwai to Neutral and substantially trimmed its price target. The bank believes OpenAI's shutdown of Sora signals a strategic contraction across the global AI industry, exposing the monetization difficulties and poor retention of consumer-facing AI products. The AI video space where Kwai is deepening its presence faces both homogenized competition and high compute-cost pressure, increasing the risk that commercialization will lag expectations, compounded by weakness across Hong Kong tech stocks and strong willingness among investors to exit.
Jaseper Tsang, vice chairman of the Hong Kong Institute of Financial Analysts and Professional Commentators Ltd., commented that the market has already grown worried about the outlook for AI development among mainland internet giants — prompted by earnings guidance from Tencent, Alibaba and others. The industry's large-scale AI investment has become severely unbalanced between input and output: persistent cash burn with meager returns has intensified selling pressure. Forrester's vice president and chief analyst, Dai Kun, said that AI video generation consumes far more compute than text products, making it difficult in the short term to build a stable profitability model; OpenAI's shutdown of inefficient services is a strategic choice under constrained resources.
Veteran industry analysts added that Kwai's sharp drop also reflects the mood across Hong Kong tech stocks. The industry is moving beyond a phase of reckless growth; capital's frenzy is fading, and attention is returning squarely to commercial fundamentals.
OpenAI's suspension of Sora and Kwai's post-earnings share collapse has sounded a warning for the AI industry and capital markets. AI technology is no longer the market's "miracle cure"; even the flashiest AI models must pass the test of "compute monetization." For investors, this may be a painful moment of valuation correction back to rationality: only companies that build a complete commercialization loop and balance costs with revenues will continue to win capital recognition.
(Reporter: Kato Ip)
Comment