Since its listing, Keep has seen its market value evaporate by over HK$2.1 billion, its monthly active users have halved from their peak, and its cumulative losses have exceeded RMB 6.9 billion. The "first stock of sports technology" is currently navigating a period of significant distress in both its capital standing and its business operations. As of September 1st, its market capitalization has dropped below HK$800 million, marking a decline of over 96% from its initial listing period. On that day, its stock price fell sharply by 10.71%, hitting an all-time low. This once-celebrated company, which attracted tens of millions of users with the motto "Self-discipline gives me freedom," is now mired in a business model paradox. Its online membership revenue is shrinking, while its own branded sports products are being forced to take on the role of primary revenue driver, yet this puts it in fierce competition with traditional industry giants. Despite loudly proclaiming an "All in AI" strategy, its research and development investment has been decreasing for years, and this new narrative struggles to mask its underlying decline. As specialized apps face pressure from both mega-platforms and AI assistants, Keep's space to operate independently is contracting rapidly. Over three years, its market value has dropped by 96%, and liquidity is nearly exhausted. On August 24, 2026, following the release of its interim results, Keep's stock closed at HK$1.76, giving it a market value of only HK$900 million. What does this number signify? On August 22, 2023, Keep's stock price reached an all-time high of HK$42.40, corresponding to a market value of HK$22.3 billion. In just three short years, HK$10 billion in market value has vanished, a 96% erosion. By September 1st, the stock had further declined to HK$1.49, an all-time low, representing a nearly 95% drop from its initial public offering price of HK$28.92. The more dangerous signal is the drying up of liquidity. Over the past 30 trading days, Keep's total trading volume was a mere HK$12.7562 million, with an average daily turnover of just HK$425,200. The interval turnover rate was 1.39%, with a daily average of 0.05%. For a listed company valued at under HK$800 million, such thin liquidity means any moderately large sell order could trigger a collapse in its share price. The capital market is voting with its feet, and the number of votes is dwindling. Keep's predicament is not unique; the dividend period for the entire online health sector has subsided. Codoon's financing stopped in 2021, and Joyrun's last funding round was in 2018. Aside from Keep's struggles on the secondary market, most online health platforms have already faded from the capital market's view. Peloton, the pioneer of internet fitness, has seen its market value shrink to US$2.4 billion, a 95% retreat from its historical peak. Niche apps are being squeezed from both sides - on one hand, super platforms like Douyin, Bilibili, and Xiaohongshu offer vast amounts of free fitness content; on the other, AI assistants can generate personalized plans with a single prompt. Not long ago, Ant Group strategically invested in薄荷健康, becoming its largest external shareholder with a stake exceeding 28%. 薄荷健康 secured funding and ecosystem support by leveraging its nearly two-decade-old Chinese food nutrition database. Keep, meanwhile, holds a similar asset - the exercise behavior data of its 400 million registered users. For major tech companies interested in AI-driven health, "diet-exercise-health" represents a complete closed loop. Could Keep be the next to be "recruited" by a major player? The "first stock of sports technology" is, in fact, becoming an online sporting goods retailer. Keep went public with significant historical losses. From 2019 to 2022, cumulative net losses over four years reached nearly RMB 6 billion, with adjusted net losses of RMB 1.966 billion. In its listing year, it reported a net profit of RMB 1.106 billion due to changes in the fair value of redemption liabilities, but its adjusted net loss was still RMB 295 million. From 2024 to 2025, it incurred losses exceeding RMB 600 million over two years. In 2025, Keep reported an adjusted net profit of RMB 25.216 million, seemingly its first annual profit since inception. However, a closer look raises questions about the quality of this "profitability." First, consider revenue. Full-year revenue in 2025 was only RMB 1.637 billion, a sharp 20.7% year-on-year decline, the largest drop in four years, bringing it back to levels seen in 2021. From RMB 2.138 billion in 2023, to RMB 2.066 billion in 2024, and then to RMB 1.637 billion in 2025, Keep's revenue scale is visibly shrinking. Next, consider the quality of earnings. The turnaround in 2025 was not due to business expansion or efficiency gains, but rather the result of extreme cost-cutting. By the end of 2025, Keep had only 645 full-time employees, nearly halved from the end of 2022. Operating costs, sales and marketing expenses, and R&D expenses were cut by 28.8%, 42.1%, and 29.4% respectively. In other words, Keep "squeezed out" its profit through layoffs, reduced marketing, and trimmed R&D. In the first half of 2026, Keep's revenue was RMB 825 million, a marginal 0.4% year-on-year increase, indicating nearly stagnant growth. Its net loss narrowed to RMB 12.19 million, with an adjusted net profit of RMB 5.877 million, down 21.4% year-on-year, signaling weakening profitability momentum. As of the first half of this year, cumulative losses have reached RMB 6.933 billion. Keep's most fundamental challenge lies in the passive restructuring of its business model. In the first half of 2026, revenue from its own branded sports products reached RMB 483 million, up 21.7% year-on-year, representing 58.5% of total revenue. Meanwhile, online membership and paid content revenue was only RMB 247 million, down 26.9% year-on-year, shrinking to about 30% of the total. From 2019 to 2023, Keep spent four years reducing its consumer goods revenue share to below 50%, successfully listing on the Hong Kong stock market as the "first stock of sports technology." Yet, it took only two and a half years to push this share back up to nearly 60%. A platform that started with online fitness content now derives most of its revenue from selling sports equipment - this makes it not a "sports technology company" but an online sporting goods retailer. What is even more concerning is the profit margin. Although Keep's consumer goods gross margin improved from 34.8% to 40.1%, it still lags significantly behind professional sports brands like Anta (around 64%) and Li Ning (around 51%). The scale disadvantage is even more fatal - Keep's first-half consumer goods revenue of RMB 483 million is less than 2% of Anta's revenue for the same period. High-margin membership business is declining while low-margin consumer goods are expanding, continuously compressing Keep's overall profit space. In the first half of 2026, Keep's overall gross margin was 51.1%, down 1.1 percentage points year-on-year, and its adjusted net profit margin was 0.7%, down 0.2 percentage points year-on-year. User data is perhaps Keep's most alarming indicator. In 2022, Keep's monthly active users (MAU) peaked at 36.39 million. Since then, it has been in continuous decline: average MAU was 29.76 million in 2023, 29.92 million in 2024, plummeting to 21.77 million in 2025, and further dropping to 18.58 million in the first half of 2026. From 36.39 million to 18.58 million, nearly 18 million users have been lost, almost halving the user base. In 2025 alone, approximately 8.15 million users were lost. Over two and a half years, MAU has shrunk by 38%. The company attributes this to a "proactive contraction" strategy - shedding low-frequency, general-traffic users to focus on high-value, high-engagement fitness enthusiasts. Management acknowledged during the earnings call on August 24, 2026, that user losses mainly came from those with weaker exercise intentions, and that the current membership benefits design leans towards content-consuming users, inadequately catering to tool-oriented users for activities like running and outdoor exercise. The "quality over quantity" narrative seems coherent: in the first half of 2026, average revenue per MAU rose from RMB 6.1 to RMB 7.4, a 21.3% year-on-year increase, and average monthly exercise duration per user grew by 15.3%. However, another set of data reveals the flip side: the membership penetration rate fell from 12.4% in the same period last year to 11.7%. This means that even among the retained "high-value users," the willingness to pay is actually declining. Online membership and paid content revenue fell by 26.9% year-on-year, far exceeding the MAU decline of 17.4% - the ARPU increase is more a result of a shrinking denominator than a genuine boost in user willingness to pay. The sentiment from an eight-year Keep user might explain it all: "Opening the app, it's all community recommendations and live-stream shopping. Just finding my own exercise records takes a long time," and courses that were once free now require a membership, even for post-run stretching. When a fitness tool app transforms into a "fitness supermarket," user departure becomes almost inevitable. On the Heimao Complaint platform [download the Heimao Complaint client], Keep has received 35 complaints in the last 30 days, averaging at least one per day, with cumulative complaints exceeding 26,000. Is "All in AI" a new story or a new bubble? In early 2025, Keep announced its "All in AI" strategy. In April 2026, its self-developed vertical large model for sports and health, Keepace.ai, was officially launched, focusing on three core capabilities: course generation, exercise Q&A, and data interpretation. To date, it has launched over 8,000 AI-customized courses, with daily token calls on the app increasing from 8.4 billion in March to 18.5 billion in July. Beneath the surface activity, concerns are mounting. R&D investment has been continuously declining. In 2025, Keep's R&D expenditure was RMB 310 million, a 29.4% year-on-year decrease; in the first half of 2026, it fell another 23.2%. The company explained this in its financial report as "improvements in artificial intelligence technology have increased labor productivity," meaning that AI investment is a replacement expenditure rather than incremental, achieving cost reduction and efficiency gains by substituting traditional content production methods. However, this explanation is itself a warning sign. Keep's "All in AI" is essentially using AI to replace human labor for cost reduction, rather than using AI to develop new revenue growth points. As the capital market's logic shifts from "burning cash for growth" to "valuing profitability," an AI narrative lacking a clear monetization path will only intensify downward pressure on valuations. More importantly, AI is becoming the "gravedigger" for niche apps. When a user wants a fat-loss plan or stretching guidance, their first instinct now is to ask a chatbot, not to download a dedicated app. AI has flattened "niche needs" into a single prompt - this is the true existential crisis for specialized apps. In an era witnessing the collective demise of niche apps, Keep requires more than just a new AI story; it needs a genuinely sustainable business model and sufficient time to prove it. The capital market, however, is unlikely to grant it much time.