30. The Revival of Chinese Stocks and Alibaba’s AI-Stock Story: China Assets Enter a Long-Awaited Era of Confidence
Summary
DeepSeek has repriced Chinese stocks from “discounted assets” back toward “globally competitive technology assets.” Its low-cost performance breakthrough challenged the stereotype that China depends on imported compute; downloads rose 2700% week on week, while Microsoft and Nvidia announced integrations with their cloud platforms. Deutsche Bank called it “China’s Sputnik moment.” Takun agreed with the shorthand that investors should look to consumer in bearish periods and technology in bullish ones: consumer is the base, while technology drives incremental growth.
This rally is stronger than the 2 rebounds in 2024 because policy improvement is finally “echoing” corporate fundamentals. Blinken’s China visit and the “September 24 policy package” produced only modest gains because earnings did not reinforce the policy signal; this time, Alibaba’s net profit was up 239% year on year by Takun’s recollection, cloud revenue grew 13%, and AI-related products delivered triple-digit growth for 6 consecutive quarters. Alibaba’s Hong Kong-listed shares jumped about 15% in a single day—“rare, extremely rare.”
Alibaba is the clearest rerating in this cycle, up about 75% from its trough as its valuation begins to “stand up from the floor.” The market once viewed it as an e-commerce company carrying antitrust risk and valued it on PB; it is now shifting back to PE and could eventually move toward PS as cloud and AI begin to deliver. That migration in valuation methodology says more about changing market confidence than a simple share-price rebound.
Alibaba Cloud, Qwen, and the RMB380B investment planned over the next 3 years form the critical chain connecting the AI story to long-term valuation. The program said roughly 60%-70% of Alibaba Cloud’s compute has gone to inference since the Lunar New Year; broader open-source model adoption drives more inference demand, which benefits cloud services. Takun therefore views Alibaba as “e-commerce plus cloud plus AI,” but with a condition: the market will only truly buy in once it can see the AI business’s “certain end state.”
The core pressure on Chinese stocks over the past several years came not from any single company’s operations, but from uncertainty created jointly by regulation, geopolitics, and capital flight. Didi’s app removal in 2021, the 42% drop in the education index under the “double reduction” policy, and the 24.66% one-day policy selloff showed overseas investors that “a single sheet of A4 paper could kill an entire industry.” Some institutions later observed that long-only capital might be leaving, and overseas firms had “stopped even allocating analysts to research Chinese stocks.” Takun said Chinese stocks were “on the grill” during those years; even excellent companies could not overcome a falling market.
Foreign investors’ first impression of China often comes from stocks they can understand, making Alibaba and Tencent important calling cards for Chinese assets. Takun called the 2 companies the “ballast stones” of Hong Kong equities: when the leaders rise, foreign investors believe high-quality Chinese assets can generate returns; if even those names remain persistently discounted, confidence is difficult to restore. Whether the rally lasts still depends on earnings from the leading companies, but he judged that Tencent’s results are “unlikely to be bad.”
The scarcest asset in this cycle is not any particular model, but the renewed confidence linking private companies, capital markets, and the policy environment. The reported Apple-Alibaba foundation-model partnership, Jack Ma’s frequent return to public view, Alibaba’s large buyback, and AI capex are all reinforcing signals; Takun said Jack Ma visiting Alibaba 3 times in a month did not produce 3 times the confidence, but “probably 9 times.” He closed by saying that “confidence is more important than gold”; the “overweight emerging-markets era” previously described by 卫诗婕 may be returning.
Deep dive
1. Takun’s move from internet product manager to public markets was fundamentally a search for “monetizing insight”
After graduating in 2008, Takun worked as a game planner and internet product manager. He worked at Tencent from roughly 2011 to 2014, moved into the securities industry after 2019, and later joined a brokerage research department. His turning point was not starting with finance, but realizing that employee stock gains could far exceed labor income.
The popularity of League of Legends and Tencent’s powerful share-price gains made him see a positive feedback loop between a company’s products, its earnings, and employee wealth: “It turned out stocks could help employees get rich, beyond their basic compensation.”
The more painful lesson came from a Japanese game he had researched in depth. It became a major success, but he never thought to buy the stock of the listed company behind it. Takun said the company rose 100x during 2013, while its share price gained 114x in a year. His conclusion: “The capital market is, at its core, the process of monetizing your own understanding.”
He opened his brokerage account in 2013, with Perfect World as his first Chinese stock and a gain of nearly 2x. He then shared gaming-industry information during the Chinese internet-stock bull market, gradually becoming a major Xueqiu investor, before choosing the securities industry as the place where his research edge could compound.
2. Everyday observation and “geographic advantage” can create information edges institutions may not have
Peter Lynch’s central lesson for Takun was that ordinary people can look for stocks they can “see and touch” in daily life: everyone plays League of Legends or Honor of Kings, and product popularity may translate into a rising Tencent share price.
He strongly agreed with Lynch’s criticism of investor habits: many people scrutinize small consumer purchases but fail to assess value and growth when buying stocks. The result is that “people are careful with small money when they should be careful with big money, and careful with big money when they should be careful with small money.”
Xueqiu’s early seed users were employees who were “inside the companies.” Tencent employees often understood Tencent’s products and operations earlier than overseas investors who only read financial statements. China’s later adoption of paid business models and its different commercialization paths further widened the local knowledge gap.
Takun cited Qihoo 360’s reversal after being shorted: its business was visible on the computers of huge numbers of Chinese users, but overseas institutions might not have understood it. A short report does not automatically prove that the fundamentals are broken; sometimes it simply exposes the researcher’s unfamiliarity with the company.
3. Shengda Games and Tencent show that information arbitrage and long-term ownership are different skills
When Shengda Games licensed Million Arthur, Takun first watched the Japanese market, then bought after seeing how strongly the game performed in China’s internal testing. Its public launch later validated his judgment. It was a representative case of turning frontline product information into a public-markets trade.
Tencent taught a different lesson: “Hold good companies with conviction for as long as possible.” Using Tencent’s roughly 100x gain from 2004 to 2014, he argued that continued product use, sufficiently long product life cycles, and recurring positive free cash flow are important characteristics of a good company.
Takun also stressed that “what you see now may not be right, but what you saw then was right.” Tencent’s later sharp drawdown showed that even investors with long-term conviction can be forced into liquidation if they bought on margin and the stock falls below their principal.
4. The 2010-2013 Chinese-stock crisis was first a crisis of credibility and institutions
Muddy Waters was founded in 2010 and shorted Orient Paper, while Chinese-stock credibility problems erupted in 2011. Takun said 22 companies were shorted that year, and more than 50 faced delisting or trading suspensions over financial issues. Some companies had in fact committed fraud, causing short reports to trigger immediate, severe selloffs.
The second shock, in 2012, came from the VIE controversy. Chinese stocks used variable-interest-entity agreements to control operating entities in China. Because this structure had not yet been incorporated into the regulatory framework, investors questioned whether companies could be controlled reliably. New Oriental’s related restructuring prompted litigation, the SEC investigated its legality, and US IPOs nearly came to a halt.
卫诗婕 emphasized that VIEs should be understood as a new form of equity arrangement at the time, rather than simply equated with noncompliance. Takun focused instead on their role as “middleware” and the control-rights questions created by routing around existing market rules. Together, the 2 descriptions point to an immature institutional framework.
The period also saw a wave of take-private delistings, with companies including Shanda Networks leaving US markets. More than 40 Chinese stocks delisted over the year. Only after China and the US signed an audit-regulatory memorandum in 2013 and opened limited cross-border cooperation did the conflict over audit papers ease and US IPO activity begin to recover.
5. Standardization and expansion arrived in 2014-2019, but whether Luckin marked a turning point remains disputed
From 2014 through 2019, the regulatory and legal framework for Chinese stocks gradually improved as China’s mobile-internet economy expanded. Alibaba and JD.com listed in 2014, while 44 companies went public in the US in 2018, including new-economy names such as iQiyi and PDD.
卫诗婕 viewed Luckin Coffee’s 2019 listing—18 months after its founding—as a key inflection point, relaying the view of an overseas IPO banker that Luckin’s fraud damaged global perceptions of Chinese companies and pushed Chinese regulators to study an overseas-IPO filing regime. Companies subsequently had to file before listing overseas.
Takun’s rebuttal is worth preserving: Luckin was primarily an isolated case, and its shares even rose after moving to the OTC market. Its severity could not be compared with the macro and institutional shocks that followed. He sees the true cycle as the era of undervalued Chinese stocks that lasted from 2020 through the end of 2024.
6. The 2021 regulatory shock convinced the market that risk could arrive suddenly from outside a company
Takun described the 2020-2024 market as a reflection of China-US relations and the macro environment: “A lot of the time, an individual stock’s fundamentals have less influence than the broader market.” Chinese stocks were “on the grill”; even strong earnings could not reverse the index trend.
In July 2021, Didi’s app removal and the double-reduction policy arrived in quick succession. The education index fell 42% cumulatively, including a 24.66% drop on the day the policy was announced. Overseas investors had not expected “a single sheet of A4 paper to kill an entire industry.” Gaming, e-cigarettes, and Hong Kong and US brokerages were subsequently called in for meetings, and the risk spread into other sectors.
When Alibaba came under antitrust scrutiny, Tencent did not rise as its competitor; it fell. The capital-markets logic was that if the hammer landed on Alibaba, Tencent could be next. The 2 companies competed operationally but were “co-dependent” in capital-market pricing.
Pessimism eventually moved from prices into organizations. 卫诗婕 relayed that Chinese institutions had observed possible withdrawals by US universities, sovereign wealth funds, pension funds, and other long-only capital. Takun’s own experience was that overseas roadshows went from having audiences to institutions “not even assigning analysts to research Chinese stocks”; at times, he could not find anyone to present to.
7. Forced liquidations and capital flight turned “the power of belief” upside down
Tencent had rarely experienced such a deep drawdown, badly hurting employees who had maintained long-term conviction, continued adding, or bought with leverage. Takun explained with a RMB10 initial investment and RMB10 borrowed investment: a RMB20 position that falls 50% leaves RMB10, and once it falls below the investor’s own principal, forced liquidation may follow.
For employees of major technology companies, visible luxury cars and homes became visible wealth destruction. Pessimism was no longer merely a valuation judgment; it became “not believing antitrust could be resolved and not believing policy would improve.” Takun called this “the power of belief turning against you.”
In 2022, the program said implementing rules for the Holding Foreign Companies Accountable Act required foreign-listed companies to increase disclosures, with delisting after 3 consecutive years of noncompliance. The Federal Reserve also turned markedly hawkish, and the China Internet 50 Index fell 41.2% for the year. Downward pressure continued through 2023 and 2024.
8. The 2 rebounds in 2024 were only previews; without fundamentals, they could not become a major rally
The first came in April 2024, when Blinken’s China visit raised hopes of improved China-US relations and pushed KWEB and the Hang Seng Tech Index higher. Musk then visited China, approvals related to autonomous driving were granted, and the German chancellor visited—signals the program treated as a succession of improving-relations indicators.
卫诗婕 also mentioned a medical ship and warships moored along the Huangpu River. Takun speculated that the vessels might have temporarily returned from their prior duty locations for rest and interpreted that as a gesture of goodwill. This was his on-site reading, not a fact further verified by the program.
The second rebound followed the “September 24 policy package,” which sent A-shares and the Hang Seng Tech Index sharply higher in tandem. But both rallies were limited because “policy guidance did not link up with fundamentals.” Geopolitical relations, the exchange rate, the possibility of war, and the policy uncertainty left by double reduction and antitrust measures had not fully cleared.
Takun’s dividing line was clear: policy improvement without earnings support produces a weak rally; if leaders’ financial results strengthen at the same time, politics and economics, and policy and fundamentals, can finally “echo” one another.
9. DeepSeek transmitted a technical breakthrough into capital-market rerating in roughly 50 days
The program used 2 dates for the release of DeepSeek R1. Takun said it was officially released on January 11 and placed the “week of release” between January 20 and 26; later, 卫诗婕 used January 20 as the starting point for the related market interval. Around the event, Nvidia fell 3.12% that day, while DeepSeek downloads rose 2700% week on week from January 20 to 26. Takun said it was probably the fastest-growing app in history and had broken the record.
卫诗婕 said Microsoft and Nvidia quickly announced integrations of DeepSeek into their cloud platforms, providing external validation of its technical capabilities. From around January 26, the Hang Seng Tech Index bottomed and rebounded, while leaders including Alibaba and Tencent gained more than 10% in a week. January completed the first stage: “the technical breakthrough drew global attention.”
February brought “technical deployment and capital rerating.” Deutsche Bank described the Chinese technology breakthrough as “China’s Sputnik moment,” and the market began debating whether the discount on Chinese assets could disappear. DeepSeek’s API price at the end of the month was said by the program to be roughly 1/27 of OpenAI’s, further lowering the barrier to AI adoption.
In March, Tencent, Baidu, and others fully integrated DeepSeek, broadening the valuation recovery. Morgan Stanley forecast that Chinese companies’ AI capex would grow 10% in 2025. Takun thought “that number may still be too low,” as capex shifted from being a headwind to becoming a growth signal.
10. Leaders and indexes rerated before small caps because foreign investors start with Chinese assets they can understand
卫诗婕 summarized the internet sector as “look to consumer in bearish periods and technology in bullish periods,” a framing Takun endorsed: consumer represents existing capacity and the base, while technology provides incremental growth and higher valuation multiples. DeepSeek made the market believe again that Chinese internet companies still had a technology identity.
This rally began as a broad advance in the Hang Seng Tech Index and KWEB, with many small caps failing to outperform the indexes. Alibaba opened sharply higher on February 3, the first trading day after the Lunar New Year, and became a leading indicator. The start of the year also coincided with a window of global capital reallocation, dense domestic policy announcements, an earnings blackout period, and active thematic investing.
When foreign capital enters China, “its first look should often be at stocks,” with priority given to leaders whose business models are easy to understand. Takun believes returns from leading stocks often represent the return on a country’s assets, which is why Alibaba and Tencent serve as the “ballast stones” of Hong Kong equities.
卫诗婕 questioned whether Morgan Stanley’s public bullishness on China was credible given its economic interests. Takun responded that sell-side research and asset-management divisions are separated by a firewall and cannot influence one another, so research views cannot simply be equated with portfolio positions.
11. US restrictions and tariffs may not directly end the rally; secondary markets still price returns
On the “America First Investment Policy” memorandum signed by Trump, Takun judged the impact limited. The restrictions discussed by the program focused on primary markets and investments in critical technologies; they did not directly prohibit US capital from buying Chinese stocks in secondary markets.
His underlying logic is that the dollar has global infrastructure and a profit-seeking nature: “As long as Chinese assets offer good returns, cheap valuations, and high certainty,” onshore or offshore dollars may still flow in. The key variable is not the slogan, but whether the fundamentals continue to deliver.
On US tariffs imposed across the world, Takun offered a relative-pricing framework: if the cost of trading with the US rises for other countries, China’s relative trade cost may fall, and China may gain “more friends.” The program used that logic to explain several days when global equities fell broadly while Chinese assets rallied, but presented it as a market interpretation rather than a firm prediction.
12. Alibaba’s core change is a valuation shift from PB back to PE, with cloud and AI seeking PS optionality
Alibaba has risen about 75% from its trough, and Takun said its valuation was “standing up from the floor.” PB values net assets, PE values earnings multiples, and PS values sales multiples. Moving from PB to PE and then to PS represents a stepwise increase in the growth elasticity the market is willing to pay for.
After the antitrust campaign, the capital market downgraded Alibaba from a technology company to an e-commerce company carrying policy risk, sending its valuation from PS toward PE and even PB. Takun believes this rally has begun moving PB back toward PE; if cloud and AI deliver, the valuation could eventually shift toward PS again.
Earnings are the anchor for this rerating. Takun recalled net profit growth of 239% year on year; the program said cloud revenue grew 13%, while AI-related product revenue maintained triple-digit, or more than 100%, growth for 6 consecutive quarters. After the earnings-related news, Alibaba’s Hong Kong shares rose about 15% in a single day, a move he stressed was “rare, extremely rare.”
Alibaba’s return from “an e-commerce asset carrying antitrust risk” to a company defined as “e-commerce plus cloud plus AI” is the central story behind the roughly 75% gain. The rally is not merely repairing the price; it is deciding which measuring stick should be used to value the company.
13. Qwen, inference compute, and RMB380B of investment are connecting the AI narrative to cloud revenue
卫诗婕 noted that before DeepSeek exploded onto the scene, the international version of Qwen already had influence among global developers, held the most GitHub Stars among Chinese foundation models, and was open source. Takun later said Qwen had surpassed 100,000 GitHub Stars, exceeding the derivative models of the Llama series. In his view, Qwen had previously been “too open source,” making it difficult for the market to identify a clear commercial innovation.
DeepSeek changed that neglect. Once the market understood that open-source models could also generate massive usage, it rediscovered Qwen’s value. The program said roughly 60%-70% of Alibaba Cloud’s compute had been concentrated on inference since the Lunar New Year: the more open the model, the more it is called, and the greater the demand for cloud inference.
That positive feedback loop explains Alibaba’s decision to invest RMB380B over the next 3 years in cloud and AI hardware infrastructure. The share price continued rising after the plan was disclosed on the earnings call; Takun interpreted the market response as recognition of both the logic and the capex.
卫诗婕 referred to the Apple-Alibaba foundation-model partnership as described by the program at the time and viewed it as a signal of Alibaba’s model, cloud, and compute capabilities. Takun saw it as another important layer of confidence. Alongside DeepSeek, Qwen, and the earnings report, Alibaba became the “most watched” company in this broad Chinese-stock rally.
14. Amazon offers a partial reference point; what the market is really pricing is a calculable “end state”
Alibaba and Amazon both began with e-commerce, and seasonal traffic leaves servers idle outside peak periods, making the path from excess compute to cloud services broadly similar. But Takun explicitly said the 2 companies “will never be completely similar, only partially similar.”
卫诗婕 pointed to Amazon’s cloud profit exceeding e-commerce profit in 2024 and asked whether Alibaba could similarly raise the ceiling for its technology valuation. Takun’s answer was that the market tolerated years of losses at Amazon because it believed it could estimate how much the cloud business would ultimately earn and what share it could capture.
This “end-state valuation” also appears in Tesla. 卫诗婕 said the market might buy in to the auto business, autonomous driving, robots, and even some SpaceX optionality. Takun summarized the current setup as investors mainly buying into Tesla’s cars and autonomous driving, with perhaps some robot exposure as well. The key is whether a business line receives enough weight and forms a market consensus.
Even without antitrust risk, Takun believes the market might not have bought into Alibaba Cloud early, because the business had experienced slowing growth, adjustments, and market-share competition. The AI story must be “worked out mathematically” before it can move from narrative to valuation.
15. Jack Ma’s return, capital-market initiatives, and buybacks together put Alibaba back on the confidence barometer
卫诗婕 viewed Jack Ma’s renewed frequent appearances at company events and government roundtables as a signal that confidence among private entrepreneurs was recovering. Takun went further: “Visiting Alibaba once a month and visiting Alibaba 3 times a month do not generate 3 times the confidence; it should be 9 times,” because the signals compound exponentially.
Alibaba’s historical role also came from capital-market innovation. Its 2014 NYSE IPO raised about $21.7B, with a $68 offer price; after more than 2 hours of bookbuilding, the stock opened at $92.7, up 36.3% from the offer price. An ecosystem partner, rather than an executive, rang the opening bell.
The program also noted that Alibaba completed its conversion to dual-primary listing in August of the prior year and entered Stock Connect, making it easier for mainland investors to participate in its valuation. In 2024, it repurchased more than RMB100B of stock, the largest buyback among Chinese stocks. Takun acknowledged that the buyback produced only a modest lift at the time because the broader market trend had not yet turned.
卫诗婕 asked how long the rally could last. Takun gave a conditional answer: confidence has recovered and uncertainty has been repriced; as long as earnings from leaders such as Alibaba and Tencent continue to improve, the trend may continue. Asked to consider the possibility of weak Tencent results, he judged that they were “unlikely to be bad.”
Sell-side firms have also begun viewing Alibaba through a technology lens. The program said Morgan Stanley raised its target price from $100 to $200, while UBS classified it as an AI stock. 卫诗婕 had previously summarized the return of an “overweight emerging-markets era”; Takun closed with the view that “confidence is more important than gold.”