Summary

风叔’s core call for 2026: reform will be driven by opening up. “The most important part of opening up is services, and the most important part of services is finance,” because across manufacturing, economic scale, finance, ideology, and military capabilities, finance is “probably the only area where the US still has a clear lead.” The key question is where, in which industries, and under what institutional framework financial opening will be piloted; he guesses capital-account opening and the supporting institutions may come “within a five- to six-year timeframe.”
The US is dismantling the dollar’s grand cycle itself. The structure consists of a military foundation, a middle layer of “globalized capacity plus deficits” and liberal market economics, and a top layer of the dollar financial empire; the US now wants to keep only the top and bottom, but “that should be very difficult” when viewed through the circulation of money. China’s response comes down to 2 unavoidable moves: RMB internationalization (“there is absolutely no suspense; it is the only choice”) and gradual capital-account opening, with the hardest work being institutional and legal. Changes in foreign-exchange reserves also partly answer claims that China’s $1T trade surplus in 2025 did not translate into an equivalent increase in reserves.
The year’s boldest prediction: the Hong Kong dollar could break its dollar peg in 2026 and move to a currency basket, with a meaningful RMB weighting. The window would require expectations of lower US rates and a weaker dollar alongside expectations of RMB appreciation, similar to the mainland’s 2005 exchange-rate reform. 李翔 objected on the spot: “I personally think it may not happen in 2026—or even in the next 3-5 years.”
Hainan’s customs-closure policy follows the logic of a repeat of Shenzhen 40 years ago. Shenzhen began as a trade experiment but unexpectedly incubated nationwide transfers of land-use rights and real estate; Hainan now combines free-trade-port status with preferential individual and corporate income-tax rates. High-margin, high-wage services should be the most tax-sensitive, and “financial trials may also start here.”
US equities in 2026 face 3 variables: the rate-cut cycle, the AI bubble, and the midterms. Money could flow out of the dollar assets that were “overweight to an extraordinary degree” in 2022-24. The historical pattern is that every bubble ends with an IPO by its hottest company—the internet in 2000, Alibaba in 2007, Blackstone in 2008, Uber/Airbnb in 2015. OpenAI’s reported $800B fundraising valuation is “unheard-of”; OpenAI, Anthropic, and SpaceX will probably see at least 1 go public, but “money has not yet exited on a large scale—the final step of the cycle has not clearly appeared.”
Chinese assets have 2 clear sources of incremental demand. Funds moving out of the 3-year time deposits opened at high rates in 2022-23 will mature progressively from the second half of 2025 onward; ETFs probably exceeded RMB5T last year. The second wave is insurers’ long-duration capital after the adjustment to equity-asset risk coefficients. Insurers are already appearing frequently in Hong Kong IPO and A-share GPU allocation lists, which should reduce volatility in new-share speculation. Foreign reallocation depends on macro stabilization: overseas investment banks cluster around 4.3-4.8, while domestic institutions see 4.7-5.4.
The AI endgame is cloud-based foundation models and compute, with software and hardware integrated. “These mega-companies may have a huge advantage”—Microsoft, Google, ByteDance, Alibaba—while startups could repeat the consolidation of the IaaS cloud industry a decade ago. Across 3 waves, the US leads in models, China and the US are roughly matched in agents and robotics, and China may gradually gain an edge at the application layer where AI actually makes money. “The only outcome nobody can predict is OpenAI”: nobody can afford to buy it, so it has to go public before the window closes.
The biggest and most distant structural issue is that data is the ultimate factor of production in the AI era. “Every problem ultimately comes back to data.” China has established a central data-management body and is piloting public-data programs in Shanghai, including the desensitization, cleaning, and pricing of hospital clinical data for different users. The US “at least for now has not had time to deal with it”; from 2026, the 2 countries will “go their separate ways” on this issue.

Deep dive

1. Opening Signal: Reform Through Opening Up, With Financial Opening at the Core of 2026

风叔 leads with the conclusion: in 2026, “the more visible direction should still be using opening up to promote reform.” The hierarchy is clear—services are the newest and most important area of opening, and within services, “finance is by far the most important,” because across manufacturing, economic scale, finance, global values, and military capabilities, finance is “probably the only area where the US still has a clear lead.” After the direction of financial opening was clarified in 2023, “2026 is the best—or at least a reasonable—time to implement it.”
There are 2 concrete points to watch: the pace, method, and speed of financial opening; and how Hainan’s policies evolve in 2026—whether Hainan’s opening could, in some form beyond the free-trade port, be combined with financial opening as a pilot. Islands and closed-off areas are the easiest places to experiment. When 李翔 asked, “We thought the same thing in 2021,” 风叔 drew a distinction: in 2021, China had only brought the pandemic under control, and “nobody knew what the outcome of the pandemic would be”; now, “the most turbulent phase appears to have temporarily passed.”
On the Venezuela discussion from the previous episode, many people on social media said, “The US has just shown China how it’s done.” 风叔 takes the opposite view: what matters more now is demonstrating that China’s way of acting is different from America’s. “In the short term, it may look as if you’re taking a loss, but in the medium term this may be a particularly rare and valuable window of opportunity.” Even if some invoke the law of the jungle, “ultimately it still comes down to where people’s hearts and loyalties lie.”
One addendum to the previous episode’s property discussion: developers do not want to carry leverage for long, and the development cycle has compressed to 2-3 years from land acquisition to launch. Land supply has fallen by “more than half” since the second half of 2022, so new-home supply over the next 2 years “will decline quite noticeably”—another factor helping the supply side of the supply-demand rebalancing.

2. The US and China Swap Scripts: Washington Embraces State Mobilization While Reversing Course on Openness

The past 2 years have produced “a great many unexpected similarities”: the Trump administration subsidizing strategic industries, taking equity stakes in companies, having the Department of Energy organize AI-industry companies to jointly formulate development plans, imposing Chinese-style price controls on biopharma, subsidizing chip manufacturing, and even “starting to use a state-mobilization model similar to China’s to build warships.” 风叔 connected this to a biography of ASML he had just read, recalling Dutch government support for ASML around 2000. 李翔 added that subsidies in the style of Japan’s and South Korea’s ministries of international trade and industry have always existed—“otherwise these mega-companies would never have formed.”
But on “openness,” the 2 countries are moving in opposite directions: the US restricts visas while China expands visa-free access; the US chooses petrochemical energy while China chooses new energy; the US raises tariffs while China cuts tariffs for developing countries; the US exits international organizations while China pushes them to keep operating. The underlying logic is: “He is no longer the biggest beneficiary, so he is done with it”—Washington wants to rewrite the rules and build a new system, while China has become a larger beneficiary and therefore wants to keep using the existing one. China’s opening also reflects its lack of channels for international narratives; visa-free access addresses the problem that “seeing is believing,” letting people discover that “you may actually be better than the impression everyone has of you.”
A worthwhile theoretical detour: 李翔 invoked modern monetary theory to explain why US spending far exceeds income. 风叔’s response was that “economic theories are born in response to economic needs”: Adam Smith emerged from Britain’s needs, while Keynes was in some sense ahead of Roosevelt’s New Deal. The central government’s large-scale leverage build-up only saw a sharp change in slope during the pandemic; by November 2020, the US had already completed the equivalent of 5 years of post-financial-crisis quantitative easing, expanding its balance sheet by more than $3T.

3. The Dollar Cycle as a Layered “Pie”: The US Wants to Keep Only the Top and Bottom

风叔 explains the post-1980s order with a layered “pie.” At the bottom is military power; above that is the globalization of production and demand plus US deficits, which send dollars abroad; in the middle is the liberal market economy promoted jointly by the US, the UK, and Margaret Thatcher, centered on reducing intervention in exchange rates and cross-border capital flows. At the top sits the financial empire: “creating enough financial products and derivatives—most of them dollar-denominated assets”—to pull the dollars back in. The cycle depended on the collapse of Bretton Woods and petrodollars; before World War II, crises were more often caused by regional excess capacity that could not be absorbed globally.
The combination the US wants today is internally inconsistent: retain military influence, albeit over a shrinking footprint because “it’s too expensive”; preserve the global reach of dollar financial assets; slash the trade deficit through broad tariffs; and abandon pure liberalism. “They originally put everything on 1 tray, with a layer of cream between 2 layers of biscuits—that was coherent. Now they want to keep the bottom and top layers they care about most, but they no longer want the middle.” Cutting the deficit means fewer dollars held by the rest of the world, which “should be very difficult” from the perspective of money circulation.
China entered this cycle through WTO accession: mandatory foreign-exchange conversion made RMB issuance dependent on the surplus, and avoiding depreciation of its dollar holdings required buying dollar assets. This also partly answers claims that China generated a $1T surplus in 2025 without a commensurate increase in reserves. China does not want the US to succeed in keeping only 2 layers, so it is buying gold, diversifying reserves, and “doing everything possible to diversify the countries and regions that make up its surplus”—supply-chain restructuring is part of the response. The remaining question is how, from 2026 onward, China can “build a financial cycle that is more favorable, relatively safer, and more diversified for you” within the broader financial system.

4. China’s Answer: RMB Internationalization Is “Without Any Suspense”; Legal Infrastructure Is the Hard Part

The logic begins with 3 common-sense questions. Does China need a system relatively independent of the dollar cycle, one that makes Chinese assets and the RMB safer? Yes. That makes 2 things unavoidable: RMB internationalization—“if only you use the RMB globally, you have no chance… there is absolutely no suspense; it is the only choice”—and gradual capital-account opening. Internationalization can come first and be advanced through China’s trade position: after China became Argentina’s largest trading partner in 2025, the 2 countries moved to dual-currency settlement, “because its currency is unstable, so settlement is mainly in RMB.”
The hardest part is the third item: institutional and legal construction—a view 风叔 has held for years that many people do not agree with. “Money moves too quickly and in too great a volume,” making property protection, asset security, and ownership determination critically important. The precedent is the capital-account pilot in 2014-15, after which “the 2015 stock-market crash happened.” The 18th Party Congress report explicitly called for “orderly opening of the RMB under the capital account.” 李翔 drew a parallel with the unified national market, which is also held up by the “people” in the free movement of people, goods, and capital—household registration. “We have spent a very long time thinking through both of these hurdles.”
The methodology is “from invitation-based access to registration-based access,” alternating between left and right feet: one foot promulgates or revises institutions, while the other pilots opening, “to see whether the system is reasonable and complete enough… then revise the system again.” A typical invitation-based move came at the 2023 Central Financial Work Conference, when a payment-clearing license was offered to a US company—roughly China’s third. Insurance, securities, and funds now permit 100% foreign ownership, but whether “you can apply and receive the license once you meet certain conditions” remains to be seen. Financial services that were once protected by licensing barriers “will gradually lose those barriers.” Recommended reading: 李晓’s Double Shocks, a microcosm of the evolution and competition of Chinese and US finance.

5. The Year’s Boldest Call: The Hong Kong Dollar Breaks Its Peg for a Currency Basket—李翔 Isn’t Buying It

Before the capital account opens, “Hong Kong is certainly a window for China and an asset-trading center”—it has a common-law system and no capital controls. 风叔’s conjecture is that in a 2026 window of lower US rates and dollar-depreciation expectations, the Hong Kong dollar could, like the mainland’s 2005 exchange-rate reform, move away from its 100% dollar peg and switch to a currency basket, with a meaningful RMB weighting. “If you do this during a dollar appreciation cycle, pressure on the Hong Kong dollar could be very high.” That means choosing a window of dollar weakness, which happens to coincide with expectations of RMB appreciation, RMB internationalization, and financial opening.
The logic is that Hong Kong trades RMB-denominated assets for overseas purchasing power, so the Hong Kong dollar should be “pegged to a foundation closer to the RMB.” The signs include the large volume of RMB assets, offshore RMB quotas, and digital-currency experiments that the mainland has provided Hong Kong in recent years. 李翔’s immediate objection is worth preserving: “That’s an extremely bold idea… It may not happen in 2026; it may not happen even in the next 3-5 years.” 风叔 leaves it open: “We can watch and see, because this is a rather unusual window.”

6. Hainan’s Customs Closure: A Trade Pilot Could “Accidentally” Incubate Financial Opening, Just as Shenzhen Incubated Real Estate

Yangpu Port is the obvious card: it affects routes to Singapore and through the Malacca Strait and would be most helpful to China’s maritime trade with Southeast Asia. Freight had already grown sharply in the second half of 2025 before the customs closure, and some traffic could be diverted from Shanghai and Zhoushan. By contrast, Gwadar Port “has taken shape, but still lacks systemic connections”—land-sea intermodal transport must cross a plateau and requires other countries to co-build the network, while Hainan “is within China’s own sphere of control.” Another new factor is the Arctic route: global warming and stronger vessels mean “ships using the Arctic route today necessarily pass Greenland,” which is partly related to the US interest in Greenland.
The tax policy is the deeper signal: preferential individual and corporate income-tax rates, plus an exemption from import duties when goods are produced locally in Hainan and achieve 30% value added. Services are relatively high-end, labor-intensive, and high-value-added. High gross margins make corporate income tax important, while hiring high-paid talent makes individual income tax important. “Services should be the sector most eager to receive simultaneous individual and corporate income-tax benefits”; its sensitivity should be the highest.
The historical parallel is the most vivid part of the discussion. Forty years ago, Shenzhen used Hong Kong’s advantages to pilot processing trade; Hong Kong entrepreneurs needed factories, and “in the end, looking at the outcome, almost all of China’s real estate and transfers of state-owned land-use rights… began in Shenzhen.” Shenzhen’s pilot was never designed for real estate. Hainan is likewise offering free trade and tax benefits, but if new demand happens to arise in finance and certain service industries, “our pilot may also start there.” Hainan has already run stand-alone pilots for innovative drugs and therapies.

7. 3 US Equity Variables: Rate Cuts, the AI Bubble, and the Midterms—Bubbles Historically End With a Leader IPO

The first variable is the rate-cut cycle, with implications for potential dollar depreciation and capital outflows, especially after global money “overweighted dollar assets to an extraordinary degree” in 2022-24. The third is the midterms: if Republicans lose the House, “people will again start anticipating challenges to policy execution and major changes in policy further down the road,” with an “enormous” impact on US equities. 风叔 also observed that Trump has refocused on issues close to home, such as Venezuela and Greenland; “in some sense, this may be because the tariffs he spent a year pushing did not generate enough political capital.”
The second variable, the AI bubble, follows a historical pattern: the peak of every boom comes when the most important company in that field is about to go public or has just completed its IPO. That means the 2000 internet bubble, Alibaba in 2007, and the 2015 sharing-economy boom, when Uber and Airbnb “some went public and some did not.” 李翔 added private-equity IPOs as a signal, pointing to Blackstone in 2008. This year, OpenAI, Anthropic, and SpaceX could raise enormous amounts of capital, and “there is a high probability that at least 1 of these 3 companies will go public this year.” SpaceX is not a conventional AI company; it is primarily commercial space. OpenAI’s reported $800B valuation for a financing round is “basically unheard-of… approaching a $1T market capitalization.”
The 3 variables are linked: “Private markets are still too flush with money… global money is overweight dollars, and even if it needs to reallocate, large-scale reallocation has not yet begun.” “Money has not exited on a large scale—the final step of the cycle has not clearly appeared.” As for what breaks the bubble, that cannot be predicted: it is “generally not something like Nvidia’s earnings suddenly collapsing,” but some unrelated event that unexpectedly triggers selling, “and then it comes crashing down.”

8. Chinese Assets: Deposit Migration and Insurer Long-Duration Capital Provide 2 Clear Inflows; Innovative Drugs Have Passed the Darkest Phase

The framework for foreign reallocation is straightforward. Before 2022, 2 questions were uncertain: China’s willingness to compete, through reform and opening, and its ability to compete with the US technologically. “The answers are basically certain” on both; the remaining question is when the macroeconomy stabilizes. The forecast dispersion itself is informative: overseas investment banks cluster around 4.3-4.8 or 5, while domestic institutions see 4.7-5.3 or 5.4. Foreign investors “are on the relatively bearish side of the quadrant, so they need more definite positive signals before bringing money back and reallocating it.” 李翔’s counterpoint is worth keeping: “For an economy as large as China’s, anything between 4 and 5 is remarkable.”
Domestic liquidity has 2 clear sources. Money was placed in 3-year deposits during a window in 2022-23 when “the interest rate was inexplicably quite high”; those deposits mature progressively from the second half of 2025 through 2026, and “there is no way all of it can roll back into time deposits,” so some will inevitably diversify. The second source is incremental insurance allocation after the financial regulator adjusted the risk coefficient for insurers’ equity assets at the end of last year. Deposit migration is already happening: China’s ETFs “probably exceeded RMB5T” last year, after the first major ETF expansion in 2024 helped stabilize the stock market to some extent.
Long-duration capital is changing the IPO ecosystem. In the past 2-3 months, insurers have appeared much more frequently in allocation lists for major Hong Kong IPOs and newly listed A-share GPU companies. Insurers have long-duration liabilities and generally plan to hold for longer, reducing the pool of shares available for short-term trading before lock-ups expire. Previously, large pools of Hong Kong capital would subscribe to IPOs and “sell as soon as trading opened,” or companies would issue only 5% of their shares to the public and then push them higher before selling into the unlock. “The first wave was ETFs; the second wave is probably long-term capital such as insurers.” The stabilizing effect should be especially significant for A-shares.
Innovative drugs delivered a full-year data point. China’s pipeline licensing contracts exceeded RMB130B in 2025. If global licensing was around $200B in 2024 and did not change materially in 2025, China’s share “basically exceeded 50%.” But upfront payments were under RMB10B, roughly 5%, still half the typical 10% ratio for buying late-stage pipelines. The reasons are that the pipelines skew early-stage, multiple rights are sold in bundles, and intense domestic competition gives buyers bargaining power. The conclusion: “At a minimum, biopharma had already passed the hardest stretch during 2025—and proved its efficiency.”

9. 3 Waves of AI Evolution and the Cloud-Based Endgame: “The Only Outcome Nobody Can Predict Is OpenAI”

The 3-wave framework is clear. In the first wave, the US is plainly ahead in foundation models. In the second, general-purpose agents on the software side and humanoid robots on the hardware side are areas where “China and the US each have their strengths”; “at a minimum, you cannot say the US is clearly ahead.” The third wave is using AI to “actually make money” across thousands of industries. History suggests that, from facial recognition to autonomous driving, “China will gradually begin to have an advantage at the application layer.” This year’s secondary-market test is “who has both an AI narrative and can genuinely use AI to generate profits.” ByteDance is the exception: it is “making serious money,” with reported net profit of RMB40B this year, and is one of the few Chinese media companies with a global footprint.
风叔 stresses that the cloud-based endgame is his personal view: foundation-model services become cloud, compute becomes cloud, and software and hardware integrate. Alibaba Cloud is moving from selling compute to selling “compute plus Qwen token calls,” so “these mega-companies may have a huge advantage.” Microsoft, Google, and ByteDance “should all be among the most competitive looking forward.” Startups face a difficult path, just as infrastructure cloud a decade ago eventually became a market where “cloud was simply infrastructure, and only a few of the largest players remained.” 李翔 recalled 王坚’s early view: cloud vendors would either build their own foundation models or bind themselves tightly to model companies. The precedent from listed non-mega-cap cloud companies is “a rapid rise during the concept phase followed by a long, slow decline.”
Microsoft is a case study in understanding the endgame. From operating systems onward, “it caught every bubble”: it moved early on browsers but was late to search, where network effects meant “once you lose it, you cannot catch up.” When 李翔 asked whether Gates attributed that to antitrust, 风叔 replied, “Yes.” SaaS, collaboration software, and cloud all made the first tier. Foundation models “do not appear to have obvious network effects”; with capex and R&D, “Microsoft should eventually catch up.” The only real suspense is OpenAI: it is not a composite of hardware cloud and model cloud, “and certainly nobody will be able to afford to buy it.” It must “get listed as soon as possible before the window closes,” while what happens after the IPO remains to be seen. Musk represents another path: avoid large-scale cloud services and take vertical applications to the limit—autonomous driving and robotics—while X.AI could still integrate with the social platform X.
Primary-market enthusiasm is rotating, with a clear dividing line. Models were hottest in 2023-24, even though some investors “refused to believe in them”; robots were hottest in 2024-25 and are now “in the second half of their hype cycle,” with fundraising frequency falling sharply. Over the past quarter, smart hardware has taken over, including the wearables, pet devices, cameras, and drones showcased at the ongoing CES. For model companies, the test case is the market: 智谱 is already listed, MiniMax is close, and 月之暗面 (Kimi) has just raised a large round at roughly $3B-plus. “Assuming 智谱 and MiniMax perform reasonably well after listing,” Kimi’s path is sustainable; otherwise, “it will have to find every possible way to list as soon as possible.” Wealth effects will keep capital flowing into AI, but toward vertical applications that previously received little funding rather than foundation models.

10. The Biggest Long-Term Issue: Data Is the Ultimate Factor of Production, and China and the US Will Diverge

风叔’s closing point is that “looking at every problem today, it ultimately comes back to data.” Strip away compute chips and the same applies to finance, models, healthcare, and vertical applications. Chip bottlenecks will ease as applications differentiate: PC gaming created Nvidia, smartphones created Qualcomm and Broadcom, and “once AI reaches applications, chips will certainly diversify for a period of time.” Data presents 2 questions: the degree and scope of digitization, which determine scale and quality; and how data is moved, priced, desensitized, and managed. Unlike land, data “can be used by me, by you, by Henan, and by Hubei.” Establishing ownership and pricing therefore becomes both the biggest source of competitive advantage and the largest bottleneck. China has established central institutions and begun pilots; the US “at least for now has not had time to deal with it.” “Starting in 2026, this will begin to send the 2 countries in different directions. It is the most long-term, but also the biggest, issue.”
Shanghai’s public-data pilot should be an important 2026 theme. The example is pharmaceuticals: after pipeline licensing began to work, Chinese drugmakers became like manufacturing entrepreneurs after 2001—“How can I make more? How can I make products different from everyone else’s? How can I sell them in more places?” Producing more, faster, and better pipelines requires clinical data from hospitals, but no single company can negotiate with hospitals one by one. In the US, “one [data record] is roughly $100 to several hundred dollars.” Shanghai wants to extract, desensitize, clean, and structure hospital data, then work through the data costs and business models for different users, including drugmakers, insurers, and early screening and diagnosis. Financial data is broader and more sensitive. 李翔 noted that protection is currently poor and personal data can be accessed if one pays. 风叔’s view: “As long as data can make money, someone will eventually sell it.” That requires law externally and management and pricing mechanisms internally. Individuals as data providers “will probably ultimately benefit too,” but the mechanism is highly complex.
Data endowments will drive divergence among the giants. WeChat began with walkie-talkie-style voice messaging, then shifted toward text to avoid disturbing people; its latest versions are heavily promoting voice-to-text input—“it is certainly because of foundation-model capabilities.” Tencent has “the best data” in contextual text and voice corpora, but “all of this data is highly tied to personal privacy, and that is the biggest problem.” Alibaba’s e-commerce, transaction, and consumer-intent data is “unmatched.” ByteDance has the best image and video data and “may do better in multimodal.” 李翔 relayed a friend’s observation that ByteDance’s Doubao speech-recognition accuracy is “far, far higher than WeChat’s.” The differences in what the 3 companies do best “may ultimately determine that their trajectories over the next few years will not be completely identical.”
The same applies to agents, where the primary market has no consensus. An agent “is essentially an application of a somewhat compressed model,” and its viability depends on 3 things: enough data from vertical scenarios, existing customers, and digitized processes. The companies that already have all 3 are often established mid-sized businesses: “You simply lack the technology to turn it into an agent.” Startups such as Manus—and a Baidu-related project whose name was unclear, possibly Genspark—have to build all 3 from scratch. Silicon Valley is also seeing a new structure: Meta and Google, including through deals such as Google’s purchase of Character AI, are no longer buying companies outright; they pay licensing fees for IP, lock up the data and team, and leave the shell operating, “to prove that I did not buy the company” and avoid antitrust scrutiny. Cash is shifted forward to founders and the earliest shareholders, which is “not particularly friendly in terms of financial returns for financial investors.” 风叔 admitted, “I never encountered this when I was investing in Silicon Valley.” Whether Manus uses the same structure is only speculation, and China has not yet developed this type of return mechanism.

Are We Wrong About A.I.? | Clip How Ladder Became #1 Strength Training App