Vol. 209 Macro Talk 102 | The AI Shift and Capital Flows Under Global “Risk Off” (Recorded Feb. 27)

Summary

• Feng Shu’s overarching read: the two-month software selloff, Nvidia falling 5%+ in a day after a strong report, Walmart breaking the $1T market-cap mark, and the crypto rout are not separate stories; “everyone is just finding different explanations for phenomena caused by changes in liquidity”—at root, global capital is Risk Off on dollar assets. Peel the onion from the outside in: crypto first, then software applications (down 30%-50%+ from their highs, with Oracle around -50% and Duolingo down more than 50%); among the Magnificent Seven, only Nvidia is still trading relatively near its highs.
• Cisco in 2000 is Nvidia’s cautionary precedent: the world’s largest company then was worth less than $600B and fell more than 80% after the bubble burst; Nortel’s potential customers dropped from 4,000-5,000 to 300-400 within a year, while Michael Burry says Nvidia has nearly $100B in long-term supply-chain commitments, close to 10x the level 18 months ago—the same underlying logic. “If it also gets peeled off the onion, that means this wave has reached the core.” Feng Shu does not want a repeat of 2001, but if it happens, history says “when one whale falls, everything else comes alive”: capital gets reallocated and other markets can return to growth.
• The first money out of the US is short-term, fast money, now racing across emerging markets—from South Korea’s memory-chip trade and Samsung Electronics/SK Hynix, to Brazil, where China and Brazil saw the largest relative tax-rate cuts after the tariff ruling, and Japan in rotation. Hong Kong stocks have underperformed for two reasons: the renminbi’s sharp intraday appreciation from 6.87 to 6.83 made Southbound investors prefer A-shares to avoid two rounds of FX spread, while the central bank cut the forward FX reserve requirement ratio the same day to suppress one-way arbitrage; fast foreign money is rotating between hot spots, with one theory holding that Korean retail investors finished trading Hong Kong’s LLM names and went back to Samsung.
• China finally has its own “onion core”: 2 Hong Kong-listed LLM companies, whose market caps briefly approached Baidu’s and exceeded Kuaishou’s, plus 2 A-share GPU companies; with only 5%-10% of their shares freely tradable, they will likely be bid up as the onion core before lock-up expiries. Ordinary investors should stay away: Cambricon fell from roughly a $100B market cap back to $20B and then languished at the lows for about a year after lock-up expiry; anyone determined to buy should wait roughly 6 months after the expiry, until the supply overhang has been absorbed.
• The winners in Chinese AI are Alibaba and ByteDance: China’s weekly token consumption has already surpassed the US, and once LLMs become “cloudified,” users care only about how much each token costs; companies with all 4 layers—chips, cloud, models, and applications—should be formidable. Microsoft “has never missed any of the biggest trends since operating systems” and will catch up; Google has all 4 layers and is still falling, which also looks more like Risk Off sentiment than a fundamental problem.
• China’s macro timetable: PPI and CPI may stop falling year on year as early as mid-to-late Q2, allowing the negative cycle to decouple; the subsequent bottoming and recovery in cyclical, value, and consumer earnings will come later. Feng Shu expects property to stabilize in major cities within the year, and believes Shanghai is essentially the first city after Hong Kong to be on track to stabilize; the rotation within Chinese assets “should happen after Trump visits China, in Q2.” Moutai is the template: baijiu overall fell about 10% over the New Year holiday, while Moutai rose 10%-20% as channel reform released demand for a product people “know they can buy genuine.”
• Three blows in America’s partisan war are attacking each side’s foundations: the ICE incident, with Democrats weaponizing immigration ahead of the midterms; the Epstein case, with Republicans countering on a near “you lose 1,000, I lose 200” basis; and the Supreme Court ruling Trump’s tariffs unlawful—leaving the US with its negotiating leverage against China “pulled out from under it.” On Russia-Ukraine, the US mediation plan is likely a ceasefire along the current line of contact followed by referendums, with Russia receiving at least 2 or 3 of the 4 eastern oblasts; Li Xiang’s warning is worth keeping: this “breaks a very important rule of the game, and things may become chaotic afterward.”
• Germany’s chancellor compressed his China trip to less than 48 hours, while the joint statement separately said that “Germany recognizes Taiwan as an inalienable part of China”—a statement from the world’s third-largest economy signaling that “the US may not directly intervene militarily in the future, and no major power will keep talking nonsense about this issue.” China returned as Germany’s largest trading partner in 2025, with more than €80B in Chinese surplus still to resolve, reinforcing the path of “cautious, gradual appreciation within a certain range” for the renminbi.

Deep dive

1. One explanation unifies everything: global capital is going Risk Off

• The year opened with a pileup of oddities: software stocks collapsed for 2 months as investors blamed AI for “replacing or consuming these software companies”; Nvidia delivered a “pretty good” earnings report but posted its largest one-day decline, roughly 5%, the next day; and Walmart became a $1T company for the first time in its history. Feng Shu offers a completely different answer: these are all the same event—“everyone is just finding different explanations for phenomena caused by changes in liquidity.”
• The framework carries forward the macro discussion from Episodes 98-100: in 2020-21, there was “more money than the entire world had ever seen”; from 2022 onward, that money underweighted China and Europe and went to the US, while higher rates gave capital more reasons to flow there. GPT in November 2022 became “the story line that emerged at the right moment,” and everything began to revolve around it.
• Extreme positioning cannot last: “Money cannot stay at an extreme allocation; it has to be reallocated and re-reallocated.” Risk Off comes in 2 forms: rotating locally from high-risk, high-valuation assets into low-risk, low-valuation assets, and moving money out of the US into relatively cheaper countries and regions.

2. Crypto sits at the front of the wave and gets peeled first

• Crypto has “fallen a great deal” since Q4 last year because it has gone from being “dollar-denominated but not traded only in dollars” to being treated as a dollar asset, especially after Trump’s policies, while remaining “an excessively volatile financial derivative—or what we might call the square of a derivative.”
• That puts it on the outer edge of Risk Off: the onion’s outermost layer is not even software companies. Crypto is the example—“hit particularly hard by Risk Off,” with the sharpest short-term declines and volatility.

3. Peeling the onion: applications lead in Risk On and fall first in Risk Off

• After the AI narrative took off in 2023, US equities first rallied in software and applications—Salesforce and Duolingo, which a junior colleague recommended late last year and which Feng Shu used while traveling with children—before the move reached infrastructure. “Imagine how an onion grows: it grows layer by layer from the outside toward the core.”
• The peeling works in reverse: “The applications that rose first on the AI concept are now the first to fall because of the AI concept.” Software and application names are down more than 30% to more than 50% from their highs: Oracle is down roughly 50%+, while Duolingo, “the application most suited to future AI,” is also down more than 50%.
• Money remaining in the US is rotating between high and low: Walmart, oil companies, and Corning have risen, prompting the market to invent the term “halo”—buying heavy assets and avoiding light-asset businesses for fear that AI will replace them. Feng Shu says that explanation is “completely reasonable,” but the underlying move is still money rotating from AI concept stocks into low-valuation names.

4. Cisco in 2000: the onion core’s cautionary precedent

• At the 2000 bubble peak, Cisco had the world’s highest market cap—the onion core of that cycle, positioned “a little like Nvidia today.” The logic was identical: “No matter whether everyone else is profitable or burning cash, you at least have to spend heavily on the infrastructure—switches and routers.” Cisco, then the world’s largest company, was worth less than $600B. “The OpenAI that hasn’t even listed in the primary market today might be able to crush it in market cap, let alone the Magnificent Seven.”
• The outcome: Cisco survived after falling more than 80% from its peak; Nortel and Lucent “gradually faded.” Nortel’s potential customers fell 10-fold in a single year, from 4,000-5,000 to 300-400. Liquidity was still the backdrop: the euro’s birth in 1999, NATO’s war against Yugoslavia, and the Asian financial crisis drove money into the US, producing the internet bubble’s “final and most intense growth phase.”
• The 2 propositions from that era still hold: the bubble burst, but 20 years later “it still became the ubiquitous internet”; and before this AI cycle, the US did in fact control the global initiative in the internet. A bubble bursting does not mean the proposition was false. Back then, people also said, “This time it really is producing productivity gains.”

5. Nvidia: Burry’s $100B commitment, with the peeling nearing the core

• Nvidia is the only member of the Magnificent Seven “still relatively near the top”; the others are down more than 10% to more than 20% from their highs. Apple had already been criticized for “not being AI,” so its decline has made it relatively safer.
• Feng Shu glanced at Michael Burry’s analysis of Nvidia’s earnings: compared with 18 months ago, the company has nearly 10x the long-term purchasing and supply-chain capital commitments, totaling close to $100B. The logic resembles Cisco and Lucent stuffing inventory into their upstream suppliers; a sharp fall in orders would create enormous problems and challenges.
• Feng Shu’s watchpoint is here: “The peeling process has not yet reached Nvidia. If it also gets peeled off the onion, that means this wave has reached the core.”

6. He does not want a repeat of 2001, but “when one whale falls, everything else comes alive”

• Feng Shu repeatedly emphasizes: “I do not want US stocks to crash, and I do not want a repeat of 2001.” A rapid decline would send capital markets around the world into violent swings. But history has another side: when the US was hit in 2001 and 2008, “the US was not the worst affected; other countries were the ones that suffered,” with Europe bearing the brunt both times.
• After the bubble burst, Europe endured 3 years of one-way euro weakness and China 3 years of deflation: “When one whale falls, everything else comes alive” because money starts reallocating. A slow peel only means “the onion loses its skin”; it does not crash the major capital markets.
• An aside: 2000-01 was the period of Buffett and Greenspan’s “most brilliant performances.” Buffett stuck to his refusal to invest in technology, while Greenspan “successfully substituted a property bubble for the internet bubble” and achieved a soft landing. Buffett’s purchase of Occidental Petroleum 3 years ago was “absolutely not” evidence that macro does not matter; it was 100% macro and geopolitical asset allocation from the US perspective.

7. Money leaving the US is running everywhere: South Korea, Brazil, Japan

• Outbound money follows the onion’s sequence: first to leave is “the most liquid, relatively most speculative—or relatively short-term and fast-arbitrage money”; the longest-duration, largest pools demanding the greatest stability leave last. South Korean equities have surged over the past month on a memory shortage, “a little like Hong Kong for a period last year”; Brazil has risen rapidly in parallel, while Japan has also been traded in the short term on the back of political stabilization.
• Brazil’s “reason” comes from a tariff comparison: after the tariff ruling was combined with Trump’s additional 15%, the 2 countries with the largest relative reductions in average tariff rates were “indeed China and Brazil.” Feng Shu’s view is unchanged: set aside the ex-post explanations and assume that the first batch of money going Risk Off out of the US will simply run everywhere.

8. Intraday renminbi appreciation: the central bank moves to suppress one-way arbitrage

• In the first 2 or 3 trading days after the New Year holiday, the renminbi saw “a rapid short-term intraday appreciation”—from 6.87 to 6.83 in a single day, a move of more than 0.1, “relatively rare in FX.” For large pools of money, 0.1 is a substantial arbitrage opportunity. With the Hong Kong dollar fully pegged to the US dollar, that created one-way arbitrage and expectations of rapid appreciation.
• On the morning of the recording, the central bank sharply cut the reserve requirement ratio for forward FX sales, explicitly seeking to “dampen this large-scale one-way arbitrage.” The renminbi’s path is “cautious, gradual appreciation within a certain range,” helping address both deflation and China’s purchasing power globally.
• The contrast is striking: last year, “countless investment banks” said the renminbi would depreciate to 7.5 or even 8, prompting exporters to retain foreign currency offshore rather than convert it. Now fast money is driving appreciation expectations, while arbitrage positions and unconverted export proceeds are rushing to convert on the fastest possible route.

9. Why Hong Kong stocks alone fell: Southbound FX math and foreign fast-money hot-spot rotation

• There are 2 pools of money in Hong Kong stocks. Southbound trading crosses 2 FX gates. If investors expect rapid near-term renminbi appreciation, “I might as well buy only renminbi-denominated A-shares, so I can completely avoid these 2 conversion and FX problems.” That explains why A-shares were relatively unaffected while Hong Kong stocks saw exceptionally high volatility within a week.
• The foreign pool is short-term arbitrage money that first went Risk Off in dollar assets and is now “chasing short-term hot spots and rapidly swapping liquidity.” A report half a month ago said Korean retail investors were pouring into Hong Kong’s LLM stocks; yesterday’s story was that the money had gone back to trade Samsung and Hynix. Feng Shu says, “I think we should just listen to that as a joke”—some of it may be true, but directionally it is the same fast money moving between emerging-market hot spots.
• The third layer is sentiment contagion: Chinese internet stocks were “to some extent” swept up by the AI Risk Off move that began in US equities 1 or 2 months ago. But Chinese internet valuations are not high overall, so “the first factor”—FX—is the larger influence.

10. China finally has an onion core: 4 new companies

• China previously had no Magnificent Seven-style core—“at most, there was Cambricon”—which is why Cambricon was once bid to an absurd level. Over the past 2 months, 4 new candidates have appeared: 2 Hong Kong-listed LLM companies, MiniMax and Zhipu, whose market caps briefly approached Baidu’s and exceeded Kuaishou’s, plus 2 A-share GPU companies.
• Why not Alibaba? First, it is too large to move easily; second, it is dual-listed, and its large US trading volume means that “if even Google is falling, the US-traded portion will at minimum materially drag you down.” ByteDance is unlisted and cannot be traded; Tencent has not positioned itself around this theme. That leaves the 4 names.

11. The lock-up-expiry wealth code: 5%-10% free float means a brief frenzy and a predictable fate

• None of the 4 companies has reached its lock-up expiry—6 months for Hong Kong stocks and 1 year for A-shares. Only the small amount issued at listing, roughly “5%-10%,” is freely tradable. The limited float makes them easy to manipulate, and “before they unlock, I suspect they will be bid up as the onion core a couple of times.”
• The advice to ordinary investors is unequivocal: “As an ordinary person, you are better off not touching them.” The eventual drawdown depends on 3 factors: how far the US onion has been peeled by then—“if Nvidia falls, it will basically pull everyone down”; Hong Kong’s history of stocks falling back toward or below their IPO prices at lock-up expiry, even when they are good companies; and selling pressure from financial investors.
• Cambricon is the ready-made template: it listed at roughly a $100B market cap, fell at its worst to $20B, then “lay at the lows for almost a year” after lock-up expiry. Only after investors reduced their positions and the free float was broadly distributed did it begin a new rapid rise with this AI cycle.

12. The math of selling pressure: the primary-market dam breaks into the secondary market

• New Hong Kong listings are “basically the dammed-up lake of primary-market investments from the past 5 years.” The arithmetic is simple: assume a $10B market cap and financial investors holding 40%—many companies are at 50%-60%. After lock-up expiry, “the market needs new money equal to another 40% of the market cap to absorb the stock.” That is a very large amount and difficult to take; there are too many Hong Kong-listed companies, and there cannot be enough money to absorb every unlock.
• Secondary-market institutions are “also very shrewd money.” They may push up and accumulate the IPO float, with an average cost potentially at a $100B-plus valuation, and need the stock to reach $200B to double their money. They may not want to use $6B to buy old shares from you when your cost was $300M, because “the moment you sell, the stock will collapse back.” So as expectations build, everyone starts selling before the unlock date, and then starts selling even earlier.
• There are 2 practical points. Anyone who wants to buy can skip the first 1 or 2 quarters after the unlock and wait roughly 6 months, for the midpoint when the market has absorbed the supply; “that is usually when it is relatively cheap.” The exception is a company with “very strong financing capacity”—one that can raise substantial capital every 3 or 4 months or every 6 months at prices above its IPO or previous financing round, perhaps by bringing in Middle Eastern money. “It will move up the steps little by little.” Li Xiang laughed: “I know which company that is.”
• A-share supply clears more slowly: investors who are also directors, supervisors, or senior executives face quarterly and monthly limits on selling. “You can never finish selling.” Feng Shu has resigned from one or two companies for that reason.

13. China’s AI winners: Alibaba and ByteDance, because of “cloudification”

• Li Xiang asked directly: setting interests aside, which Chinese AI giants does Feng Shu favor? His answer: “Actually, it is these 2—Alibaba and ByteDance.” They have huge AI applications at the front end, cloud, and LLMs; Alibaba has explicitly announced chips, while ByteDance has only market rumors. The 4 layers are essentially connected.
• A small news item used to trade the A-share and Hong Kong power sectors contains the key: China’s weekly token consumption has surpassed the US. The broader implication is to set aside model quality and benchmark gaming. At the application layer, users “do not care which model the back end uses for me; I care only about the cost of doing this.” That cost is tokens: “How much did I actually spend to get this done?”
• The analogy is the first cloud boom of 2014-15. Everyone talked about superior architecture and vertical use cases; by 2017-18, when cloud usage began to grow, most people did not care about those things. They cared who could provide it “more cheaply, quickly, and reliably.” The company with all 4 layers wins. That was Amazon’s path: use a sufficiently large application base to support its models and compute cloud, then extend them to others. Cloud companies focused only on data centers will ultimately differ on technology iteration speed and cost.

14. Will Microsoft and Google fall behind? Feng Shu says no

• Google has all 4 layers—TPU, cloud infrastructure, Gemini, and front-end applications through YouTube and Search. It “should have strong accumulated capabilities at all 4 layers, with the ability to connect them and deliver services.” “Almost nobody is talking about Perplexity today.” Its continued decline is “quite strange” and looks more like Risk Off than a fundamental problem.
• A debate with an Alibaba executive is worth recording. The executive said Microsoft had fallen behind in this cycle; Feng Shu pushed back: “In every major trend in history since operating systems, Microsoft has never missed.” It caught up in collaboration software, IM, search, enterprise cloud, and big data; OpenAI was also its first investment in a major tech company. “So it can absolutely catch up in the end.” It is being hit by Risk Off because it is both a software company and not the innermost onion core.
• The same applies to Alibaba: from roughly 170 to 150, it is down by the low-teens percentage in this cycle, for the same reason as Microsoft and Google. A very large, dual-listed company is highly exposed to US trading sentiment. Li Xiang observes: “Every time everyone thinks it should be doing well, it doesn’t; then when everyone thinks it is not doing well, it does.”

15. The tech gala and robots: expectations are the hardest thing to manage

• Feng Shu originally expected the robot frenzy to fade. Last year, nobody knew robots would appear at the Spring Festival Gala, so the performance materially exceeded expectations; DeepSeek, a little Ne Zha, and the symposium with private-sector entrepreneurs then reinforced one another and pushed the theme higher. This year, “everyone knows robots will appear, so it is almost impossible to exceed expectations”—after prolonged pre-event promotion, the final product is unlikely to be universally acclaimed.
• Unitree’s performance “slightly exceeded my own expectations”: multiple robots coordinated with people through precise positioning. The shaking in the legs may actually show that the system first became unstable and then rapidly stabilized, adjusting its posture in real time. Once the hardware is good, the algorithmic layer is “not that extraordinarily difficult,” though getting the overall scheduling, positioning, and precision right remains hard.
• The hedge remains in place: “I am only half-convinced that this can sustain the robot frenzy.” The social-media view that it is “technology for technology’s sake” is itself evidence that expectations are fully priced in. Germany’s chancellor also visited Unitree on this trip.

16. China is also going Risk Off: Moutai’s counterintuitive data

• China’s market is beginning the same rotation between high and low: “chemicals and cyclicals, plus a little consumption—that is a very clear asset rotation.” Feng Shu disclosed a position but said he was discussing only the phenomenon, not recommending the stock. Before the holiday, “everyone in the secondary market was adamantly bearish on baijiu… only Duan Yongping was adamantly bullish on Moutai.” By early January, everyone thought Duan was trapped. “We weren’t—we only bought in early January.”
• Consensus was half right. Baijiu consumption overall fell roughly 10% year on year during the New Year holiday, as expected. But statistically, “Moutai increased by 10%-20%.” The reason was channel reform Feng Shu had discussed 4 years ago: shifting from distributors to direct sales, direct sales to e-commerce, and e-commerce to self-operated channels.
• Three data points support the case: i Moutai added several million new registered users before the holiday—free registrations whose customer-acquisition value would be enormous if bought as traffic—and more than half of new buyers were born in the 1980s or 1990s; restaurants saw an extremely high Moutai opening rate during the holiday; and the demand came from certainty—“people used to be unsure whether they could buy a genuine bottle.” A distributor’s old remark captures the problem: in some years, the amount of Moutai sold with invoices was far greater than the amount leaving the factory. As Deng Wenbin asks, “Will young people never grow old?”

17. From elite mystique to the family table: 3 stages of consumption

• Moutai’s consumption setting has changed 3 times. More than 10 years ago, it belonged “only to leaders and certain closed circles” and carried a sense of mystery. After the sharp fall following the 8-point regulation, it became “regular medium- and high-end business liquor.” Now it is shifting again to “family or ordinary social consumption,” with social media filled with bottles opened while returning home, drinking with parents, and gathering with friends. The parallel is familiar: 3 or 4 years before the pandemic, Lululemon and Arc’teryx were worn mainly by the upper-middle class; now ordinary people in first-tier cities wear Descente as normal sportswear.
• Price-band compression is reshaping the market: Fenjiu below 200, Wuliangye at 500-800, and Moutai formerly above 2,000 left a 1,000-1,500 gap that had been “empty for a long time” and is now being aggressively contested. Once channels opened up, Moutai became available at 1,500-1,700, so “the 1,200-1,300 price point is at least gone in the short term.” During the holiday, only Fenjiu, Wuliangye, and Moutai were flat or slightly up year on year, with Moutai posting the largest increase.

18. New Year consumption met the state’s objective, but was not a recovery

• Total tourist trips and total spending both rose year on year—nearly 20% and more than 15%, respectively—but the main reason was 1 additional holiday day. Per-capita spending was “only a few yuan below last year—this year it was another 3 yuan lower.” Feng Shu’s verdict: “It achieved the state’s objective, but you still cannot say consumption has clearly recovered.” With a longer holiday, “you are not going to spend for all 9 days.”
• Li Xiang added his own experience: apart from a handful of leading companies performing very well, most listed consumer companies delivered disappointing revenue and profits. “The mass-consumption market remains extremely, extremely challenging.” Feng Shu agreed.

19. The deflation decoupling timetable: PPI/CPI in Q2, property stabilization within the year

• China’s biggest macro challenge consists of 2 mutually reinforcing problems: deflation, meaning CPI is flat or falling year on year, and property “stopping its decline and stabilizing—not rising.” They map to household and economy-wide balance sheets, consumer confidence, and supply-demand equilibrium. Historically, when PPI and CPI fall together, “roughly 60%-70% is external factors” in the roughest sense; those external factors then reinforce the remaining 20%-30% of domestic causes, locking the cycle in place.
• The timetable is specific. Foreign investors expect property to stabilize in 2027; Feng Shu believes “property will stabilize in major cities within this year,” with Shanghai essentially the first city after Hong Kong to be on track to stabilize. CPI needs the negative cycle to stop intensifying and external demand to improve; “that factor appears almost in place by Q1 or early Q2.” Renminbi appreciation supports asset prices and purchasing power over the medium term, but “it is not immediate.”
• His forecast, explicitly a personal view and not investment advice: excluding the New Year effect, “by Q2, either in the middle or toward the end—perhaps a little earlier,” PPI and CPI will stop falling together year on year. Last year’s low base helps, and PPI may escape faster when viewed through the middle and upstream sectors. Once the negative cycle decouples, earnings bottoming and recovery in midstream cyclicals, value, and consumer stocks will be a lagging result—not a precondition for decoupling. The rotation from high to low within Chinese assets “should become clear after Trump visits China, in Q2.”

20. America’s 3-hit partisan war: ICE, Epstein, and the Supreme Court

• Feng Shu calls his interpretation “a little conspiratorial.” The ICE incident in Minnesota in January was amplified into protests across the US, “just as Black Lives Matter appeared at a particular stage of an election.” He believes it was almost certainly the Democrats targeting Trump’s weakest points—immigration, high tariffs, and the return of manufacturing—with the midterms as the objective.
• Republicans countered with the Epstein case. The disclosures mainly involved Democratic-linked officials, the president of Harvard, “this important businessperson who just came from China,” and a former president about to testify. The tactic was “you lose 1,000, I lose 200,” accepting damage to America’s political and official image as the price of retaliation.
• Then the Supreme Court ruled Trump’s tariffs unlawful, “returning a counterattack to the Republicans at nearly twice or several times the force.” Although Republican-appointed justices are a majority and Trump participated in 3 nominations, “there must be some result of partisan conflict.” The practical consequences are severe: before the most important negotiations with China, the US has to manufacture leverage from nothing each time, and suddenly “someone pulled the leverage out from under it.” The damage to US debt, fiscal policy, and credibility is also significant. The overall judgment: “Their partisan conflict has reached the point where it is affecting the foundations of both sides,” going somewhat beyond Feng Shu’s prior expectations.

21. Why Russia wants the Donbas whatever the cost

• Feng Shu’s synthesis after reading 7 or 8 geopolitical books by Wang Wei, the lead writer at Guigu Studio, has 5 parts. Eastern Ukraine was originally allocated to Ukraine by the Soviet Union, with Crimea transferred later in 1954. The Soviet planned economy’s industrial division of labor never anticipated the breakup, so removing eastern Ukraine left Russia’s industrial system incomplete for building aircraft carriers and large aircraft. Yet Russia was the only country in the world that could be fully self-sufficient in resources: it has roughly one-fifth of the total supply of every resource the world needs, more than the US. NATO expansion left the great plain without strategic depth; Russia had historically relied on distance and winter to withstand Napoleon and Hitler. Donbas, Zaporizhzhia, Kherson, and other areas once accounted for much of Ukraine’s GDP and nearly all of its industry, much of which is now “blown to pieces.” Finally, the territory protects Crimea’s Black Sea access and permanent naval base, with eastern Ukraine lying between them.
• Feng Shu labels his view of the US mediation plan “a wild guess, completely irresponsible as international politics”: declare a ceasefire along the current line of contact, then hold referendums, with Russia receiving at least 2 of the 4 oblasts, perhaps 3. The US has already secured the minerals agreement and “does not really care who gets those 4 oblasts”; it wants some form of weak cooperation with Russia, “at least in some sense, an alliance.”
• Li Xiang’s pushback is worth preserving: “If it really strips certain areas of eastern Ukraine away and gives them to Russia, it breaks a very important rule of the game. Things may become chaotic afterward… Nothing like this has happened since World War II.”

22. The mirror of oligarchization and the challenge of a unified great power

• Ukraine’s path was rapid and violent privatization after the breakup, followed by corruption, oligarchization, and the transfer of large volumes of state assets to individuals; the oligarchs then returned to politics. Before Zelensky, Ukraine’s presidents were all economic oligarchs—“the chocolate king and so on.” Russia followed the same shock-therapy path. Yeltsin actively elevated Putin, after which they agreed never to pursue Yeltsin or his family retroactively. Putin then fought corruption, arrested oil and gas tycoons—possibly Khodorkovsky, “the best-looking one”—and forced media tycoons into exile in Britain, while bringing part of the energy lifeline back under state control: “Otherwise, it would have no foundation as a country.”
• Feng Shu’s biggest takeaway for China is that “China has experienced every problem Russia experienced, good or bad; it just happened that China never swung into the uncontrollable phase.” Russia took every issue “to 10,” while China corrected course in time 20 or 30 years ago. Even the sequence and trajectory are somewhat similar. The 2 books most relevant to China are the one on China’s surrounding geopolitics and the one subtitled The Oil Wars.
• Unity is China’s trump card against offshore balancing: “For an offshore-balancing country like the US, China’s biggest problem is that it is a unified country.” Europe can be fragmented, reunified, and sent back to war—Britain could even be pulled out of the euro and the European Community through Brexit. “You cannot hold a referendum in California while sitting in the US; California would certainly become independent.” Dismantling the Soviet Union made the problem much easier for the US.

23. Germany’s chancellor visits China: one Taiwan line carries 3 signals

• The China-Germany joint statement contains a standalone sentence: “Germany recognizes Taiwan as an inalienable part of China.” Its weight comes from Germany’s position as the world’s third-largest economy, “probably ahead of Japan.” The signal to people on the island is that, first, the US may not directly intervene militarily in the future; second, no major power will keep talking about the issue. Japan is the contrast: because it spoke about Taiwan, “we can directly sanction it.”
• The economic backdrop can be stated in 2 lines: China returned as Germany’s largest trading partner in 2025, implying that the US held the position in 2024, with energy likely a major factor; China runs a surplus of more than €80B with Germany that still needs to be resolved. That reinforces the logic of moderate renminbi appreciation and stronger Chinese purchasing power globally.
• History rhymed with 1999. Schröder visited China during an economic challenge, compressing the trip to 28 hours, and brought the largest wave of China-Germany commercial cooperation 5 months later. This time, Germany faces a huge number of major economic and political issues at home yet still compressed the trip to less than 48 hours. “The history is very similar at this point.” Germany’s complexity is that it is both Europe’s leader and a defeated power: it lacks France’s military-industrial system and relies on external sources, especially Russian energy. Its position on Russia-Ukraine is therefore “as complicated as France’s, but more complicated”; Britain is the most resolute because it has already left the EU, and the more Europe is thrown into turmoil, the better for it. The China-Canada joint statement was also a signal to “a member of the core American sphere.”