Author: Xinhuo Technology
Mr. Fu Peng, Chief Economist of Xinhuo Group, was invited to participate in Wiki Finance EXPO Hong Kong 2026 and deliver a keynote speech. Drawing from a global liquidity framework, Mr. Fu shared his core insights on current global asset classes and market trends, and provided a systematic analysis of the underlying logic of the crypto market.
Fu Peng, Chief Economist of Xinhuo Group, delivered a speech at Wiki Finance EXPO Hong Kong 2026
Today, I’d like to share with you some insights from this session, where we’ll explore perspectives across major global markets from several angles. First, let’s discuss liquidity. Regardless of asset class, as of today—including major crypto assets—all are fundamentally tied to global core liquidity.
After the 2008 financial crisis, global liquidity reached a阶段性 peak between 2008 and 2021. Liquidity cannot be understood solely by news about interest rate hikes or cuts—it has three dimensions that everyone must remember.
Interest rate hikes and cuts only reflect changes at the ends of the yield curve and do not represent the full liquidity environment. Observing liquidity can be broken down into three aspects: the volume of water in the pool, the water temperature, and the distribution of liquidity pressure within the pool. The underlying logic can be simply understood as P/Q×G. From a professional perspective, liquidity can be tracked through interest rates, the yield curve, the Federal Reserve’s balance sheet, open market operations, and other factors. However, there is also a simpler way to observe it: in current traditional trading, most participants treat mainstream crypto assets like Bitcoin as leading indicators of liquidity conditions.
01. From the speculative hype of low-quality assets to “circle contraction”: Two faces of the liquidity cycle
After the 2020 pandemic, the world entered an extreme period of monetary easing characterized by low interest rates and central bank balance sheet expansion. During this easing cycle, global financial assets exhibited a typical pattern: speculative frenzy targeting low-quality assets.
For example, the retail short squeeze in the U.S. stock market with GameStop, and the surge of numerous crypto meme coins, are both fundamentally the result of excessive liquidity—when money is abundant, any asset can be pumped up. But when liquidity begins to contract overall, the market enters a “rationalization” phase: capital actively distinguishes between good and bad assets, abandoning inferior ones. This process of bubble compression was already underway in the second half of 2021.
From the second half of 2021 to 2022, a typical example was the cryptocurrency market, where Bitcoin dropped from over $70,000 to around $20,000, and NVIDIA’s stock in the U.S. market fell approximately 64% to 65% throughout 2022. This process was like squeezing a sponge, continuously removing market bubbles.
This year, the real turning point occurred in November last year. 2021 marked the peak of central bank balance sheet expansion, while the end of last year was the critical juncture where liquidity was tightened by both monetary tightening and balance sheet reduction. Simply put: when a reservoir starts draining, market liquidity doesn’t immediately tighten—it’s only when balance sheet reduction reaches a certain scale that the market truly feels the pressure.
In November and December last year, when Bitcoin was around 110,000, I made a bet with Li Lin that crypto assets would likely halve in value over the next year—if that came true, it would once again confirm that the fundamental logic of crypto assets is entirely tied to global liquidity.
Key observation in November last year: Fed’s SRF open market operations. This indicator signifies that structural funding pressures have emerged in the market after balance sheet reduction reached a critical threshold. It’s important to understand that banks tightening credit and liquidity contraction does not mean everyone is short on cash—funding pressure is transmitted in layers: highly leveraged, weaker entities are the first to face liquidity crunches, while high-quality, leading institutions still maintain ample funding. As liquidity stress filters through layers, capital markets experience a “shrinking circle”: Funds first sell off peripheral, liquidity-sensitive weak assets and continuously concentrate toward the most core, highest-certainty assets.
Many retail traders in crypto have a misconception: when the crypto market is quiet, all the money flows into trading U.S. stocks. This is a very retail-oriented mindset. Objectively speaking, during a liquidity tightening cycle, capital will first liquidate all high-volatility, highly speculative assets from their portfolios—cryptocurrencies and small-cap speculative stocks are prioritized for reduction. When money is abundant and liquidity is loose, capital is willing to speculate on all kinds of junk assets; when money is scarce and liquidity tightens, capital focuses only on truly valuable core assets—that’s the essence of a “consolidation” market.
02. Key Turning Point for the AI Industry: Free Cash Flow Hits Zero, Capital Expenditure Narrative Fully Collapses
Since November last year, global capital has continuously flowed toward the long-term productivity upgrade theme—namely, the AI sector. The logic behind the AI sector can be compared to large-scale fixed asset investment; it’s easier to understand by drawing an analogy to domestic infrastructure development.
In 2001, the central topic in the market was large-scale domestic infrastructure construction, as the old saying goes, “If you want to get rich, first build roads.” At that time, Lin Yifu and Xie Guozhong were both discussing how highways and railways could stimulate economic growth. In 2002, the two sessions confirmed the direction of infrastructure development; in 2003, central government and land-related fiscal funding were fully implemented, triggering a wave of nationwide road and bridge projects and ushering in a long-term capital expenditure cycle. In 2004, institutional portfolios centered on upstream infrastructure equipment and raw material companies such as Sany Heavy Industry and Anhui Conch Cement.
The logic of analogy is fully aligned with the current AI sector and will not alter the underlying industry dynamics simply because of the label “AI.” The first phase of AI was driven by the real-world deployment of applications like ChatGPT, which spurred corporate capital expenditure. Starting in 2023, global tech companies collectively invested in digital infrastructure—namely, computing power and data center construction. Large-scale digital infrastructure development benefits upstream hardware, storage, optical modules, HBM, and other related产业链 components, with Samsung Electronics, SK Hynix, and TSMC serving as the steel, cement, and construction machinery of this infrastructure era. However, the second quarter of this year marks a critical turning point for the entire industry chain, amplified by the dual forces of liquidity contraction.
After yesterday’s Google earnings report, seasoned investors could clearly sense that the market’s dominant logic of the past two to three years has become obsolete. The market rules for 2023, 2024, and 2025 were simple: when major internet companies increased investment in AI infrastructure and expanded capital expenditures, the market rewarded them with high valuations. But after the Q2 earnings reports from major tech firms this year, even as capital expenditures continued to grow rapidly, stock prices declined.
The core reason is that investors have captured a key data point: all major companies that have heavily invested in AI infrastructure have zero free cash flow. Last night, the most critical metric in Google’s earnings report was free cash flow. Many investors still cling to the old logic that increasing capital expenditures will drive stock prices higher—but that era is over.
The market pricing logic has completely shifted: previously, competition was based on the scale of capital investment, now capital will question whether infrastructure investments can generate sustained traffic and revenue that recoup costs. Zero free cash flow is the defining signal marking the transition from the first to the second phase of the AI industry. Companies planning to continue increasing capital expenditures can only raise external funding through equity issuance or bond offerings; external capital comes at a cost, and investors’ evaluation standards will become extremely stringent.
Here’s a key metric to track closely: the ratio of capital expenditure (CapEx) to cloud revenue growth. Currently, Google’s ratio is approximately 1.9, meaning that for every $1.90 invested in infrastructure, only $1.00 in cloud revenue is generated—this is the core reason why capital markets are reluctant to sustain high valuations.
The overall liquidity environment has tightened, with global capital continuously flowing into a limited number of high-conviction assets, compounded by a shift in the industry cycle—leading to significant risk volatility in a market characterized by contraction.
Taking NVIDIA as an example, I’ve fully mapped out the industry cycle: 2022 marked the confirmed starting point of NVIDIA’s industry cycle, when its market capitalization fell from $1 trillion to $100 billion; after the launch and explosion of ChatGPT, NVIDIA officially entered its value growth phase. In 2023 and 2024, NVIDIA’s narrative became fully self-reinforcing: consistent earnings growth, expanding orders driven by global AI capital expenditures, and market capitalization surging past $1 trillion, $2 trillion, and $3 trillion, with extremely low stock volatility and virtually no risk of deep corrections.
But after returning from my research trip to Singapore in June 2024, I alerted major financial institutions to the risk: NVIDIA’s corporate operations, industry supply and demand, and fundamental business conditions were all sound—the risk came entirely from off-exchange financial leverage.
Today, many Gen Z investors hold serious misconceptions: they believe stock price movements must perfectly align with fundamentals. This view is completely wrong! Capital markets price in market expectations, which often significantly precede a company’s actual fundamentals.
For example: The current industry reality is that HBM capacity is scarce and supply is insufficient—this fundamental fact is accurate—but it does not imply that stock prices will continue to rise. This is a classic cognitive bias: earnings reports and production capacity reflect the present reality, while stock prices trade on future expectations; fundamental data lags significantly behind market pricing. This is my hands-on experience from over two decades in the industry.
In July 2024, NVIDIA plummeted 20% in just a few trading days, while Japan’s stock market dropped 10% in a single day. At the time, many analysts released reports attributing the decline to the Bank of Japan’s interest rate hike and the unwinding of yen carry trades—these were merely surface-level explanations. The underlying truth is: global capital flooded into a handful of assets perceived as certain, and extreme certainty bred extreme greed, manifesting directly in investors wildly leveraging up.
Here’s a simple trading analogy: You and I are playing cards—you have a 6, I have a 5, and you know for certain your hand is stronger. An average retail trader might go all in, but a skilled trader would use maximum leverage to bet everything. Key takeaway to remember: Certainty breeds greed, and everything has two sides; the action corresponding to greed is adding leverage. The market widely expects NVIDIA to have strong long-term orders, so traders will continuously pile on leverage to amplify returns—that’s instinctive trading behavior. Once leverage accumulates to a critical point, it will inevitably trigger sharp volatility and a rapid decline.
The current market is replicating a similar pattern: some memory chip stocks, with no negative industry catalysts, stable business operations, full order books, and steady earnings growth, are experiencing frequent and sharp price drops. Many young traders in the Korean market saw substantial profits one day but suffered large losses the next. The root cause is not Samsung or SK Hynix, nor is it the supply and demand dynamics of the HBM supply chain—the core issue is excessive leverage within the market.
The underlying logic is identical to NVIDIA’s crash in July 2024: Precise, high-confidence assets fuel leverage bubbles; when leverage hits a critical threshold, a collapse is inevitable—no leverage-driven rally can last forever. Here’s a simple risk indicator: when recent graduates with no practical experience pour all their funds into leveraged bets on Samsung and SK Hynix, danger is near. When speculative retail investors flood niche, specialized sectors, a bubble burst is only a matter of time. The market has been in a prolonged liquidity contraction phase for years. Everyone knows which core assets offer certainty—but the real risk isn’t in industry fundamentals; it lies in liquidity and leverage. This must be taken seriously.
The market has now reached a critical juncture in the first phase of AI, where the narrative based solely on capital expenditure expansion has run its course, and we can expect significant volatility and valuation corrections. For the overall outlook on U.S. equities this year: maintaining a sideways trading range is already an optimistic expectation. Some may argue that after the steep decline in March, U.S. markets rebounded in May and June. However, it’s important to recognize that the May-June rally was an extreme structural move, driven by only a handful of stocks lifting the indices, while the vast majority continued to decline. The structure of the A-share market over the past year has been identical: 55% of stocks are trading below their levels corresponding to the 3,000-point mark, with the index supported solely by a few leading AI-focused companies.
Summarizing the current market environment: liquidity is tightening, market differentiation is extreme, and the AI industry cycle is entering a critical turning point. Once again, emphasize that the long-term development logic of the AI industry remains unchanged—productivity upgrades are the definitive主线—but blind long-term holding of assets is not advisable; instead, adopt a comprehensive industry cycle approach and deploy strategies in phases.
I have built a five-layer analytical framework: industry layer, economic layer, inflation layer, liquidity layer, market layer. At present, there is no need to invest significant effort in dissecting the economic layer; the core focus should be on the industry, liquidity, and market layers, with the weight of macroeconomic analysis significantly reduced. Some may ask whether it is still necessary to deeply analyze the U.S. economy. The answer is completely unnecessary. The reason is that U.S. corporations are continuously expanding capital expenditures, and households completed deleveraging as early as 2008. In other words, there is no need to closely examine high-frequency economic data—the defining characteristics of the U.S. economy can be summed up in two words: resilience.
03. Global Market Landscape: The Only Main Theme Is AI
At the market level, the only global theme is artificial intelligence—currently, all global capital follows this single core investment logic. Looking at globally allocatable assets, the only core markets for the future are Japan, South Korea, Taiwan, Mainland China, and the United States; all other regions have minimal allocation value. In Europe, only ASML warrants attention—other assets hold no meaningful allocation potential.
Consider two questions: Is the current movement of the Korean stock market related to Korea’s domestic real economy? Completely unrelated. What about Japan’s stock market—is it tied to Japan’s domestic economy? Also unrelated. Looking closely at Japan’s core stock assets, they are all upstream equipment manufacturers in the AI supply chain. Market attention is mostly focused on Samsung and SK Hynix, but the core production equipment these two companies purchase comes entirely from Japanese firms, fully integrating the entire supply chain. In the Taiwan region, the only core asset is TSMC; there are no other industrially significant companies with comparable investment value.
The entire AI sector is fundamentally an industry investment driven by productivity, operating with fixed cyclical patterns. Let me clarify the core conclusion: the point in Q2 when major tech firms’ free cash flow hits zero is the market’s critical turning point. Before and after this inflection point, the entire market’s asset pricing logic reverses completely—this is crucial to remember.
The AI industry chain is divided into upstream, midstream, and downstream segments, each with its own independent industrial lifecycle and clear cyclical rotation and allocation windows. Do not treat AI as a blind belief and hold it long-term without consideration—purely speculating on AI concepts will inevitably lead to losses. Many people ask me if I’m bearish on AI, but this question itself contains a logical flaw. Over the past decade, the market has reached a consensus: artificial intelligence is the core driver of the next generation of productivity, and this is unquestionable. Believing in the sector does not mean holding any single asset blindly at all times. NVIDIA, as a core upstream hardware player, has already completed its high-growth phase and entered a mature, blue-chip stage starting in 2025, which is why its price appreciation has significantly slowed from last year to this year. You won’t have to wait long—Samsung and SK Hynix will also enter their mature cycles. Overall growth in the upstream hardware segment is slowing, and growth momentum will gradually shift downstream.
Full AI industry cycle rhythm forecast: 2022 dominated by upstream hardware, 2026 sees software layer valuation digestion and reshaping, and around 2030, terminal application layer valuation adjustments and repricing occur. The full AI industry cycle lasts approximately 20 to 25 years, with 10 years already completed; the first decade was driven by upstream hardware infrastructure, and the next decade will be led by terminal applications.
However, there is currently a cycle gap, and the next 10 to 18 months represent a transition window for the industry. During this window, avoid going all-in; instead, strictly follow industry cycle patterns and deploy investments step by step to mitigate the risk of large fluctuations. Here, it’s important to distinguish between two key concepts: AI coding tools and development assistance tools, from a programmer’s perspective, belong to the industry support tool layer and are not part of the end-user application layer—there is a significant difference in their valuation logic.
04. Karen Wash and the Paradigm Shift in Liquidity: Central Banks No Longer Backstop, Cryptocurrencies Enter Maturity
Finally, I’ll return to the liquidity dimension, which is a core variable closely tied to crypto assets. Why is new Federal Reserve Chair Karen Wash a key signal? Her appointment signals the complete overhaul of the central bank’s core policy framework established by Bernanke after the 2008 financial crisis.
I wrote an analysis note in January: this personnel adjustment signals a return to the pre-2008 path of central bank policy. A brief overview of the policy context: the 2008 financial crisis exposed massive systemic financial risks. Policymakers drew lessons from the 1929 Great Depression: allowing free markets to operate entirely unchecked meant markets could not stabilize themselves during crises. As a result, Keynesian stimulus policies were widely implemented globally after 2008.
Bernanke and Yellen, former chairs of the Federal Reserve, all followed the same core principle: after a financial crisis, the central bank must step in to stabilize and rescue the market. However, every policy has two sides, just like leverage in investing—leverage can rapidly amplify returns, but it can also lead directly to total account liquidation. Central bank intervention can quickly calm market panic and prevent a repeat of the Great Depression, but prolonged, unconditional market bailouts by the central bank can foster speculative behavior and fuel large-scale asset bubbles.
The market has a professional term called “the Fed put”: whenever the market declines, capital boldly buys without hesitation, as all traders bet that central banks will always intervene to rescue the market. When a consensus forms that profits go to investors while losses are borne by the central bank, all financial assets become severely overvalued.
The central theme of Karen Wash’s public speeches can be summarized in one sentence: Central banks only fulfill their statutory responsibilities. Central banks’ two statutory core objectives: stable employment and inflation control, and they do not routinely bail out stock markets. With ongoing technological advancement and steady productivity growth, central banks now have the conditions to exit long-term market support models.
It can be compared to family education: when a child enters high school and gains the ability to live independently, parents should not handle everything for them, as this fosters dependency. Since Karen took office, many market participants have misinterpreted the situation as signaling expectations for rate cuts or balance sheet reduction, but the real focus is on balance sheet reduction, which has relatively low correlation with short-term interest rate movements. The central issue is how to orderly complete balance sheet reduction and return the central bank’s role to its pre-2008 standard positioning.
This marks the definitive end of the largest global liquidity easing cycle in human history, spanning from 2008 until Karen’s appointment. Therefore, do not harbor illusions: over the next 5 to 10 years, we will not see a repeat of the comprehensive monetary flood and broad-based asset price increases that characterized 2008 to 2026. Capital will flow back to core assets with genuine long-term value—this is a pivotal turning point in liquidity that will fundamentally reshape everyone’s investment strategy. The investment logic will shift from the previous approach of broad diversification and synchronized asset appreciation to a focused allocation on a limited number of high-quality core assets.
The crypto market is also undergoing similar changes. Many traders have observed that the market capitalizations of Bitcoin and Ethereum are gradually stabilizing, volatility continues to decline, market liquidity is becoming more stable, and market participants are becoming more institutionalized. These characteristics are typical of core assets that have survived after the bubble has been cleared. The narrative logic that once fueled speculation around empty crypto projects has completely lost its validity.
Liquidity is the top-level core factor influencing all financial assets, and everyone must fully understand this analytical framework this year. By applying the liquidity framework to break down industries and corporate fundamentals, your analysis of various assets will become much clearer. Due to time constraints today, I won’t be able to meticulously unpack each of the five layers of this analytical framework.
I prefer to engage with everyone on the underlying logic and analytical methodologies; once the foundational thinking is clear, you won’t overreact to short-term market fluctuations. That concludes my sharing—I hope it has provided some insight. Thank you all.
Leave a comment