Chief Economist of New Fire Group, Fu Peng's latest speech: Cryptographic assets are deeply tied to liquidity, and the global asset "contraction" market differentiation is intensifying

Author: New Fire Technology

Mr. Fu Peng, Chief Economist of New Fire Group, was invited to participate in the Wiki Finance EXPO Hong Kong 2026 and delivered a keynote speech. Mr. Fu shared his core views on the current global asset classes and market trends from the perspective of the global liquidity framework, and made systematic judgments on the underlying logic of the cryptocurrency market. Mr. Fu Peng, Chief Economist of New Fire Group, delivers a speech at Wiki Finance EXPO Hong Kong 2026 Here is the full text of the speech:

Today I would like to share with you some perspectives on the global major markets from several dimensions. First, let’s talk about liquidity. Regardless of the type of asset, as of today, including mainstream cryptocurrencies, they are all fundamentally linked to global core liquidity.

After the 2008 financial crisis, global liquidity reached a peak from 2008 to 2021. Liquidity cannot be assessed solely by interest rate hikes or cuts reported in the news; it can be divided into three dimensions, which everyone must remember.

Interest rate hikes and cuts are merely changes at the end of the interest rate curve and do not represent the complete liquidity environment. Observing liquidity can be broken down into three aspects: the amount of water in the pool, the temperature of the water in the pool, and the distribution of funding pressure within the pool. The underlying logic can be simply understood as P/Q×G. From a professional perspective, liquidity can be tracked from the interest rate side, the interest rate curve, the Federal Reserve's balance sheet, open market operations, and so on. However, there is a simpler way to observe: in current traditional trading, mainstream cryptocurrencies like Bitcoin are generally regarded as leading indicators of liquidity strength.

01. From "Crazy Speculation on Poor Quality Assets" to "Contraction": The Two Faces of the Liquidity Cycle

After the pandemic in 2020, the world entered an extreme easing window characterized by low interest rates and central bank balance sheet expansion. During this easing cycle, global financial assets exhibited typical characteristics: crazy speculation on poor quality assets.

For example, the short squeeze of retail investors in GameStop in the U.S. stock market and the massive surge of numerous worthless cryptocurrencies in the crypto market are essentially results of liquidity flooding—when there is more money, any type of asset can be speculated upon. However, when liquidity begins to contract overall, the market will experience a "contraction" trend: funds actively differentiate between good and bad assets, and poor quality targets are abandoned by capital. This process of squeezing out the bubble had already started in the second half of 2021.

From the second half of 2021 to 2022, a typical case is Bitcoin in the crypto market, which fell from over $70,000 to around $20,000, and Nvidia in the U.S. stock market, which saw a decline of about 64% to 65% throughout 2022. This process is like squeezing water from a sponge, continuously expelling market bubbles.

This year, the real core turning point appeared in November last year. 2021 was the peak of the central bank's balance sheet expansion, while the end of last year marked a critical juncture of dual tightening from liquidity and balance sheet reduction. To explain simply: draining the pool does not mean that funds become tight the moment the balance sheet reduction starts; only when the balance sheet reduction progresses to a certain extent will the market genuinely feel the pressure of funds.

In November and December last year, Bitcoin was around $110,000, and I made a bet with Li Lin at that time: that cryptocurrency assets would likely be halved in the coming year. If this comes true, it would further confirm that the underlying logic of cryptocurrency assets is completely tied to global liquidity.

The core observation indicator in November last year was the Federal Reserve's SRF open market operations. This indicator represents that after the balance sheet reduction reaches a critical value, the market has already shown structural funding pressure. It is important to understand that when banks tighten credit and liquidity contracts, it does not mean everyone is short of money; the pressure of funds transmits in layers: high-leverage and weak-quality entities are the first to experience funding breaks, while high-quality leading entities still have ample funds. The layered transmission of liquidity leads to a "contraction" in the capital market: funds first sell off peripheral, liquidity-sensitive weak assets, continuously concentrating towards the most core and highest certainty assets.

Many retail investors trading cryptocurrencies have a misconception: when there is no market in the crypto space, all funds run to speculate on U.S. stocks. This is a very retail-oriented way of thinking. Objectively speaking, during a tightening cycle, funds will first clear out all high-elasticity and high-speculation assets from their portfolios; cryptocurrencies and small-cap speculative stocks will be prioritized for reduction. When there is plenty of money and liquidity is loose, funds are willing to speculate on all kinds of junk assets; when money is scarce and funds tighten, they will only focus on truly valuable core assets, which is the essence of the "contraction" trend.

02. Key Turning Point in the AI Industry: Free Cash Flow Goes to Zero, Capital Expenditure Narrative Completely Fails

Since November last year, global funds have continuously gathered towards the main line of long-term productivity upgrades, which is the artificial intelligence track. The logic of the AI track can be compared to large investments in fixed assets, making it easier to understand with examples from domestic infrastructure.

In 2001, the core issue in the market was large-scale infrastructure construction in the country, as the old saying goes, "To get rich, first build roads." At that time, Lin Yifu and Xie Guozhong were discussing the economic pull of road and railway infrastructure. In 2002, the two sessions finalized the direction of infrastructure development, and in 2003, central fiscal and land finance matching funds were all implemented, with nationwide road and bridge projects starting in batches, entering a long-term capital expenditure cycle. In 2004, the core targets for institutional allocation were companies like Sany Heavy Industry and Conch Cement, which are upstream equipment and raw material companies in infrastructure.

This comparative logic is fully applicable to the current AI track; it will not change the underlying rules of the industry just because it is labeled "AI." The first half of AI was driven by applications like ChatGPT, which stimulated corporate capital expenditure willingness. Starting in 2023, global tech companies are concentrating on digital infrastructure, which includes computing power and data center construction. Large-scale digital infrastructure construction will benefit upstream hardware, storage, optical modules, HBM, and other industry chain targets, such as Samsung Electronics, SK Hynix, and TSMC, corresponding to the steel, cement, and engineering machinery of the infrastructure era. However, the second quarter of this year is a critical turning point for the entire industry chain, resonating with the dual variables of liquidity contraction.

After the release of Google's earnings report yesterday, mature investors can clearly perceive that the market's main line logic from the past two to three years has already failed. The market rules for 2023, 2024, and 2025 are very simple: if internet giants increase AI infrastructure and expand capital expenditure, the market will grant high valuations. However, after the earnings reports of major companies in the second quarter of this year, even if capital expenditure maintains rapid growth, stock prices have instead fallen.

The core reason is that investors have captured key data: all leading companies heavily investing in AI infrastructure have seen their free cash flow go to zero. The most critical indicator in Google's earnings report last night was free cash flow. Many investors still cling to the old logic, believing that as long as capital expenditure continues to expand, stock prices will rise; this era has ended.

The market pricing logic has completely switched: previously, the competition was about the scale of capital investment; now, funds will question whether infrastructure investment can bring sustained traffic and revenue to recoup costs. Free cash flow going to zero is a symbolic signal of the transition from the first phase to the second phase of the AI industry. If companies plan to continue increasing capital expenditure, they can only finance externally through issuing stocks or bonds, and external funds come with costs, making investors' review standards extremely stringent.

Here is a refined tracking indicator for you: the ratio of capital expenditure (CapEx) to cloud business revenue growth rate. Currently, Google's ratio is about 1.9, meaning that for every 1.9 yuan invested in infrastructure, only 1 yuan of cloud business revenue is generated, which is the core reason why the capital market is unwilling to continue granting high valuations.

With the overall funding environment tightening, global funds continue to concentrate on at least a few high-certainty assets, combined with the industry cycle switch, the "contraction" trend will inevitably lead to severe risk fluctuations in the market.

I will take Nvidia as an example to outline the industry cycle: 2022 was the confirmation point for Nvidia's industry cycle, as its market value fell from a trillion to hundreds of billions that year; after the explosion of ChatGPT, Nvidia officially entered the value growth phase. In 2023 and 2024, Nvidia's logic is completely closed-loop: continuous performance growth, global AI capital expenditure driving order expansion, with market value breaking through 1 trillion, 2 trillion, and 3 trillion in succession, and stock price volatility being extremely low, with almost no deep correction risk.

However, after I returned from research in Singapore in June 2024, I alerted various financial institutions to the risks: Nvidia's business operations, industry supply and demand, and industry fundamentals are all sound, but the risks come entirely from off-market financial leverage.

Currently, the new generation of post-00 investors has a serious cognitive bias, believing that stock price movements must completely match fundamentals. This view is completely wrong! The capital market prices are based on market expectations, which will significantly lead real fundamentals of companies.

For example: the current industry situation is that HBM production capacity is tight and supply is insufficient, which is a factual fundamental, but it does not imply that stock prices will continue to rise. This is a typical cognitive bias: financial reports and production capacity reflect the current reality, while stock prices trade on future expectations, with fundamental data lagging significantly behind market pricing. This is my practical experience from over twenty years in the industry.

In July 2024, Nvidia's stock plummeted by 20% in just a few trading days, while the Japanese stock market fell by 10% in a single day. At that time, many researchers released reports attributing the decline to the Bank of Japan's interest rate hikes and the unwinding of yen carry trades, which is just surface reasoning. The underlying truth is: global funds are concentrating on a few certain assets, and extreme certainty breeds extreme greed, directly reflected in investors' reckless leverage.

To give a simple trading analogy: if we are playing cards, your card is 6, mine is 5, and you clearly know your hand is superior. Ordinary retail investors would heavily enter the market, but qualified traders would leverage their entire position. The core conclusion must be remembered: certainty breeds greed, and everything has two sides; the operation corresponding to greed is to increase leverage. The market uniformly predicts that Nvidia's long-term orders are sufficient, and traders will continuously add leverage to amplify returns, which is instinctive for traders. Once leverage accumulates to a critical point, it will inevitably trigger severe fluctuations and rapid declines.

The current market is replicating similar trends: some memory chip targets have no industry headwinds, stable business operations, full orders, and steadily growing performance, yet stock prices frequently plunge. Many young traders in the Korean market made substantial profits one day, only to face significant losses the next day. The root of the problem lies not with Samsung or SK Hynix, nor with the supply and demand of the HBM industry; the core issue is the excessive accumulation of leverage in the market.

The underlying logic is completely consistent with Nvidia's flash crash in July 2024: high certainty assets breed leverage bubbles, and once leverage reaches a critical point, it will inevitably collapse; there is no such thing as a perpetually sustainable leverage market. Here is a simple risk judgment standard: when freshly graduated young people with no practical experience leverage all their funds to bet on Samsung or SK Hynix, it means that risk is approaching. Originally niche professional tracks are flooded with speculative retail investors, and the bubble bursting is just a matter of time. The market has been in a funding contraction cycle for the past few years, and everyone is aware of a few core certain assets, but the risk does not lie in the industry fundamentals; it is hidden in liquidity and leverage, which must be closely monitored.

The current market has reached the core critical point of the first phase of AI; the narrative logic relying solely on capital expenditure expansion has run its course, and the market will experience significant fluctuations and valuation adjustments. Regarding the overall judgment of U.S. stocks this year: maintaining a sideways trend for the index is already an optimistic expectation. Some may argue: after the significant drop in U.S. stocks in March, there was a rebound in May and June. However, it is important to distinguish that the rise in May and June belongs to an extreme structural trend, with only a very small number of stocks lifting the index, while the vast majority of stocks continue to decline. The structure of the A-share market in the past year is completely consistent: 55% of individual stocks are priced below the corresponding level of 3000 points, relying solely on a few leading stocks in the AI track to support the index.

To summarize the current market environment: liquidity is tightening, the market is extremely polarized, and the AI industry cycle is reaching a key turning point. I want to emphasize again: the long-term development logic of the AI industry has not changed; productivity upgrades are the certain main line, but one cannot blindly hold onto targets for the long term; it is essential to phase investments based on a complete industry cycle approach.

I have built a five-layer complete analysis framework: industry layer, economic layer, inflation layer, liquidity layer, and market layer. Currently, there is no need to invest a lot of energy in dissecting the economic layer; the industry layer, liquidity layer, and market layer are the core of the analysis, while the weight of macroeconomic analysis has significantly decreased. Some may ask whether it is still necessary to deeply dissect the U.S. economy? The answer is completely unnecessary. The reason is that U.S. companies are continuously expanding large-scale capital expenditures, and the household sector completed deleveraging as early as 2008. In other words, there is no need to closely examine high-frequency economic data; the core characteristic of the U.S. economy can be summed up in two words: resilience.

03. Global Market Landscape: The Only Main Line is AI

From a market perspective, the only main line globally is artificial intelligence; currently, global funds only have this core investment logic. Looking at globally allocatable assets, the future core markets are only Japan, South Korea, Taiwan, mainland China, and the United States, while other regions have very low allocation value; in Europe, only ASML has attention value, while other targets have no allocation significance.

You can think about two questions: Is the current trend of the South Korean stock market related to the local real economy? It is completely unrelated. Now looking at the Japanese stock market, is it linked to the domestic economy? It is also unrelated. Upon closer examination of the core assets in the Japanese stock market, they are all upstream equipment manufacturers in the AI industry chain. Most market attention is focused on Samsung and SK Hynix, but the core production equipment purchased by these two companies comes from Japanese firms, fully connecting the entire industry chain upstream and downstream. The only core target in the Taiwan region is TSMC, with no other core industry companies of allocation value.

The entire AI track is essentially an industry investment driven by productivity, with fixed cyclical operating rules. I want to clarify the core conclusion: the point at which major tech companies' free cash flow went to zero in the second quarter is a significant turning point for the market. Before and after this turning point, the entire asset pricing logic of the market has completely reversed; this point must be remembered.

The AI industry chain is divided into upstream, midstream, and downstream, with each segment having an independent industry lifecycle, and there are clear sector rotation switches and allocation windows. Do not treat AI as a blind faith for long-term holding; simply speculating on AI concepts will definitely lead to pitfalls. Many people ask me if I am not optimistic about AI; this question itself has a logical flaw. Over the past decade, the market has reached a consensus: artificial intelligence is the core main line of the next generation of productivity, and there is no dispute about this. Being optimistic about the track does not mean holding onto a single target blindly at any time. Nvidia, as a core upstream hardware target, has completed its high-speed growth cycle and will enter the mature blue-chip stage starting in 2025, thus the growth rate has significantly narrowed from last year to this year. There is no need to wait too long; Samsung and SK Hynix will also enter the mature cycle, and the overall growth rate of the upstream hardware sector will slow down, with growth gradually shifting to the downstream segments.

The complete rhythm prediction of the AI industry cycle: in 2022, the upstream hardware dominated the market, in 2026, the software layer will undergo valuation digestion and restructuring, and around 2030, the terminal application layer will see valuation adjustments and repricing. The complete AI industry cycle lasts about 20 to 25 years, and we have already completed 10 years; the main line in the first ten years was upstream hardware infrastructure, while the main line in the next ten years will be terminal applications.

However, there is currently a cyclical disconnection, and the next 10 to 18 months will be an industry switching window period. During this window period, do not go all in; strictly follow the rules of the industry cycle for phased investments to avoid significant volatility risks. Here, I want to distinguish a set of key concepts: AI code tools and development assistance tools from a programmer's perspective belong to the industry support tools layer, not the terminal application layer, and there is a significant valuation logic gap between the two.

04. Karen Walsh and the Liquidity Paradigm Shift: Central Banks No Longer Provide a Safety Net, Cryptocurrency Assets Mature

Finally, I will return to the liquidity dimension for further explanation, which is also a core variable highly related to cryptocurrency assets. Why is the new Federal Reserve Chair Karen Walsh a key signal? Her appointment signifies a complete rewrite of the core policy framework established by Bernanke after the 2008 financial crisis.

I wrote an analysis note in January: this personnel adjustment means that central bank policy is returning to the old path before 2008. To briefly outline the policy background: the 2008 financial crisis exposed huge systemic financial risks. Policymakers learned from the experience of the Great Depression in 1929: completely allowing the free market to operate means that when a crisis occurs, the market cannot stabilize on its own. Therefore, after 2008, Keynesian stimulus policies were widely implemented globally.

Bernanke and Yellen, who served as Federal Reserve Chairs, all followed the same core thinking: after a financial crisis, the central bank must intervene to stabilize the market. However, any policy has two sides, consistent with the logic of leverage in investing. Leverage can quickly amplify returns but can also lead to account liquidation and zero. Central bank interventions can quickly calm market panic and prevent a repeat of the Great Depression crisis, but if the central bank continuously provides an unlimited safety net for the market, it will foster speculative sentiment and create large-scale asset bubbles.

There is a professional term in the market called "Federal Reserve put option": as long as the market declines, funds dare to buy recklessly, with all traders betting that the central bank will definitely step in to stabilize the market. When the market forms a unified expectation: profits belong to investors, while losses are covered by the central bank, all financial assets will appear severely overvalued.

The core message of all of Karen Walsh's public speeches can be summarized in one sentence: the central bank only performs its statutory responsibilities. The central bank has two core statutory goals: stabilizing employment and controlling inflation, and it will not routinely support the stock market. Now that technology continues to advance and productivity steadily improves, the central bank has the conditions to exit the long-term safety net model.

This can be compared to family education: when a child enters high school and has the ability to live independently, parents cannot handle everything for them, as it fosters a dependency mentality. After Karen took office, many market participants misinterpreted it as expectations for interest rate cuts and balance sheet reductions; the core focus is actually on balance sheet reduction, which is less related to short-term interest rate fluctuations. The core issue is how to orderly complete the balance sheet reduction and return the central bank's functions to the standards set before 2008.

This signifies the complete end of the largest global liquidity easing cycle in human history from 2008 to Karen's appointment. Therefore, do not harbor illusions: in the next 5 to 10 years, there will not be a repeat of the comprehensive flooding of liquidity from 2008 to 2026, with all categories of assets rising simultaneously. Funds will flow back to truly valuable core assets, which is a key turning point in terms of liquidity, and will comprehensively rewrite everyone's investment strategies. The investment logic will shift from a previously fully diversified layout with all types of assets rising in sync to focusing on a small number of high-quality core targets.

The cryptocurrency market will also undergo similar changes. Many traders have already observed: Bitcoin and Ethereum's market values are gradually stabilizing, volatility is continuously decreasing, market liquidity is tending towards stability, and market participants are becoming more institutionalized. These characteristics are typical manifestations of core assets that have survived after the bubble has cleared.

Liquidity is the top core influencing factor for all financial assets, and this year everyone must thoroughly understand this analytical logic. Following the liquidity framework to further dissect the industry and corporate fundamentals will make the analysis of various assets much clearer. Today's sharing time is limited, and I cannot detail the five-layer analysis framework one by one.

I hope to communicate with you about the underlying logic and analytical methodology; once the underlying thinking is smooth, observing short-term micro fluctuations in the market will not be overly entangled. My sharing ends here, and I hope it can provide inspiration to everyone. Thank you all.