Author | @plur_daddy
Compiled | Odaily Planet Daily (@OdailyChina)
Translator | DingDang (@XiaMiPP)
Editor's Note: Gold and silver both plummeted, U.S. stocks fell across the board, and the cryptocurrency market was even more brutal, with over $26 billion liquidated in 24 hours. Bitcoin once flash-crashed to the $60,000 mark, plummeting nearly 20% in a single day; from its high of $126,000 in October last year, the price of BTC has been halved. What's even more frightening is that the market showed almost no significant resistance.
Everyone is frantically searching for reasons: U.S. tech stocks dragged down crypto market; Trump's nomination of Warsh sparked hawkish expectations; the strong dollar, poor employment data... These explanations sound reasonable. But in the view of the author of this article, they are more superficial than the core of the problem. The real underlying reason is: Money in the world is becoming insufficient. The massive AI capital expenditure cycle itself is shifting from "injecting liquidity" to "draining liquidity," leading to a substantial shortage of global financial capital. The following is the author's original text, which will deconstruct step by step how this mechanism operates.
We are experiencing a paradigm shift in the market, due to a shortage of financial capital caused by the AI capital expenditure cycle. This has profound implications for asset prices, as capital has been excessively abundant for a very long time. The Web 2.0 and SaaS paradigm that drove the market boom of the 2010s was essentially a extremely capital-light business model, which allowed a large amount of excess capital to flow into various speculative assets.
Yesterday, while discussing the market landscape, I had an "aha moment." I believe this is the most differentiated article I've written in a long time. Below, I will deconstruct, layer by layer, how all of this operates.
There is actually a highly similar mechanism between AI capital expenditure and government fiscal stimulus, which helps us understand the underlying logic.
In fiscal stimulus, the government issues treasury bonds, so the private sector absorbs the duration risk; then the government gets the cash and spends it. This cash circulates in the real economy and creates a multiplier effect. The net impact on financial asset prices is positive, precisely because of this multiplier effect.
In AI capital expenditure, mega-cap tech companies either issue bonds or sell treasury bonds (or other assets), again the private sector absorbs the duration risk; then these companies get the cash and put it to use. This cash also circulates in the real economy and creates a multiplier effect. Ultimately, the net impact on financial asset prices is still positive.
As long as these funds come from the "dry powder" (idle, unused capital) within the economic system, this process runs smoothly. It worked very well, almost "lifting all boats." In the past few years, this has been the dominant paradigm—AI capital expenditure acted like an incremental stimulus policy, injecting adrenaline into the economy and markets.
The problem is: Once the dry powder is exhausted, every dollar flowing into AI capital expenditure must be pulled from somewhere else. This triggers a convex battle for capital. When capital becomes scarce, the market is forced to reassess: where is the "most useful" place to deploy capital? Simultaneously, the cost of capital (i.e., the market-determined interest rate) rises.
Let me emphasize again: When money becomes scarce, a "knockout tournament" occurs among assets. The most speculative assets suffer disproportionate losses—just as they disproportionately benefited when capital was extremely abundant but lacked productive uses. In this sense, AI capital expenditure actually plays a role of "reverse QE," bringing negative portfolio rebalancing effects.
Fiscal stimulus typically doesn't face this problem because the Fed often ultimately becomes the absorber of duration risk, thus avoiding "crowding out" other uses of capital.
The "money" mentioned here can be used interchangeably with "liquidity." But the word "liquidity" is confusing because it has different meanings in different contexts.
Let me use an analogy: Money or liquidity is like water. You need the water level in the bathtub to be high enough for the financial assets (those floating rubber ducks) to all rise together. There are several ways to do this:
- You can increase the total amount of water (rate cuts / QE)
- You can unclog the inlet pipes (operations like the current RRP (Reverse Repo Program)/RMP (Reserve Management Purchases) "plumbing work")
- Or you can reduce the speed at which water drains from the bathtub.
Discussions about liquidity in the economy almost always focus on the money supply. But in fact, the demand for money is equally important. The problem we are facing now is: Demand is too high, leading to significant crowding-out effects.
Media reports suggest that the world's "deepest pockets"—Saudi Arabia and SoftBank—are basically tapped out. The whole world has been gorging on assets for the past decade and is now "stuffed." Let's look specifically at what this means.
Suppose Sam Altman (founder of OpenAI) reaches out to them, asking them to fulfill their previous commitments. Unlike in previous periods when they still had dry powder, now they must first sell something to free up money for him. So, hypothetically, what would they sell?
They would look at their investment portfolios and pick the assets they have the least confidence in: sell some underperforming Bitcoin; sell some SaaS software assets facing disruption risk; redeem funds from hedge funds with long-term poor performance. And these hedge funds, to meet redemptions, must sell assets. Asset prices fall, confidence weakens, the availability of margin tightens, triggering passive selling in more places. These effects cascade and amplify through the financial markets.
Worse still, Trump chose Warsh. This is particularly problematic because he believes the current problem is too much money, when in fact, we are facing the opposite problem. This is why the pace of these market changes has noticeably accelerated since he was selected.
I have been trying to understand: Why have memory chip manufacturers like DRAM / HBM / NAND (e.g., SNDK, MU) performed far better than other stocks. Sure, the underlying product prices are indeed soaring. But more importantly, these companies are now and will be in the near future in a state of supernormal profits—even though it's clear their profits are cyclical and will eventually fall back. When the cost of capital rises, the discount rate increases accordingly. The result is: Speculative assets with longer duration, reliant on future expectations, get hit, while assets with near-term cash flows benefit relatively.
In such an environment, crypto assets naturally get "decimated," as they are the frontline probes for changes in conditions. This is also why the market feels like it's "falling endlessly."
Highly speculative retail momentum stocks can hardly hold any gains, and even sectors with improving fundamentals are struggling hard.
As demand for money exceeds supply, sovereign bond and credit interest rates are both rising.
This is not a time to be complacently extremely long. This is a stage for defense, being extremely picky with holdings, and seriously managing risk. I am not telling you to sell everything; this article is not a trading directive. You should treat it as a contextual framework to help you understand what is happening.
I personally sold gold and silver near the highs, and most of my positions are now in cash. I'm in no hurry to buy anything. I believe that if you are patient enough, extremely rare opportunities will emerge this year.
Finally, thanks to the brilliant friends in the group chat who helped me thoroughly discuss these issues, including @AlexCorrino, @chumbawamba22, @Wild_Randomness.
