Arthur Hayes: AI Bubble Popping Soon, Crypto Market Under Short-Term Pressure

Original Author: Arthur Hayes

Original Translation Compiled by: Luffy, Foresight News

Is all of this just my illusion, or is investing in artificial intelligence nowadays as simple as subscribing to Citrini Research's services and blindly buying all the stocks they recommend?

Am I dreaming? Or has the price of oil long lost its influence over the economy and politics? That's why Trump and the Islamic Revolutionary Guard Corps can taunt each other on social media, while a large number of ships remain stranded in the Strait of Hormuz.

The 2-year US Treasury yield is 0.5 percentage points higher than the effective federal funds rate. With the market sending such a clear signal, will the Fed really hold its ground and refuse to raise rates at its upcoming meeting?

Will all the dividends that AI creates for the United States really only end up in the hands of a few tech workers?

Faced with this chaotic world, I have to stop and conduct a reality check, confirming whether I am awake or deep in a dream. Once the test results prove it's all an illusion, I will immediately adjust my investment portfolio. This article is my examination process. After typing these words and sorting out my thoughts, my position layout will undergo significant changes, or simply remain as is.

Let me first state my core judgment: the current market state feels more like a dream. Within the entire investment system, the price of oil and other hydrocarbon energy sources is a core variable with a reverse feedback effect. The essence of human perception of the world is converting energy into biological intelligence, and the logic of artificial intelligence is the same. This law will never be broken. The market might deviate from this common sense in the short term, but reality will eventually exact its revenge.

This article will start with oil prices and ultimately touch on the US elections. The current situation is very likely to trigger a crash in the AI stock market bubble, dragging the entire crypto market down with it. Only after the dust settles will Bitcoin have a chance to bottom out and rebound. I previously asserted that Bitcoin would never touch the $60,000 mark again. Clearly, that judgment was wrong, which is also the norm in market predictions. I always adhere to one principle: opinions can be strong, but there's no need to be stubborn.

Let's begin the analysis.

To Negotiate or Not: The Core Dilemma

Politicians always act in their own self-interest. Only Trump himself probably knows the real reason behind his unprovoked military action against Iran. Faced with the constant stream of rhetoric from him and his advisors, the outside world simply cannot discern the truth. Since that's the case, dwelling on the cause is meaningless. The real question is whether Trump and Iran's Islamic Revolutionary Guard Corps will choose a ceasefire and how they will end the standoff.

This conflict is now entirely orchestrated by Trump, and for him and the Republican Party, starting a war in an election year puts them in a difficult position.

In the United States, the prices of necessities like gasoline and food often directly determine the outcome of an election. Currently, passage through the Strait of Hormuz is hindered, and energy and food inflation continues to rise. This is fundamentally rooted in the Trump administration's military action against Iran, undertaken without the consent of the American people. Some might point fingers at Israel, but that argument doesn't hold water. Understanding US history makes it clear that domestic forces will never follow orders from abroad.

As long as the war doesn't affect their daily lives or cause casualties among friends and family, the American public isn't opposed to foreign wars. Trump has repeatedly emphasized that only thirteen US soldiers have died in this so-called "special military operation." This is also why the US favors using high-precision, long-range weapons to wage "video-game wars." Even though launching this Middle East conflict lacks a clear winning strategy and goes against the expectations of some supporters, the Republican base still chooses to stand with the party. The fact that some Republican congressmen who wavered in their stance were pressured by Trump's faction and lost their primary bids confirms this.

Trump's core vulnerability isn't that his base won't vote in November; it's that soaring prices will drive a large number of swing voters towards the Democrats. The cost of living issue has become the biggest problem on the path to Trump's election victory.

To win over swing voters, Trump must at least stabilize current oil prices. Now that the supply chain is beginning to digest the pressures from rising energy and various production raw material costs, completely curbing inflation is unrealistic. What Trump can do now is manage market expectations for inflation, rather than changing inflation itself.

Whether Trump is willing to reach a settlement with Iran depends entirely on oil prices. As oil prices climb, his rhetoric will soften; but once the market anticipates negotiations and oil prices fall in response, he will change his tune. After all, from a geopolitical standpoint, any agreement reached through such negotiations would likely be more disadvantageous than the deal the Obama administration signed with Iran. In the eyes of many voters, this would be akin to "defeat," costing the Republicans politically.

Negotiation always requires concessions from both sides. The Islamic Revolutionary Guard Corps has similar considerations. If oil prices are too high, their major trading partners will pressure Iran to compromise with the US. However, once Iran signals willingness to negotiate and oil prices drop, the pressure from those trading partners will also ease.

At current oil price levels, neither the US nor Iran has a strong incentive to back down. While oil prices are significantly higher than before the conflict, they haven't yet reached a level that triggers a full-blown crisis. The commodity market is relatively stable, there isn't a global famine, and most countries can supplement key industrial raw materials from other sources.

However, this delicate balance cannot last forever. A significant reduction in global core energy supply without a corresponding price surge is itself a contradiction of market principles. Once global spare production capacity is exhausted, spot prices will inevitably rise sharply – a consensus among many commodity analysts. The crisis hasn't fully erupted yet simply because global energy inventories were ample before the conflict.

If the US-Iran standoff persists until the end of Q2, we will inevitably see a sharp surge in the spot prices of hydrocarbon energy and various basic commodities in Q3 of this year.

To borrow Churchill's famous quote: Politicians will always make the right choice after exhausting all other options. Only when the situation spirals completely out of control will Trump and Iran truly sit down at the negotiating table. In my view, the disruption to shipping in the Strait of Hormuz is likely to continue until early Q3.

Let's assume oil prices will rise gradually amidst volatility. Given this backdrop, how will rising oil prices interact with Trump's campaign rhetoric?

The November Election Showdown: Republicans vs. Democrats

According to the odds on the prediction market Polymarket, the Republican Party can currently only maintain a razor-thin majority in the Senate, while facing significant losses in the House of Representatives.

Generally, it's believed the Republicans will lose the House, but I hold a different view. Trump still has a chance to turn the tide. The breakthrough lies in shifting the public discourse by making regulatory and tax-related statements targeting data center construction and the AI industry.

The current seat distribution among the parties is as follows (218 votes are needed to pass a bill):

Based on Polymarket's current odds, here is the projected party composition after the election:

The outlook for the Republicans' seats in both chambers after the election is not favorable. However, the Republican Party can change the situation through redistricting. When the existing rules guarantee defeat, changing the rules becomes a necessity. Assuming Polymarket's predictions are correct, the Republicans need to gain 19 seats. Redrawing district lines can reduce this number.

Here is the potential impact of redistricting:

Now the Republicans only need to gain another 11 seats. Next, let's look at which races are competitive. According to current polls, which districts might slightly lean Republican, within the margin of error?

There are 35 seats with significant uncertainty. As mentioned earlier, high inflation and rising living costs are negative narratives that are hard for Trump to reverse. Another major topic capable of engaging voters from both parties is the expansion of data centers and the impact of AI on the job market.

Almost everyone, except for the ultra-wealthy, is worried about data center construction driving up costs and fearful that AI will take away jobs. Many places have already implemented policies to halt new data center projects, and calls for taxing AI companies to subsidize the general public are growing louder. After all, the vast majority of people are not AI company executives or high-earning employees in the field.

For voters in competitive districts, these issues are highly influential. Trump could potentially secure the remaining key seats by taking a stance on the AI industry. At this stage, he only needs to make related statements; no concrete legislation is necessary. He simply needs to promise the general public that if the Republicans win, they will crack down on the AI industry after the election.

As a seasoned politician, Trump has always been adept at making campaign promises that he rarely fulfills afterward. His handling of the Epstein-related files is a classic example: he loudly declared during the campaign that he would thoroughly investigate those involved, but only released a small amount of information after taking office. He can employ a similar strategy now: during the campaign, he could announce plans to legislate slowing down data center expansion, impose windfall profit taxes on AI companies, and use the revenue for a new round of stimulus checks. Then, after the election dust settles and the Republican Party secures power, he could gradually backtrack on these statements.

Some might find it hard to understand Trump emulating left-wing Democratic politicians. But remember, he implemented the largest universal relief program in the US since the New Deal, without restricting how the money was spent, even if used for daily consumption. To secure his political position, temporarily distancing himself from AI moguls like Elon Musk and projecting an image of supporting the common person is not difficult for Trump.

If Trump were to make tough-sounding statements targeting the AI industry, the market wouldn't just see them as campaign tactics. Instead, it would assume the US is genuinely moving to restrict capital expansion in AI and increase taxes on the industry. Panic would spread instantly, and the AI stock market bubble would burst.

The recent public spat between Elon Musk and Trump on social media – where Musk's affiliated departments publicly questioned Trump, who responded by threatening to cancel government contracts linked to Musk's companies – caused Tesla's stock to plummet 18% in a single day. This shows how sensitive the market is to such political risks. Politics can foster an industry, but it can also deliver a swift blow.

That feud was later revealed to be a public relations stunt; the two quickly reconciled, and Musk was even invited to attend Trump's recent summit in Beijing with the Chinese President. But at the time, the market took it seriously, triggering a massive sell-off.

This was just the shockwave from a private disagreement between two individuals. If Trump officially states, on behalf of the Republican Party, a plan to impose heavy taxes on businesses related to AI models and agents, the impact would be far greater. Similar remarks from South Korean political circles previously caused their local composite stock index to nearly hit the daily limit drop the next day, only recovering after an official denial, allowing the market to resume its upward trend.

The current optimistic market expectations for the AI sector are built on the assumptions that industry revenue will continue its exponential growth and that the concentration of new technology and wealth won't trigger public backlash. This kind of thinking is detached from reality, more like being immersed in a dream. Trump's statements could become the reality check that bursts this illusion. Whether he actually acts depends, again, on oil prices.

The longer the Iran situation drives up oil prices, the more severe the inflation problem becomes, the fewer rhetorical options Trump has, and ultimately, he will be forced to target data centers and the AI industry.

The reason Trump desperately wants to avoid the Democrats taking control of the House is clear. If the Democrats win the House, they can exercise their subpoena power, repeatedly summoning Trump, his family, and his key advisors to testify, bombarding them with difficult questions. If the Democrats also regain the White House in 2028, the Justice Department, armed with leads, could pursue legal action, investigating Trump's business entities.

Let's trace the logical chain: A prolonged US-Iran standoff leads to higher oil prices; rising consumer prices cause voter dissatisfaction; Trump must then court votes by regulating and taxing the AI industry.

From now until the November election, even if AI-related stocks are cut in half, it could be an acceptable price for Trump to pay in exchange for escaping endless Democratic investigations. After the election, he can easily reverse his previous statements on data centers and AI. The industry could recover, and the S&P 500 might even challenge the 10,000 mark.

But for investors, market movements are interconnected. A crash in the AI sector would fundamentally alter market expectations of its future returns. After experiencing the shock of regulation and high taxes, investors will never be able to blindly favor this track like before.

The California Dream: Where Does Liquidity Flow?

Before analyzing the impact of the planned IPOs of the three giants – SpaceX, Anthropic, and OpenAI – on global financial markets, let me first address a question: Why hasn't Bitcoin experienced a simultaneous surge despite the continued easing of dollar liquidity since the end of Q3 last year?

On November 30, 2022, ChatGPT was officially launched to the public, marking the beginning of the AI super-bubble. Almost simultaneously, the scandal of FTX founder SBF misappropriating user funds was fully exposed. Bitcoin, after hitting a low of around $15,000 that year, rallied to $125,000 by October 2025, a cumulative increase of over six times. However, during the same period, Nvidia's stock rose elevenfold, and many small-to-mid-cap tech stocks, reliant on computing power to convert electricity into intelligence, also surged. The returns from the AI sector far outpaced the crypto market, and the gap has been widening since late 2024.

Even as Bitcoin (white) reached its all-time high, Nvidia (gold) still delivered superior returns

Bitcoin (white) performed worse after its ATH, now down 50%. Nvidia (gold), the world's most valuable company, has still gained 10% since late 2025

Based on my previous logic of judging the crypto market by fiat liquidity, Bitcoin should have seen higher gains in the current environment, but the reality is quite the opposite. Where is the problem?

I used to focus on the total amount of fiat issuance but overlooked the specific destinations of that liquidity. I originally thought that liquidity would eventually flow into Bitcoin, pushing prices up. This time, my judgment was off.

My conclusion is: almost all the newly created dollar liquidity has been absorbed by the AI sector. AI is a highly capital-intensive industry. To build the massive data centers required to run AI, enormous amounts of energy must be consumed. Hydrocarbon energy, nuclear power, and renewables are converted into electricity, which is then sent to data centers to power specialized chips for model training and inference computations.

Starting in 2024, global capital expenditures on data centers began to surge, with further increases in 2025. The industry's financing needs exploded correspondingly. Based on publicly available data, the total debt financing raised by the AI-related sector from November 2022 to date is $1.5 trillion. Coincidentally, the increase in the US broad money supply (M2) over the same period was exactly $1.