Crypto
Reality Test – by Arthur Hayes
(Any views expressed here are the personal views of the author and should not form the basis for making investment decisions, nor be construed as a recommendation or advice to engage in investment transactions.)
Want More? Follow the Author on Instagram, LinkedIn and X
Access the Korean language version here: Naver
Subscribe to see the latest Events: Calendar
Am I dreaming – or is AI investing as easy as buying a subscription to Citrini Research and buying every stonk they write about?
Am I dreaming – or does the price of oil not matter to economics or politics and that is why Trump and the IRGC can continue trading rhetorical barbs over social media while ships languish in the Strait of Hormuz?
Am I dreaming – or does the 2-year US treasury yield trading 0.5% over the effective funds rate not lead to the Fed raising rates at an upcoming meeting?
Am I dreaming – or can the entire gains of AI to America accrue to a few tech bro douche nozzles?
To check whether I’m lucid dreaming about this FUBAR universe in which I inhabit, I must perform a reality test. And if this test fails, I know I am dreaming and must alter my portfolio. Let this essay be that test, and at the end of the cathartic experience that is tap-tapping on my word processor, my portfolio will have undergone immense changes or none at all.
My theory at the outset is that I am indeed dreaming. The most important variable that exerts a reflexive effect upon the whole investing complex is the price of oil and other hydrocarbons. Our ability to experience the universe hinges on the efficiency with which we convert energy into biological and or artificial intelligence. Nothing violates this truth. And while the markets may, for a blip in time, appear to negate this law, shit always comes back to bite you in the proverbial ass.
We start with oil and end with an election in Pax Americana. This story arc could produce a situation whereby the AI stock bubble pops and takes the entire crypto complex down with it. When the dust settles, then and only then, can Bitcoin rise from the ashes. Yes, I know I proclaimed we wouldn’t see $60,000 Bitcoin again, and obviously I erred, but that’s forecasting for you. But I deeply believe in having strong opinions loosely held.
Let’s begin.
Deal or No Deal – That is the Question
Politicians respond to incentives. US President Donald Trump launched a war of choice against Iran for some unknown reason. Only Trump knows why he did it; I certainly can’t discern the reason from the cacophony of justifications given every hour by him or one of his lackeys. It ultimately doesn’t matter why; we are here now, and the question is how, if at all, do Trump and the IRGC play their parts to end the conflict.[1] This is Trump’s war now, and unfortunately for him and his Team Red Republicans, he decided to strike in an election year.
Tell me the change in price of gasoline and other essential goods like food, and I will tell you who wins an election in Pax Americana. Trump and his Team Red Republicans are up against high and rising food and energy inflation, primarily driven by the closure of the Strait of Hormuz, which directly results from a war they launched against Iran with no consultation of the American public. If you are dumb, you might believe that Israel made them do it … lolz just lolz. If you think a WASP takes orders from a Jew, please read more American history. Americans are bloodthirsty, war-loving people as long as it doesn’t affect their daily lives and none of their family members return home in body bags. Trump routinely touts that ONLY thirteen brave American soldiers perished in this special military operation. That is why America wages video game war using high precision extremely expensive stand-off munitions. As much as Trump betrayed his base by launching another Middle East war with no coherent theory of victory, they have stuck by the party because it’s cool to kill some non-Christian towel heads. If you don’t think that is the case, Trump primary’d several disloyal Republican legislators, and these traitors to the party lost. The problem isn’t that his base won’t come out and support the team in November; it is that undecided voters facing higher food and energy prices because of the Iran war, decide to vote for the Team Blue Democrats. Therefore, “affordability” is Trump’s worst issue.
To bamboozle undecided voters, Trump needs to at a minimum maintain oil prices at these levels. Squashing inflation in its tracks is almost impossible at this point because supply chains are just beginning to reflect the rise in energy and other input costs. All Trump can do now is manage the perception of inflation, not the reality. This means that Trump’s motivation to accept a deal with Iran depends on the price of oil. As the oil price rises, Trump’s rhetoric shifts conciliatory. But then, if the market delivers lower oil prices in anticipation of a deal, Trump backs off because there is no reason to accept what objectively will be a worse geopolitical situation between Iran, the US, and the broader Middle East than before the war. The Trump-Iran deal is almost certain to be worse than the Obama-Iran deal, and a large part of the electorate will punish the Republicans at the ballot box for “losing” the war in their minds.
It takes two to TACO, and the IRGC is under a similar decision matrix. If the oil price rises too high, their main economic trading partner, China, puts pressure on them to make concessions to the Americans. But as soon as the IRGC signals they are ready to deal, the oil price sells off and Chinese pressure subsides.
Both the US and Iran have no incentive to meet in the middle while the price of oil is at this level. Yes, it’s up materially from pre-war levels, but it ain’t that bad yet. Regarding the rest of the commodity complex, there aren’t any large populations starving, and many countries have been able to secure supplies of critical industrial commodities from other parts of the world. This fucked up Goldilocks scenario cannot continue forever. You cannot remove 10s of percentage points of supply of a global commodity and prices remain moribund. Once excess capacity is exhausted, the system will ration supplies by ramping up spot prices. This is the doom and gloom many commodity analysts espouse. The reason the mangled chickens haven’t come home to roost is that the excess supply was quite substantial going into the war. If the US-Iran war muddle-through continues deep into the second quarter, the setup for the third quarter is ballooning spot prices for hydrocarbons and other essential commodities.
I’m going to bastardize a Winston Churchill quote, “You can always count on [politicians] to do the right thing, after they have exhausted all other possibilities.” Only when shit is truly fucked, will Trump and Iran do a deal. Which, in my mind, means this severely reduced Strait of Hormuz traffic situation will persist, possibly into early Q3.
For this thought experiment, let’s assume the price of oil slowly grinds higher with intense volatility around the trend-line. What then? How do rising oil prices and Trump’s Team Red Republican re-election campaign rhetoric move reflexively?
November Kickoff: Team Red vs. Team Blue
If the price of oil slowly rises, but not enough to force Trump and the IRGC to deal, how else can Trump snatch victory from the jaws of defeat?
These Polymarket odds indicate that the Republicans, as it stands now, will retain control of the Senate by the hairs of their chinny chin chins. But they will receive a walloping in the House.
While it looks like a foregone conclusion that the Republicans will lose control of the House, I want to challenge this view. I believe Trump still has a path to victory, and it goes through a change in his rhetoric towards data centers and AI regulation/taxation.
To set the stage, currently, the distribution between parties is the below (218 votes needed to pass a piece of legislation):
Based on current Polymarket odds, here is the projected party composition after the election:
That looks super-duper bad for Team Red. But the empire struck back with redistricting measures. When the rules of the game point to defeat, you’ve gotta change the rules. This is politics, degen. Let’s assume that Polymarket is correct, then the Republicans need to make up 19 seats. Redistricting can reduce that number.
Here is a table of the likely impact of redistricting:
Now we are down to 11 seats the Republicans must make up. Let’s move onto which races are a toss-up, and within the band of statistical error based on current polling could lean Red.
There are 35 races which could go either way. I already spoke about affordability and the limited things Trump can do to shift the narrative on the disastrous impact of inflation on undecided voters. The other issue that animates the entire electorate on both sides is the construction of data centers and AI’s impact on employment. Unless you are uber-wealthy, everyone is worried about data center inflation and AI taking their jobs. This fear manifests itself in many local moratoriums on the construction of data centers and a crescendo of talk about taxes on AI companies to fund welfare payments to the 99.99% of the population that aren’t CEOs of AI-related companies and their highly paid employees.
For electoral maths consider a few local initiatives in contested districts regarding data centers (Perplexity generated tables):
Trump can capture the remaining 11 seats needed by selling the AI tech bros down the river. All that is required at this stage is rhetoric, not concrete legislative action. The point is to convince plebes that if you vote for my party, we will deal with AI after the election. As a consummate politician, Trump can promise the world and never or partially deliver post-November. Just look at his handling of the Epstein files. Maximalist campaign rhetoric about exposing the alleged pedos, but once in office a minimalist release of materials; you might say that Pam Bondi dressed herself in black face “REDACTED”. In a similar vein, Trump could declare while campaigning that he will sign legislation to slow down data center build-up and impose an excessive profits tax on AI companies to fund another round of stimulus checks, then backtrack once the election concludes and Team Red’s rule remains.
For those of you who recoil in horror that Trump would adopt measures championed by disgusting pinko commie Democrats like AOC and Elizabeth Warren, please remember that Trump conducted the largest government handout to Americans since the New Deal.[2] He even gave welfare kings and queens in the hood a few thousand dollars and didn’t prevent them from buyin’ 40’s at the corner store. Now tell me he won’t turn on Elon and Jensen, at least for a few months, profess his desire to level the playing field between AI and humans, to save his political hide.
If Trump executes this AI pirouette, the market will not be intellectually flexible enough to consider this purely an expedient electoral caper but believe that Trump actually will curtail AI capex build-out and tax the AI bros. Almost immediately, the market will convulse, and this will pop the AI bubble. For those who do not appreciate how skittish the market could be when Trump turns on the AI bros, remember the made for social media spat between Elon and Trump?
Elon left his DOGE department and started throwing shade at Trump.[3] Trump hit back, threatening to cancel government contracts given to various pieces of Elon’s business empire. $TSLA traded down 18% intraday. What politics giveth, it can taketh away.
Obviously, this was a psyop. Elon made up with daddy, got back into his good graces, and even received an invitation to the recent Trump-Xi summit in Beijing.
Don’t worry, Elon. Daddy still loves you.
The market took the feud at face value and panic-sold. This is small potatoes to Trump signaling his Team Red Republicans would support massive taxation on AI models and agents. When a Korean politician suggested something similar, the Kospi went almost limit down the next day.
The Korean government quickly walked back the statements, and it was back to Up Only.
The broader AI story assumes corporations will grow revenues at an accelerated exponential rate in perpetuity and will face no popular backlash against the societal disruption this technology and wealth accumulation will create. Only when one is dreaming can these delusions persist. Trump can provide a reality check, but his desire to do so depends on the price of oil.
The higher the price of oil goes because of the Iran special military operation, the more limited Trump’s electoral messaging options become on how to extricate the Team Red Republicans out of the hole they dug for themselves.
But why does Trump care about whether the Democrats control the House? Because with control of the house the Democrats have the power of subpoena, and they can make life a living hell for Trump, his family, and his lieutenants by asking many uncomfortable questions under oath. With that information, assuming a Team Blue president in 2028, the newly weaponized DOJ in the opposite direction could seek retribution upon the Trump Organization, possibly using information gleaned during testimony.
To reiterate, the inability to come to a deal immediately between Trump and the IRGC ultimately leads to higher oil prices. The higher oil prices anger voters because shit is more expensive. At that point, the only hot-button issue Trump has agency over is data center build-out and AI regulation/taxation. Therefore, the only path to victory is turning on the AI tech bros. If the market falls 50% between now and November, that is a small price to pay not to face an endless stream of House Democrat subpoena requests. Trump knows that once the election is over, he can walk back any of the anti-data center build-out and AI rhetoric, and it’s back to business as usual … and hopefully S&P 500 to 10,000! The problem for your portfolio is that things are path dependent, and watching one’s portfolio dump will permanently change future expectations for AI stock returns. You can never unsee the impact progressive taxation and regulation has on the AI complex; and as such, you will not invest with such insouciance as before.
California Dreaming
Before I explain the impact of the three mega AI IPOs (SpaceX, Anthropic, and OpenAI) on the global financial markets, I must present my argument for why Bitcoin failed to rally even as the pace of dollar liquidity expansion accelerated starting in late Q3 last year.
The commercial release of ChatGPT on November 30, 2022, started the great AI bubble. Around the same time, the brazen theft of customer funds by the right kind of white boy fake vegan Sam Bankman-Fried became known as fact. Bitcoin bottomed at ~$15,000 and then 7x’d to $125,000 by October 2025. However, over the same time frame, Nvidia grew 11x, and a plethora of other smaller-cap names integral to the conversion of electricity to intelligence recorded sick gainz as well. AI stocks left crypto in the dust. And AI’s outperformance sped up from late 2024 to the present.
Even when Bitcoin (white) hit an all-time high, Nvidia’s (gold) returns were still superior.
Bitcoin (white) fared even worse post hitting its all-time high as it is down 50% since then. Nvidia (gold) even as the largest company in the world by market cap rose a respectable 10% since late 2025.
Given my underlying theory of crypto markets based on fiat liquidity, Bitcoin should be much higher. But it ain’t. What did I get wrong?
Because I’m a lazy human, I only observe and model the headline amount of fiat creation. But I do not study where it goes in great detail. I assume that eventually enough of the liquidity will find its way into Bitcoin and create Up Only. But this time I got it wrong, so back to basics.
My initial hunch is that AI sucked up all available fiat liquidity, especially of the US dollar flavor. AI, as we know it, is very capital-intensive. We must build data centers that convert electricity into intelligence. Hydrocarbons, nuclear, and renewable energy push turbines that create electricity. That electricity travels to data centers where it passes over specialized silicon chips that train models and conduct inference.
As the above chart depicts, data center CAPEX spending went ape shit starting in 2024 and ramped up big time in 2025. And as such, the capital needs increased as well. In terms of the AI CAPEX spends effect on dollar liquidity, we only care about the borrowed amount. Perplexity came up with these estimates from public disclosures:
$1.5 trillion of issued debt of various types between November 2022 and the present is a huge amount, but how does that compare to the amount of dollars created? The rough and easy way to estimate the amount of dollars created is by observing M2 over the same timeline. M2 rose by $1.5 trillion. And that, ladies and germs, is a wrap. AI sucked up all created dollars. Bitcoin never had a chance. The only reason Bitcoin rallied strongly off the Nov 2022 FTX-bankruptcy low is that the AI debt binge really kicked into gear starting in 2025. $1.3 out of the total $1.5 trillion of issued debt occurred from 2025 to the present. And low and behold, Bitcoin peaked in October 2025 right as AI CAPEX spend reached previously unimaginable heights.
This matters because if AI stocks crater, there will be no excess capital with which to invest in Bitcoin. Banks will pull back on lending, and some will find that their loans were based on fugazi cash flows. The bank loan officers will only fret about a potential lack of cash flow generation to repay their debts when the hyperscalers’ stock prices decline by -50% or more. The credit destruction is real, and it will tighten liquidity. And if the politics is negative towards AI, there will be no immediate rescue. Eventually, the government will bail out the banksters, but in my model, Trump must talk tough towards the AI tech bros and their financial enablers to hoodwink the public into believing he is serious about creating an equitable distribution of the wealth generated by the AI productivity miracle. A rescue is only politically feasible after the election in November.
The connection between Bitcoin and AI stonk prices means we must have a view on whether AI stock prices are in a bubble and when it is likely to pop, and why.
The AI Lead Balloon
There are three darts that will pierce the AI bubble. They are: higher energy costs, the inability of the market to absorb the three mega IPOs (SpaceX, Anthropic, and OpenAI), and Trump’s anti-AI rhetoric.
The goal of this whole escapade is to convert energy into intelligence as efficiently as possible. Humanity is a biological vessel to convert calories into intelligence and allow the universe to experience itself. AI is no different, but instead of Nodoguro nigiri, it converts electricity into intelligence. The electricity that powers AI data centers on the margin comes from hydrocarbons like natural gas. If the price of energy rises, it raises the cost of intelligence or token production. That directly reduces the margins of AI model companies like Google, Anthropic, and OpenAI. At a certain point, they raise the price to customers, and token usage growth decelerates. Therefore, the war waltz danced by Trump and the IRGC affects AI companies, and ultimately, if either side steps on the other’s foot and causes a spike higher in oil prices, eventually profits because of AI will materially fall.
If token usage drops alongside earnings, it calls into question the continuation of planned data center CAPEX spending. When the market questions that, it’s game over. The forward earnings multiples contract, and Rrrrrraaawww, the grizzly bear tears your head off.
Source: Goldman Sachs
Combined, SpaceX, Anthropic, OpenAI, and other tech shares expiring lockups will raise more money than all dot-com IPOs in total. It’s fucking nuts. Can the market absorb all this supply? Every AI name that has pumped over the last few years can only continue going higher if investors believe earnings growth will accelerate into the future. As no one has a crystal ball, if faith in the AI story falters, investors will pay a lower multiple for future earnings. Therefore, if these IPOs don’t pop, investors will take that as a cue that the market topped and begin selling.
I want to look deeper into the SpaceX IPO, as we have more details about the mechanics and timing than the other two. Always remember, he who sells first, sells best. Elon is the best marketer on the planet. Going first will allow him and his shareholders to extract the most capital from gullible muppets. Crypto degens will be familiar with the IPO allocation setup. Essentially, it is a low-float, high-FDV shitcoin.[4]
Per the SpaceX S-1 filed with the SEC, investors in this IPO will pay ~100x sales. ARE YOU FUCKING OUT OF YOUR FUCKING HUMAN MIND! If that’s not bad enough, the company will only float 4-5% of shares initially. Given the red-hot AI market, it almost ensures that the stock will pop. But there is a downside to this: the expectations are so high that the law of large numbers suggests it will be almost impossible to satisfy the market. SpaceX will instantly be a $1.8 trillion company and the seventh largest globally by market cap. To go up 50% would put it as the fifth largest company in the world just eking out Amazon, with nowhere near the earnings to justify membership in that rarified club. I know I’m shitting on the AI story, but Nvidia deserves to be valued so high given the insane gross margins of the business and the amount of revenue it generates. SpaceX is a speculative bet on data centers in space, which, if Semi Analysis is correct, are 4x more expensive than terrestrial data centers, with cost parity to be reached in the next decade. If he had chosen a much lower valuation, it would have been easier to paint the chart green. And crypto degens know this story well: if you cannot create the situation for secondary market punters to make money, then it’s Down Only, as the insiders have no motivated buyers to dump their newly vested shares upon.
That brings me to the vesting schedule. Between now and early September, the float will increase by 5x. This creates a tough environment for the stock to rally, as all this supply comes to market. Worse yet for stonk punters, Anthropic and OpenAI both indicated a potential listing date in September. They also will issue stock at trillion-dollar valuations.
SpaceX can do well from June until September … maybe … but when all three companies are simultaneously selling shares at insanely high multiples, it guarantees market disappointment. Merely grinding higher is not enough; the expectation is of an explosion higher in the stock price.
Finally, if the oil price refuses to fall, and the plebes blame Trump and the team Red Republicans for inflation in necessities because of the Iran war, Trump will have no choice but to adopt an anti-AI stance to win over voters. The politics of data center build-out and AI will become quite nasty starting in Q3 as American politicians jostle to win votes.
If the three darts hit true to mark, none of these IPOs will do as well as expected. And once the market believes in the unlikelihood of exponential earnings growth for all things related to AI, it will demand lower multiples for many stonks resulting in lower prices. Given the explosion of stock margin loans and lending to AI CAPEX, the banking system will take massive losses on the credit extended to the industry.
Bitcoin cannot rally in the short term if the entire world takes serious losses from the deflation of the AI bubble globally. Eventually, it will bottom, then rise as Bitcoin forecasts an increase in liquidity to put Humpty Dumpty back together again. But right now, it’s about protecting one’s crypto capital.
Before I move on to how I’m positioning Maelstrom’s equity and crypto portfolio, I want to touch on whether the Fed will loosen or tighten monetary policy.
Warsh the ?
Should newly appointed Fed chairperson Kevin Warsh’s nickname connote strength/courage or weakness/pliability? It all depends on how he navigates the contradictory position in which his august institution finds itself.
2-year treasury yield minus effective fed funds rate (shaded, white), WTI front month oil futures contract (blue)
Warsh’s pickle is simple. Trump appointed him to cut rates, and he hinted he could look through this “transitory” war-created inflation straight to the AI productivity miracle. This would allow the Fed to cut rates. Unfortunately, the market requires higher rates. The 2-year treasury yield’s spread versus the effective fed funds rate indicates whether the market believes the Fed should hike (positive spread) or cut (negative spread). At over 0.5%, the market believes the Fed, because of high and rising inflation, should hike rather than cut rates at its June 16-17 meeting.
The most likely outcome is a hold. But then the market will focus on Warsh’s press conference rhetoric, and any changes to the RMP program.[5] The market will view the hold as hawkish or dovish. A hawkish hold is just as bad as a hike because the market will discount a future of higher rates at the same time oil prices rise because of a lack of resolution in the US-Iran war, and the indigestibility of three mega AI IPOs. This toxic cocktail screams for a slight or significant correction in all risky assets.
The absolutely worst outcome for risky assets is if Trump instructs Warsh to do what the market suggests and immediately hike rates to help improve affordability for the plebes. Unfortunately, unless Warsh raises rates by percentage points immediately and sells bonds from the Fed’s balance sheet in the open market, then the Fed will remain behind the curve. This is a replay of the 1970s in that the Fed hiked aggressively but not enough to squelch inflation.
It is extremely unlikely that Warsh cuts in this environment, and with that off the table, odds are the market interprets the Fed’s actions as tightening future liquidity. This will act as another anchor to drown the AI bulls.
My view that oil will rise morphs into a bearish stance on all risk assets because of the logic presented above. Let me finish this essay with a discussion on the direction I took Maelstrom’s portfolio.
Portfolio Positions
If every piece of our existence is a derivative of energy, then I must buy energy if I believe it will get more expensive in the future. As it stands now, Trump and the IRGC still believe each can wait the other out. The amount of lost barrels of oil and cubic feet of natural gas because of the war continues to climb day after day. This is even in the face of the limited amount of shipping traffic through the strait. While the market is sanguine today, should the muddle through continue, the spike higher in energy prices becomes mathematically predestined.
There are many similar charts making the rounds, but the message is clear: inventories are at the lowest levels in recent history and declining day by day as the war persists. At a certain level the system breaks, and we don’t want to experience what that means in terms of higher energy prices.
In the best-case scenario, the war ends immediately and traffic via the Strait of Hormuz rebounds to pre-war levels. Once that happens, hydrocarbon inventory replenishing and just-in-case hoarding by sovereign nations will push up prices.
Both scenarios point to higher oil and natural gas prices three to six months out, even if in the short-term energy prices fall after the signing of a peace agreement. Therefore, the play is to go long companies that produce these essential commodities. Maelstrom’s stonk portfolio holds large positions in US-listed energy producers.
Going long energy pays off in almost all scenarios, and it is trading relatively cheap vs. the companies that require it to exist. Now I must focus on the assets whose valuations depend on cheap energy.
Can the AI complex continue to rip at $150 oil? Doubtful. Therefore, à la poubelle with my AI stonks.
If all marginal capital busted their proverbial nut into the gaping hole of AI stocks, then when these stocks fall rapidly, investors cannot purchase crypto even if it outperforms in a relative sense on the way down. Therefore, à la poubelle with any crypto that is not a structural core position. I dumped $HYPE, $NEAR, and $WLD last week. I also dumped $ZEC because of the Orchard Pool bug. I wish I didn’t have to do that, but capital preservation is more important than capital appreciation.
That leaves Bitcoin and Ether. Ether is dead but functional. I have no immediate large capital demands that require liquidation of my Ether, so it shall stay unmolested. Because ultimately, I believe that once the AI bubble pops, it will cause a financial crisis that will usher forth the Big Print, I am confident that Bitcoin will dump then pump. I shall hold through the tempest but take some cheeky tactical short positions using derivatives in order to satiate my desire to trade. I’m still a certified degen.
Finally, if I awaken from my dream and find reality much the same, then I just took profit before my pilgrimage to Mediterranean house dance clubs. I can revisit the markets and the outcomes of my predictions in early September and buy back in at higher levels. Without the pressure from LPs to generate arithmetic returns annually, Maelstrom can focus on geometrically compounding capital over the long-term. As such, we have the luxury of investing in both the dream and the real world.
Want More? Follow the Author on Instagram, LinkedIn and X
Access the Korean language version here: Naver
Subscribe to see the latest Events: Calendar
[1] IRGC – Iranian Revolutionary Guard Corps
[2] AOC – Alexandria Ocasio-Cortez
[3] DOGE – Department of Government Efficiency
[4] FDV – Fully Diluted Valuation
[5] RMP – Reserves Management Program aka Stealth Quantitative Easing