Yves here. Yours truly is a very poor judge of how far a mania can go. I recall in December 1996 when Alan Greenspan wondered if dot com prices were a sign of irrational exuberance. When the Dow fell by 143 points due to his remark, Greenspan made a point of never getting in the way of that bubble again. Even so, it was remarkable to see delusion become normalized, such as valuing companies that would clearly never earn a dime based on eyeballs, or it becoming routine for mere MBA students to be able to raise $5 million with a 5 page plan for an Internet venture. Today, it has taken Ed Zitron and other financial sleuths to try to unpack the extremely unprofitable economics of AI companies, ex the chipmakers, due both to games like circular financing and obscuritanist methods of reporting.
We skipped over the latest Strategy tsuris although we covered it in detail back when doubts first arose:
🚨 STRATEGY IS NOW IN VERY DANGEROUS POSITION
The company spent $1.38 billion, nearly 61% of its total cash reserves, to buy back debt at a discount.
Strategy now has just $870 million in cash left while holding $64.8 billion worth of Bitcoin.
One sharp BTC crash could change… pic.twitter.com/v90DJRfr4b
— Crypto Rover (@cryptorover) May 26, 2026
Another cheery assessment:
Bitcoin dumped to $68K for a reason.
History is repeating itself.
Bull trap at $97K, then $83K. Exactly as expected.
Now $BTC dumps like this:
$65K → $61K → $58K → $55K → $48K.
Next stops:
→ $60K in days.
→ $48K by September.I called the $126K top in October 2025… https://t.co/CPDrE1pApp pic.twitter.com/PN30tS9aTH
— Nonzee (@0xNonceSense) June 2, 2026
By Wolf Richter, editor of Wolf Street. Originally published at Wolf Street
Bitcoin, currently trading at around $63,000, has plunged by over 50% from its all-time high last October and is down 37% year-over-year. In terms of market cap, expressed in despised inflation-wracked US fiat, about $1.2 trillion of bitcoin value has gone up into thin air.
Ethereum, the second-largest crypto by market cap, has plunged by 64% from its all-time high last August, is down 26% year-over-year, and $364 billion in market cap has vanished since the peak.
XRP, another top crypto by market cap, has plunged by 68% from its high in July last year. About $138 billion vanished.
The crypto market cap index by CoinMarketCap, which tracks all major cryptos, has plunged by $2.08 trillion since October. Over $2 trillion in value, expressed in despised worthless fiat, gone up in smoke. And yet, outside of the crypto world, there haven’t been any significant ripples.
On the other side of the equation are the massive gains these cryptos produced until their highs last year. Betting early on these cryptos was among the winningest bets of all time. Bitcoin went from zero to a market value of $2.5 trillion over the span of about 16 years, without ever having to produce any kind of product or service, revenues or profits, or even financial statements, or walk befuddled analysts through an earnings call or whatever.
Other cryptos, by now many thousands of them, came along to duplicate that feat. Cryptos were the best-ever get-rich-quick scheme of all times. But anyone can create their own cryptos, and there are many thousands of other cryptos now, many of them have been left behind for dead.
But who needs cryptos to get rich quick if the stock market’s tech sector is now providing that service for free and with less hassles?
And there are the mega-IPOs coming up, including by SpaceX, which is literally going to the moon, not just figuratively, and its valuation at the IPO price has already gone to the moon, at $1.77 trillion. At this price, it is valued at 93 times its trailing 12-month revenues of $18.7 billion, and now it’s time to sell some cryptos to shake loose some cash to buy SpaceX to hitch the next mania to the moon? With SpaceX, folks will at least get a slice of an actual company with amazing products and innovations, and not just digital units on a blockchain.
Or sell some cryptos and buy the semiconductors that have produced WTF charts on a daily basis? The other day, we talked about Micron whose market cap had spiked to $1 trillion, from $500 billion, in 48 trading days. The share price had exploded by over 850% in 12 months and by over 1,300% in 14 months. The shares rose further over the past few days, but today are down a little.
If you look closely, you can see the Dotcom Bust, during which the shares collapsed by 98%, and then remained below the 2000 high for 24 years (data via YCharts).
For a good look at the mania in the broader semiconductor space: The Direxion Daily Semiconductor Bull 3X Shares [SOXL] has spiked by 550% over the past two months, since the end of March. Over the past 12 months through yesterday, it spiked by over 1,400%, meaning it multiplied by 14. Today, it gave up some of those gains.
In turn, it’s not unusual for this triple-leveraged semiconductor ETF to collapse by over 90%, which it most recently did in 2022. In terms of the math, SOXL would have to drop by 93% this time to wipe out the 1,500% gain of the past 12 months.
Is this still a good time to chase this mania? This chart shows the percentage gains of SOXL and MU over the past 12 months (data via YCharts).
Why bet on cryptos if you can have so much fun, and so conveniently, and instantly, with semiconductor stocks that became a mania and went parabolic, and triple the fun with triple-leveraged ETFs?
But the dollar amounts are bigger with the semiconductor industry and the AI-related stocks. Therefore, the dollar amounts of the gains in portfolios are bigger, and they show up in mundane ETFs that are widely held, such as S&P 500 index funds, dominated by a dozen stocks that combined have a value of $30 trillion. And when they turn south eventually, they’ll make a much bigger dent in dollar terms.
Crypto’s 50% drop took out $2 trillion, and it didn’t produce any ripples outside of the crypto space. A 20% decline of the top 12 stocks by market cap, not including SpaceX yet, would take out $6 trillion. The total market value of the stocks in the S&P 500 is currently close to $70 trillion. A 20% drop would take out close to $14 trillion.
But as with cryptos, these stock holdings are spread around the world, not just in the US, and a portion is held by institutional investors, not just retail investors.
So any damage gets spread around the world, not just in the US, and it gets spread around investor classes, not just retail investors.
And in the past, drops of around 20% didn’t produce significant economic ripples in the US. But the much bigger drops during the Dotcom Bust did eventually produce economic ripples, though most of the economic damage occurred in the cities where these companies were located, while the ripples further afield were minor.
A trillion here, another trillion there, and one is soon talking about real money. Problem is that such money is only a smidgen more real than cryptos are. Bessent, Bessent will you be able to deal with this?
I expect Elon to be a Trillionaire for a day, perhaps for as long as six months.
Thankfully the Economy on the whole is as stable as Donald Trump.
It’s going to be a lively Summer.
Lots of people will be billionaires, when a dozen eggs cost a million.
Trillionaire will probably remain more rarefied air, as you say.
It all depends on where you ‘get in’ and when you ‘get out’, right? Get in at the beginning at $1 and riding it all the way up AND down is different than doing it with $100.
It’s even better when you can say:
As long as you make money for those with money, you are golden.
The smart money must be slowly selling into the AI mania and accumulating crypto and real assets. 🧐
2 different categories, unless you’re talking about jerrican build houses, equally flimsy,probably a lot in crypto.
Always wondered why crypto is valued in dollars.
Crypto is the replacement to fiat currencies but the replacement process is essentially a revolution and that takes time. 💵 ➡️ 🪙
“What wins revolutions are viable alternate power structures, which run parallel to, then eventually overcome and replace, existing power structures” – Atun-Shei Films on the American Revolution.
It was true during the Founding Times and is true now.
Crypto was started or better yet based on about 64 people “printing” digital tokens with an artificial limiter on its creation[tm] due to increasing computational demands. All based on an ignorant idea/belief about money/currency and a huge ideological chip on its shoulder about all governments bad and if only democracy of money and atomistic individualism traders in a no holes barred market would ring in utopia for everyone …. lmmao …
Yet to date it has violated just about every ideological axiom that was sold by its proponents. Its price is dictated by whales, it makes the 1800s free banking period look benign in social effect, takes fiat to front load it to buy it and its CapEx, zero socioeconomic good, had it all been spent on things that fix the economy, support local enterprise, and engage in the Capitalism of social up lift …. short list ..
As I have learned by my involvement in the Silver mania, (I dollar cost averaged, so I’m doing better than most,) any ‘market’ can be manipulated. [Look at today’s Comex action. Silver down eight plus percent in a day.]
Anything that can be traded on an exchange is vulnerable to Price Discovery Distortion.
Casino Manager: “What exactly is your complaint Mr. Regulator?”
Mr. Regulator: “Gambling, gambling in this Casino! I am shocked! Shocked I tell you!”
Croupier: “Your winnings Mr. Regulator.”
Mr. Regulator: “Don’t forget the ten percent for the Big Dogs.”
Croupier: “Already taken care of sir.”
“I came he for the waters”
“I was misinformed”
Israel marketing department
I’m proud to announce the launch of Bitchcoin, whose value goes up every time holders of Bitcoin kvetch about the market collapsing on packaged air.
Who would have thought that the Seinfeld of Semollians would drop like a rock?
It’s a money about nothing!
Like the show, it’s an aggressively unfunny story about aggressively unpleasant people.
Iran is taking Hormuz tolls in cybercurrency, maybe the tanking is on account of that?
You noticed lmmao …. but seriously its a multivariate issue no different to FX via development level of said nations and ideological driven contractually outcomes via force – see West. Vibrating investors are a thing too behold, milliseconds now, algo/AI driven on rumor, and not fundamentals … but yeah … rational free market hyper unenlightened Ranoid atomitaistic individuals seeking potential – the one with the most numbers denoted in tokens wins and ascends to the Ferengi heaven of the eternal treasury … where all value is stored …
Numismatic trading with no broad socioeconomic fundamentals until its traders woe about how the neighborhood is getting scary … see history … not that you are bad thingy ..
Why with the constant pithy insults?
You do it all the time and it diminishes you greatly.
You’re better than that-
I said you were not bad.
Ahhha – response in some void … until then you are and your like the worst enemy of yourselves due to a cockamamie ideology … aka Randoid thingy …
Does not even remotely mean they keep it in crypto.
Meanwhile, who knows what horrors lurk in private equity?
“extremely unprofitable economics of AI companies”
A tangent question: I’m not following the AI mania closely, but isn’t “Deepseek” free and open-source?
So putting aside the US companies financials, isnt just AI itself, from a technological viewpoint, still economically viable?
Yes, a company could choose to use Deepseek.
In that case, they would need to buy or rent the GPUs ($$$) and the electricity.
They would need enough of both to provide the throughput needed for their application.
Running LLMs on even high-end CPUs is very slow.
AI costs too much, that suggests that AI dependent on huge watt expense costs far more than it produces.
Not only is AI expensive in the broad attribute of efficient, it is not improving productivity. That is before you deal with hallucinations.
It is limited by rigid allusion to thinkin when it is basically a parrot.
DeepSeek is reputed to cost 10% what U.S. labs peddle, the effectiveness of DeepSeek remains to be proved.
It is Google in full middle school grammar paragraphs.
Imagine xAI, has leased out just shy a billion bucks worth of AI chips. I doubt it started out to be Taylor rental for AI labs.