The stock has a history of collapsing by 50% to 98% after every spike, regularly falling below its Dotcom Bubble high. But this time is different?
By Wolf Richter for WOLF STREET.
The stock of memory chip maker Micron Technology has a history of fantastical mania spikes that then collapse. So now there’s another fantastical mania spike, but it’s an AI mania spike, and everything else pales compared to it. Since the low of April 3, 2025, Micron’s stock [MU] exploded by 1,315%, and its market capitalization exploded from $72 billion to $1.01 trillion. Over the past 12 months, shares exploded by 854%. Yesterday, they spiked by 19%. Today, they’re up about 2% at the moment, at $916 a share.
But for Micron, spikes have invariably been followed by collapses. For example, from September 1995 to July 1996, the stock collapsed by 82%; and from Micron’s Dotcom Bubble’s peak in July 2000 ($95.13) to December 2008, so in about 8 years, it collapsed by 98% to $1.85, most of which in the first two years. There were numerous spikes followed by post-spike plunges of 50% or more. But it took the shares 18 years (till March 2018) to exceed the Dotcom Bubble high for the first time, and then shares plunged 50% again, well below the Dotcom Bubble high.
The Dotcom Bubble high wasn’t seen again until mid-2020; the stock continued to surge until mid-January 2022, but then plunged by 50%, once again below the Dotcom Bubble high. And the stock plunged by 58% again between June 2024 and April 2025, to $64.72, which was 32% below its Dotcom Bubble high 25 years earlier. At each spike along the way, it was: “This time it’s different,” and it was never different. But nothing compares to the current AI mania spike.
Trump was helping along the way. On Friday, he said at a rally, “Boy, Micron is great.”
In terms of market cap, Micron’s path to a $1 trillion stock from a $500 billion stock broke all records: only 48 trading days to reel in the second $500 billion after the stock hit $500 billion in market cap for the first time on March 17.
And we know: “This time it’s different,” and this time it is different because the amounts of dollars involved – they’re no longer tens of billions, but a trillion this time.
In terms of speed from $500 billion to $1 trillion, that 48 trading days was by far a record. Nvidia, the poster boy for this sort of stuff, took 490 trading days to get there. Nvidia is now a $5 trillion company. Now there are 12 publicly traded US companies valued at over $1 trillion, including Micron.
There is no telling how much further this AI mania will go, and what new crazy thing will happen next, and there is no telling which of these stocks will tank first by 50%-plus.
When this stuff goes up in smoke once again, some serious dollars will vanish. Even a minor drawdown of 10% of these 12 stocks combined incinerates trillions of dollars of imagined value that then becomes unimagined.
But investors are spread around the world as the entire world has piled into these few stocks to get rich quick, and they got rich quick, and when the AI stock mania deflates, as all manias eventually do, the damage will be spread around the world too. But wait, no, this time it’s different….
Enjoy reading WOLF STREET and want to support it? You can donate. I appreciate it immensely. Click on the mug to find out how:
Thank you for your article. One observation, however, is that the new artificial intelligence revolution has an enormous demand for Micron’s product. As long as the hyperscalers keep forking over all their cash in order to build out AI infrastructure, Micron stands to benefit. The company seems to be in a very rare situation of having suddenly become one of the most important corporations on this planet. It was always important, but this time, it is making the thing the big technology players need most. Besides of course, water and electricity itself… moreover, relative to future earnings, the stock seems to be still heavily undervalued. And, if the company keeps delivering and executing like it has, there is no reason to believe that their forecasts and projections are inaccurate.
So, until demand begin softening, they lose pricing power, their margins suffer, or they become too richly valued, Micron seems to have earned its rightful place in this market.
Yes, this time it’s different!
💔
It is. “Trust me.”
I’m serious, actually. There’s a reason MU has hopped so much so fast. It’s a healthy freaking company with unfathomable demand. Yeah, it’s cyclical, but after going up a bazillion percent, we can afford a nasty 50% drop after 5 years or so.
Same with the “housing bubble” that turned out to not actually be a bubble.
Your audience craves pessimism and you deliver the product very well, I think. No offense intended. Just observing from my standpoint.
Look at MU’s history, which is what this article is about. I lived through the Dotcom Bust when MU lost 98% of its value.
Memory chips are commodities, whose prices surge and collapse, which is why MU is so cyclical. And then Micron has billions of dollars in net losses. It lost $6 billion in 2023, which is what happens when memory chip prices drop again, as they always do eventually.
“but after going up a bazillion percent, we can afford a nasty 50% drop after 5 years or so.”
if it only drops 50%, it means that you still lose half of the entire Bazillion. But 50% is a small drop for MU. It did that routinely. It dropped 98% after the Dotcom bubble, which also came with unlimited demand for memory chips and endless replacement cycles lol
“Your audience craves pessimism and you deliver the product very well, I think.”
People have been accusing me here for years of wearing “rose-colored glasses” and being eternally optimistic and blind to reality or whatever. Including this very minute in another thread…
https://wolfstreet.com/2026/05/20/credit-card-delinquencies-balances-debt-to-income-credit-limits-and-collections-in-q1-2026-americans-and-their-revolving-credit/#comment-681811
So everyone gets to cherry-pick. I just look at the data.
Good thing past performance is no guarantee of future results.
Permanent datacenter infrastructure. Not a bunch of ridiculously overvalued e-commerce companies that could easily go poof overnight. The Magnificent 7 are not going anywhere.
The AI shift is fundamental and all-encompassing. I own two companies and AI makes the work faster, which means we can outdo competitors who think AI is a farce. We use it for more than software but also in developing physical products which is really, really awesome. Which reminds me, physical AI is just getting started. Efficiencies everywhere.
I do worry about lost jobs. Some companies are not satisfied with working faster. They want to throw people off the cliff and capture their salary. Frankly, though, as far as the stock market goes, that’s a “good” thing. Cloudflare is a good example. When they announced a 20% cut I felt bad things in my heart but good things in my brain, since I’m big, long and early on NET.
The AI fan boys are entertaining, you have to give them that.
Unfathomable demand? As I understand it a lot of these orders are coming from hyper scalers who are putting the servers in warehouses because the data centers aren’t fully built yet. They’re assuming it’s going to take a couple months to get the power hooked up. But it’s sounding like it could take a couple of years. That unfathomable demand could evaporate in a heartbeat if hyper scalers can’t get the power they need.
Also, everyone is trying to make LLM’s more efficient. One way is by having much smaller, much more focused models that use far less compute resources. If there’s a breakthrough or even progress in that area, the unfathomable demand could shrink rapidly.
I don’t doubt AI is here to stay, but I’ll be shocked if we haven’t found more efficient ways to run these models in the next 12 months. And that could be very bad news in the short term for companies like micron and Nvidia.
Come for the pessimism.
Stay for the information.
… soon to join Wolf’s list of imploded stocks. There are quite a few charts like this in the AI related space and everyone thinks it’s early days like a 1998 dot-com period. This is not what the broader market needs now irrational exuberance always ends in massive corrections.
Outside of MU thesis is greater adoption of AI will lead to massive earnings expansions … hence even the Canadian banks are breaking vertical now too. Boy this is crazy.
FYI Goldman just boldly issues at 8000 easy target for the SP500
Boy o boy …
What you fail to grasp is mu has many new markets other than cell phones and computers: data centres, robotics, vehicles, industrial, edge computing. HBM memory is harder to produce than cpus. Can mu go lower? yes. Can mu go higher? yes. but it certainly won’t go down 90% unless america and the world enters a depression which the way it’s headed is a possibility.
In its history, MU has always had new markets and huge and growing demand. For example, PCs used to have 512,000 bytes of memory back in the day. Now they have terabytes of memory. Cell phones went from almost no memory to hundreds of gigabytes. Data centers (pre-AI) came along with huge demand for memory chips. All kinds of commercial and industrial equipment has massive amounts of memory, etc. etc. But that doesn’t matter. What matters are chip prices. MU makes commodity chips that are priced like commodities, with prices spiking and then plunging. Chip makers over the decades have ramped up production, so supply increased in huge wages every time prices surged, and chip makers are now in a massive buildout of production, once again, to profit from the price spike of chips, and that ramp-up of production always leads to gluts and price destruction, which is why memory chip stocks are so massively cyclical.
What is different in the last few years is that we are transferring wealth from the public to the private sector. The current federal budget deficit is over 2 tril per year. That’s real money that has to appear somewhere. It is a heck of an experiment that will continue for who knows how much longer. Perhaps until bond buyers wake up and smell the … inflation?
I also lived through the .com… thats when I started investing. I bought Cisco after it dropped 50% and got my ass handed to me :).
The way to think about it imo is that Micron will probably make 100b in proftmit which justifies a 1 or 2t market cap….
But in some number of years (my guess 2-5) that will conceivably drop to low single digits or even negative when the infinity money meets the increased production.
People point at long term contracts… but those are not that useful because when the bottom falls out you cancel the long term contract and pay the penalty or get sued.
Micron could EASILY see -6b to +10b in profit in 2-/ years at which point a 30-100b dollar valuation is reasonable in a good economy… it could go down to 15 in a pessimistic market… thats a face ripper.
I have lot of MU stocks and am cashing out slowly but I think MU would fall big time.
Semis are very cyclical and AI bubble would burst. It does not mean AI is useless.
Burry just wrote an interesting article on this.
But some things to consider:
1. The market is forward looking so it will drop before actual demand drops. It will be on the perceived future demand dropping
2. The hyper scalers are in a training phase
3. Infrastructure is currently being built out, at a certain point we’ll fall into a more regular chip cycle since chips don’t need to be replaced every year.
Thing is hard to say when though the market could double again before it collapses or maybe this time is different. Haha.
Looking at past data, every time something came along that was revolutionary and people started buying up those stocks, it all collapsed. The hyperscalers will eventually realize they are overbuilding capacity and start to slow down data center builds, if not halt them entirely. This is where the memory manufacturers, the chip manufacturers, the cooling and server rack hardware companies, etc start to go down. These are the companies taking on debt to grow to meet demand. The debt they’ll struggle to pay back when suddenly they aren’t getting the numbers they expected. This time isn’t different. It will happen. It’s a matter of when.
The race to AGI is a different beast. There is a reason it’s compared to the Manhattan Project and the government is so involved.
You might disagree with the possibility, but the first company (entity?) to reaches 1.0001 * human intelligence with self replication had in principle created “God in a box” or “skynet” and has the potential to control human development moving forward. In such a scenario, the share price of Micron is as immaterial as the health of Shainkolobwe miners.
What does self replication mean for a computer program or robot? That it builds another robot like itself? Are there any very simple robots that can replicate or accomplish some part of replication?
Yes, essentially it means a superhuman AI that can design improved versions of itself–effectively perpetual compounding or “takeoff”.
I get it’s science fiction-ey and debatable whether such ability at lower levels exists at the moment in software, but the analogy is not to robots but rather intelligence itself or “minds”.
This field moves REALLY fast–in less than five years we already have models that can write better code at 100X+ the speed and available to the public for a few thousand dollars a month. Coding has gone from the job of the future to 500K layoffs so far in 2026 in the US alone.
Micron’s memory market share is ~20%. The dominant producer is Samsung by a wide margin. Once capacity utilization falls a few percentage points, prices will plummet, and all producers’ gross margins will be negative.
AI will go down as the worst investment tool ever. Open AI walked back everything they claimed would never happen. Chat gpt for dummies all at the expense of the common folk. Now, open AI has monthly payment plans with ads. Privacy is nothing today given how open AI showed its consumer. It won’t be too long companies scale back (if they haven’t already) their AI ambitions.
Ed Zitron and other skeptics are predicting that all of these companies will eventually have to drop the subscription model and go to token-based billing. That will mean much higher bills for people who are actually using AI. That might be enough to pop the bubble, but it probably won’t happen until they run out of investor cash to feed into the furnace.
To the AI boosters in this comments section, let’s see what happens when:
1) Users of AI are forced to pay the real cost of the service.
2) The internet is fully contaminated with AI slop and there is no more “low-background” training data to use.
3) Companies using vibe-coded software and customer service chatbots are competing with companies using hand-written code and live CSRs.
4) When people like Musk actually try and follow through with their IPO promises of data centers in space to meet the promised endless, exponential demand growth.
5) When AI labs must actually post a profit, and can no longer rely on their ability to vacuum up VC money just by publishing press releases of giant “deals” they’ve made which really just reflect circular financing.
I’ve been wondering about that for a while now. Let’s say that the data center infrastructure costs $20 per search. How many people would actually be willing to pay that?
It’s easy to think that “AI is making my job so much easier” when you’re not paying the full cost of it, or even close to it.
Additionally, while I do think there are some benefits to AI tools, I think there’s a lot of groupthink going on as well. That is, people are saying how great it is because they think everyone else is expecting them to think that.
A search query is not that expensive. What is expensive is training the new models. And that is an easy cost to cut as they could just slow down.
It’s critically important to separate out training costs from query costs.
It’s the training costs for developing incrementally better models that are driving up data center and query costs.
The best analogy I’ve heard is with petrochemicals–right now there are monster trucks consuming a million liters per hour (hyperscale training), which drives up the cost of fuel (cpus, memory, energy, talent) for mundane query work.
In principle models will reach really steep points of diminishing returns (0.99998 to 0.99999) that don’t justify further investment, or it will get to 1.0001 and start designing ways to make itself more efficient at an exponential rate.
Love Ed Zitron. He’s wordy (much wordier than Wolf), but similar level of attention to detail. Another voice of reason in an insane world
Many corporations are being billed per token already
That’s right, and what I’ve heard – and again this is from the skeptic’s side – is that these companies are burning through their token budgets in no time. It’s a meme right now that these managers, who don’t really understand AI but are frantic about not missing the boat, are forcing their employees to burn as many tokens as possible, and are even basing performance reviews on it.
I don’t think we’ll get a good idea of how much AI is actually helping these companies until we’re past the mania phase and their executives and shareholders have to actually look at the cost-benefit.
Personally, in the case of software, I’m not a professional but I’ve done a lot of coding. I can’t imagine getting rehired to one of these jobs where they laid a bunch of people off to vibe code everything, and now I have to come back and deal with a bunch of AI spaghetti code. What a nightmare.
Matt B It’s the exact opposite actually–we let Claude code loose on github and it made the code 30% more efficient in a day, while also discovering hundreds of bugs we hadn’t found after working with the code for thousands of hours. You think Mythos, which is finding vulnerabilities in some of the most “reviewed-by-humans” code in the world, is a hoax?
Spot on. Replacing humans with human mimics for $$$$$? Anyone who thinks like that is profoundly ill.
Companies are already doing this and the CEO/CFO of Uber said they blew through their entire AI budget in four months with little to show in ROI from actually using it. The comments from Uber executivea are not the only ones saying that either.
Microsoft says Bing search is powered by AI (in partnership with OpenAI). I did a search for AT&T and it was giving me search results for Anthony Bourdain. On the first page. If this is what AI can do today, I’m not impressed.
log in to your google account and just try gemini. There are a few different models you can choose from. The higher perfoming ones are limited by how much you can use them per day but they will give you a much better understanding of the capabilities. There is a big difference between the free AI slop via search and the pro models that compare their answers with search resulst.
…and don’t forget. Google doesn’t sell searches. It sells ads for those searches. They are, at the core, an advertising company. Unless they can somehow quantify how an AI search delivers 2x more value for an advertiser than a standard search, they will NEVER monetize their AI investment. Just like Amazon and AI. I am getting spammed by Amazon with AI slop to try to get me to buy more stuff. At the end of the day, I’m not spending more… since I only have so many dollars to spend. So it is a game of AI musical chairs – everyone talking about how AI is going to generate revenues, but no one able to explain where exactly those revenues are going to come from, in a macro economic way.
About 70% of Alphabet’s revenues are from ads and its ad network (the behind-the-scene tech infrastructure to deliver ads, including to this site). About 30% of its revenues come from enterprise services it sells via subscriptions. This includes Google cloud services, and the Google cloud is HUGE. It sells enterprise productivity services, including a package of “Workspace” services that include enterprise email and security services, cloud services, collaboration tools such as sharable docs, spreadsheets, presentations, and calendars, a messaging service, and video meeting services. The new thing is that now it includes a subscription for Gemini, whether you want it or not. Because of the addition of Gemini, the subscription was raised by 16.7% from $12 per month to $14 per month per account (based on pricing of my corporate subscription, large outfits probably got a better deal).
Tulipmania! I know but I cannot forego trading on this stuff daily. Been a very good few month$. Keeping a good dose of powder and some diversification just in case.
You all around here said BTC was tulip crap for like 10 years and look how it has been doing. I was on board but the stuff has gone institutional. Yeah, it is in my opinion a failed currency disguised as a strangely volatile “stored of value”, but there is no tulipmania and the same goes for AI.
“volatile “stored of value””
It is a tall midget.
Supply vs. Demand Analysis for Digital Tulips:
The digital tulip buyer of last resort might be the government.
After that, there isn’t any substantial pool of future tulip buyers.
But there are many potential suppliers of improved digital tulips!
As the older digital tulips wilt, no one will want to own them anymore.
Bitcoin is a ponzi scheme and the bubble may someday burst. We’ve moved into phase of Idiocracy where the financially uneducated out number the educated and the masses have free easy access to markets. Btc will have value for as long as other people think it has value. The moment everyone thinks it has no value it will be worthless because there is no underlying asset.
Whether that’s in a year, 10 years, or never with the continuation of people getting stock tips from social media and reddit hard to say. It’s an asset that’s only value comes from hype – but there’s definitely a lot of hype
It’s as if Micron found the cure for cancer. The bubble popping can’t happen soon enough so that we get the cure for all the collective hopium poisoning.
This won’t be Micron’s last cycle. The sheer volume of AI buildout demands upgraded parts every few years for many years to come, much like PCs did back in the good ole days. The infrastructure costs of AI dwarfs the PC and internet buildouts.
I agree. It’s permanent infrastructure. AI will pervade virtually every area of our lives in 10 years. We’ll say “How did we ever live without AI?” like we do now about the Internet. Electricity. The wheel.
MU, SNDK, Samsung… will moderate production to keep memory prices up whenever necessary. Why wouldn’t they? Yeah, it’ll be a jerk thing to do but who will stop them? Today’s stock market (and housing market) is different than 25 years ago.
It’s for the rich to get richer and somehow us normal folks get a little invitation.
My wheels do cost more now!
There is no OPEC for memory chips and never has been. Plenty of chip history now, and it shows nothing but boom-and-bust cycles.
This time is not different.
Most of these stocks cannot go up another 2x, certainly not another 10x. There isn’t enough GDP in the world for them to ever end up at reasonable earnings multiples without price haircuts.
Bonus thought: When chip prices inevitably peak, unloading of hoarded stocks will push prices down fast.
Chips are much easier to hoard than petroleum. How much of the current “chip shortage” is from people stockpiling now, in order to sell at a higher price later?
After all, that looks like a no-brainer free-money trade right now. And if you’re whale, chip stockpiling would goose your chip stock portfolio too!
Wow, you’ve got it all figured out! How do I sign up for your newsletter?
Why I’d call that parabolic.
I’d even go as far as calling it exponential. This is stuff that K-shaped economies are made of.
I’m curious who’s buying retail or institutions? Parabolic usually means retail….
But who’s going to be the first $10T company? and how fast and far do they drop after that spike?
Why, NVDA, of course. Replacing their suppliers with themselves and all that. I hope they never raise an army.
you mean replacing their customers with themselves.
Yes, totally stupid. Now go ahead and try to short it. I triple dog dare you.
None of what lead to the great financial fraud of 2008 has been addressed/fixed. There is no spoon(“market”).
Hedge accordingly.
That guy fell off the wall 6 of 6 times?
😆
You slept though the prior five episodes and just woke up? 🤣
I guess I’m going to hafta rerun the whole series so you get a chance to see all 6. They’re really hilarious.
lol! I miss too much really.
Too much life stuff going on, I glance at the market, scroll Reddit and here. Look for a stock forecast and then it seems like the day is over.
Yes Reddit will do that to you.
You will be happy to know Reddit just made it a lot harder to waste time there. Let alone read interesting things there, which I admit are hard to find.
Some of the stuff over there is just hilarious, anyhow no more wasting time with the changes. I’m def never joining a social media platform etc.
1,300% up. Ok so it drops 50%, there will still be winners who were there holding for years. What’s the difference.
PS- I wrote this to get comments/reply. I await you all. ;-)
That’s not how it works :).
If profit goes from 100b annually to 6b valuation can go from 1t to 60b very fast in a “reasonable” market.
In a panic driven bloodbath it can go much lower than 60b.
If memory is over produced *which was the case a few years ago* profit can go to -4b at which point these memory businesses can be worth almost nothing.
I bought micron when it was about 40b because its replacement value alone was over 60b and no one was even trying to replace it.
At 1t its worth more than Walmart which seems absolutely insane long term. There just isnt a world where a commodity can command such massive margins for ant reasonable period of time.
So this company’s market value is = 1/32 of U.S. GDP? I know that’s an apples vs. bananas comparison of two differnent sorts of measures. But it’s still bananas.
MU PE Ratio (TTM)
43.54x
Historical EPS Growth (5-Year Avg):
25.9% per year, outpacing its
annualized 5-year revenue growth of 19.9% due to expanding operating margins
Net Income Growth (5-Year CAGR):
49.96%, reflecting high operating leverage during industry upcycles
The market is forward looking”
MU investors obviously expect much faster earnings growth in the future
During the 2022 market correction, MU experienced a 49.8% decline
During the 2024 market correction, MU experienced a 24.3% decline
Your last number is wrong:
18/Jun/2024: $153.45
04/Apr/2025: $64.72
Declined: -57.8%
there was a stock market correction in 2024, but it only reached “official” correction territory for the tech-heavy Nasdaq Composite (MU i there)
The largest overall decline (or peak-to-trough drawdown) of the Nasdaq Composite Index during 2024 was a 13.15% correction that occurred between mid-July and early August
During the mid-July to early-August tech market correction of 2024, Micron Technology (MU) stock declined by 30.57% … I previously wrote 24.3% which appears to be wrong
i wanted to show MU volatility during market corrections
18/Jun/2024 to 04/Apr/2025 is not a market correction
But it is a large decline for MU stock
I enjoy some posters with optimism about the AI revolution. I think its refreshing to think that the housing market, bitcoin, and now AI have broken historical economic models and now the sky is the limit. The problem with a lot if these magical things is that its all economic rent seeking. The FAANGs know nothing else, Uber and Lyft used it to make sure their books are balanced, and AI will do the same in order to survive. First they have to capture the market.
I can’t wait for the AI edition of Parthenon of Exploded Stocks. Will it come out later this year or next year or the year after? Taking bets now..
Stock market continues to rise because the Fed turned treasury bonds into a risky investment via its strategy to inflate problems away.
25% inflation in five years.
Many people now think growth stocks are safer than bonds over the long term, evidenced by a shrinking equity risk premium.
Indeed. The Fed’s mission evolved into “robbing bond investors in order to pay for our politicians’ follies”.
They say the fair market value of something is what another person is willing to pay for it in an open market.
We have one person with no finance background who got their stock advice from social media or reddit buying from another who did the same. I’m actually surprised this scenario didn’t make the movie Idiocracy since so many other things did.
It’s very 2001, every retail investor thinks they’re an expert. I get offered stock tips by my hairstylist. Lol.
Thing is I thought the whole pump dump thing that happened back in chatrooms pre dotcom bubble was illegal? I thought you had to disclose positions when talking about stocks.
Regardless institutions will ride the wave and then cash out before retail knows anything has turned and retail will be the bag holder like always.
Margin is again at all time highs. Trump will try to pump things until the election after that who knows though…
There is Always Another Election right around the corner
Taiwan’s stock market became the fifth largest in the world in value overtaking India. At least 40% of the value is Taiwan Semiconductor. That’s some concentration!
It’s interesting that Taiwan’s housing market is frozen and prices are dropping due to loan restrictions because the government artificially keeps Interest rates low and can only cap the number of loans per person or business. Now that money isn’t chasing the property bubble, it’s going into the stock market. Seems eerily similar to the US.
The reason HBM stocks such as Micron are rocketing up is that AI data centers face a critical ‘memory wall’ bottleneck where GPU processing power vastly outpaces memory bandwidth and capacity. As a result, the latest Nividia GPUs cannot be fully employed until the memory bottleneck is overcome. Think about the dollars invested in the latest Nividia GPU and it cannot be fully utilized.
You had me at the bottleneck. That champagne cork is going to the moon!
And Micron isn’t even the craziest stock in the AI bubble!
Micron (MU) is _only_ up +865% in the past calendar year.
American Crystal Tech (AXTI) is up +7872%. That Is Not A Typo.
SanDisk (SNDK) is up over +4000% as I type. … Not a Typo.
Lumentum (LITE) up +1051%… not a typo.
Western Digital (WDC), a major shareholder of SNDK, is itself +928% in past year.
These are not micro-cap pump-n-dumps… these are all multi-billion dollar market caps!
And this is not copy-pasted bogus internet crap. I charted each one myself to verify.
P.S. One way to tell that the bubble is nearing exhaustion is to note that Micron now has a $1 trillion market cap. Not far from NVDA, GOOG and Apple. At this rate, if the bubble went on for another year, the AI sector stocks put together would be worth more than the combined value of all other assets worldwide!
This reminds me of the Japanese real estate boom in the 1980s, when land in Tokyo cost up to $20,000/sf. And the land under the Imperial Palace in Tokyo (1.15 sq km – less than 300 acres) was briefly “worth” more than all of California.
All great stocks. Is it too late to invest? I mean long term.
But of course! You can invest at any time the market is open.
Even better, whoever is selling to you will be happy to let go of their shares at the market price. This is much different than a buy-the-dip or bear-market trade, where people are often forced to unload in a fire sale. Here, you can take pride in bringing great joy to others, as they lock in their massive profits!
However, as they say, their past performance is no guarantee of your future returns…
Is it me, or is irrational ai optimism showing up here in the comments the true harbinger of the bubble popping?
It’s pretty clear to me now this is just another flavor of the dot com bust.
Most companies are not only spelling their own doom by piling money into ai products they don’t need, but also hamstringing their own employees and systems with what will be the tech detritus of tomorrow. It’s just like dot com. Most companies will later settle on some rational, small, but very real efficiency gain. In the meantime a lot of companies will be wiped out by this. And yes micron and nvidia and amd and all the data center construction and all the dark financing is going to crater into oblivion and wipe trillions off the books
The many comments here saying it truly is different this time, and defenders of MU price as anything approaching realistic valuation, mean we must be close to a correction. Perhaps all it will take is SpaceX to suck the last marginal buyers out of equities markets. I’m decidedly in Wolf’s corner on this argument. The sell-off is going to be historical, just as the rise has been. I wouldn’t be surprised by a 90+% drop peak to trough in some of these AI names. Anyone who thinks we have not entered the landscape of the irrational is suffering from delusions. Must be close now, given the hyperbole … but as the cliche goes, markets can remain irrational far longer than I can remain solvent. I am not shorting yet, but I am close and started accumulating names of the more marginal businesses that have been pulled along in the slipstream, doing my diligence for when the bell tolls.
Margin hit all time highs again last week.
However our president tweets and stocks jump.
I think a large broad based pullback before the election is unlikely.
One thing to note though institutional investment seemed to be between the March bottom and $500. Then retail jumped in and drove it parabolic. The institutional investors will take gains at some point and then retail will be left holding the bag. Raising price targets is a way to drive price up more before they slowly cash out.
Don’t bother arguing or trying to convince the “This time is different” crowd. They will have the same arguments akin to past bubbles, parrot talking points and will have to learn the hard way, once they get a margin call.
Wealth is relative. Many people (perhaps most on this site) don’t realize what Chuck Prince articulated so well: “As long as the music is playing you’ve got to get up and dance.”
“Wealth is relative.” Yes, there is no absolute store of wealth. That’s why we have to jump from one sector to another.
“As long as the music is playing you’ve got to get up and dance, WITH OTHER PEOPLES’ MONEY.”
FIFY
I guess I shoulda bought Micron stock in 2025…. And sold right now !!
Dang it !!
I see that I bought MU at $67 and sold at $257 this year. It was an obvious buy but was glad to cash it out at what felt like a high price. Guess I don’t have a clue about what high is!
I wish my pockets were deep enough to short it but $28000 for a one year contract at current price feels like it would bite me in the ass somehow.
For the fearless there’s a 2x bull ETF for Micron and also 1x short ETF. Could be an interesting opportunity to grow or eviscerate a pocket full of cash you don’t otherwise know what to do with.
MU was certainly widely discussed around the water coolers of America today.
“Could be an interesting opportunity to grow or eviscerate a pocket full of cash you don’t otherwise know what to do with.”
Eh,have classic 4×4’s,they are a monetary black hole,eviscerate enough cash!
Given there isn’t a lot of every day public demand for AI, I am curious how fast this bubble pops.
Don’t confuse automation( people like machines that make life easier), with AI( a chat bot version of an Encyclopedia Britannica using a basic google search)
Data centers are simply a circular money machine – The money spent to make them is from the companies supplying the parts and software, the people making the parts and software then make up some useless slop to use the machines.
No one turned on Google AI so Google killed the search engine to force people into AI as the default.
Do you get the feeling that large pools of money (hedge funds) are being operated by AI driven program trading that seem self indulged in AI connected stocks?
AI buying AI. Price goes up, AI says buy more.
It could be the first sign of the danger of AI.
Maybe.
Hello Wolf,
Will there be an article on the margin debt?
Might still be in the picks and shovel phase of AI mania. There is going to be a couple of big IPO’s and then the small cap crazy AI companies IPO’s.
So this bubble might still have legs.
Should see more and more AI related acquisitions and mergers in the stratosphere.
There is surely an awful large run-up in these AI-related stocks.
However, designing and manufacturing advanced semiconductor products is: 1) difficult to do, so supply will always be limited 2) part of life in 2026 such that a disruption in their distribution would be catastrophic and deadly to society
They printed too much.
I haven’t seen any comments on capitalism’s solution to higher prices and exploding demand…. of course that is new competitors hitting the market.
Often a lot of these new competitors will come from China. Here’s some news I saw the other day. You might want to reseacrh whats going on w/ the company ChangXin Memory Technologies (CXMT) to see what can and will happen moving forward.
—————————–
According to Tom’s Hardware, Wccftech and techspot, Chinese semiconductor firms have begun flooding the market with domestically produced DRAM and NAND chips in a move that analysts say will drive down memory and storage prices, offering consumers some much-needed relief. Reports suggest that several leading global PC component manufacturers have already started using the chips in upcoming products.
Corsair has integrated memory chips manufactured by Chinese DRAM maker ChangXin Memory Technologies (CXMT) into its next-generation memory modules. While Corsair typically sources memory chips from Micron Technology, elevated market prices have reportedly pushed the company to explore more cost-effective alternatives.
I’m curious as to the quality of these chips. I’m not ragging on Chinese manufacturing, but I’d be weary of failure rates/performance goals from these fly-by-night companies, especially in the SSD market.
The music will stop when it always does. Warsh will tighten. Right now the 10mo rate-of-change in our means-of-payment money supply is still accelerating (latest April #s)
SK Hynix expands near Perdue U with the US government support. It validates MU expansion in upstate NY, ID and W. VA. It’s a systemic change. The AI revolution caused a severe shortage and bottlenecks. A pause for 60 days with Iran beyond the 250Y jubilee can send MU > 1,600. During the pause oil might rise, not fall, due to restocking empty strategic emergency reserves, before the next “event”. Thereafter winter 2027. Oil might rise to the $150 area. Lower oil: bs !!
Memory may be the most technically demanding product to manufacture. It’s also the most brutally competitive market on earth. Large swings in the stock price isn’t a system bug, rather it’s a systemic feature of the industry.
This article and the comments is like a written description of a ride on the “Devils Canyon Mystery cave ride”.
Micron and friend are in a bubble. Read the article “Global Crossing is Reborn” by Kuppy over at Praetorian capital.
The US economy is failing and the powers that be are pumping hype into a cycle that will end in tears.
It’s obvious that AI cannot overcome its capital investment curve any time in the near future. The tip off is that AI is a long term investment and Wall Street only cares about the short term. All the hyperscalers will be force to pull back on investment by later this year or next.
Interesting website I found, and the numbers seem about correct but I wouldn’t bet my fortune on them.
isaiprofitable.com
There’s so many people on here so bullish on AI and Chips that is precisely how I know this time is NOT different! I lived through 2000 and many of the MU (and AMD, and others) busts through the years. While AI is definitely good and game changing unfortunately these markets always get out of balance and when the commodity turns direction, look out below.