Memory giant Sandisk — which fell 11 per cent on Monday — plunged more than 13 per cent on Tuesday. The stock has lost more than 50 per cent of its value this month, but nevertheless remains more than four times higher since the start of 2026.
There is no logic to any of that.
And we are told markets should be relied upon and that they know the answers to all questions.
If there is one thing I know for sure, it is that they don’t.
AI uses lots of momery, memory prices rise, makers of memory rise as their profits soar – their production costs have not gone up, only their selling prices.
Then, some time later, we find out that maybe AI isn’t going to be quite the thing we thought it was going to be. Thus memory prices, the margins of memory makers, might fall. Moemry maker prices fall.
Prices in markets have reacted, twice, to new knowledge about what prices should be. This is proof that markets don’t know? Rilly?
He is possibly the stupidest man alive – certainly with any reach. He doesn’t understand even simple things like this.
Pretending not to understand gets him a new column written.
He’s been doing this for years now.
And:
There is no logic to any of that.
Logic applies to statements – not events…
“AI uses lots of memory, memory prices rise, makers of memory rise as their profits soar – their production costs have not gone up, only their selling prices.”
Yet we also know that memory production is a massively cyclical industry. Production is constrained in the short to medium term, because expanding production is difficult and expensive. So short term prices (and profits) can rise dramatically. But we also know that higher profits will encourage extra production in the longer term, the marginal production cost of one extra chip off a production line is very low, so eventually both prices and profits will crash.
So why is the market’s reaction to react the current rise in prices and profits, but completely ignore the longer term implications? All the memory stocks have been priced by the market as if the current level of profitability will continue forever. Its all so short term. I mean its hardly that long ago that the market hadn’t caught on at all to the implications of the AI boom on memory prices. Memory stock share prices were low only a year ago.
The stock market acts like a hyperactive toddler, chasing the next shiny thing waved in front of its nose. It never thinks more than 5 minutes ahead. Its hardly a rational actor. Which is hardly surprising, given the stock market is just the aggregate of millions of humans making individual decisions. Humans are notoriously prone to pack behaviour. Markets will be no more rational than the average human being is.
The stock market has already priced in the rise in memory prices, both for the current stock and all the stock coming from the factories they’re building.
A new factory for memory takes years and multiple billion dollars to build. When the factory comes online is often just when prices are going down.
So any news that perhaps there won’t be quite as much memory sold as they thought will make the stock fall a lot.
Particularly since anyone buying memory company stock knows (or should know) quite well just how boom/bust memory prices are.
“So any news that perhaps there won’t be quite as much memory sold as they thought will make the stock fall a lot.”
But that such a day would come is obvious. Its not like its never happened before. So why bid the company stock up to such stratospheric prices in the first place? Those prices were never a rational valuation of the company.
So that if you guess the peak correctly you can sell to one of those born every minute?
One of the reasons I don’t do stocks & shares is because it seems to me to be a game of more or less predatory psychology rather than having much do do with actual production of value. Yes, a casino. And personally I’m not particularly good at that.
Keep in mind that spinning up a new memory fab isn’t quick or cheap so the market knows that the supply won’t increase in the short term. Its not irrational for a person to jump on that opportunity. As soon as it looks like demand will not remain high is when people will jump ship. That is what a rational actor will do. They’re playing a game of chicken to an extent.
“That’s not what ‘rational market’ means. Asset prices reflect all available knowledge. Those companies were worth that *on that day*.
That prices will drop is known. *When* it will happen is not. Until it is.”
Except prices (of memory)haven’t dropped have they? They are still through the roof. Yet stock prices have cratered. If the stock was worth 10X yesterday, and memory prices have not so much as twitched lower why is the stock only worthy 4X today? Either the market was being irrational when it bid them up, or its being irrational now. Which is it?
The facts have not changed from the 19th June (when the Kospi index peaked at 9000) to today when it stands at 5600. What has changed is human perceptions of those unchanged facts Which is not a rational method of valuing something. As I keep pointing out.
The valuation of a stock should reflect its current net assets plus a discounted value of its future net distributable profits offset by a risk factor. If there is a reduction in the expectation of future profit margins then a reduction in the share price will follow.
What is irrational about that?
“If there is a reduction in the expectation of future profit margins then a reduction in the share price will follow.
What is irrational about that?”
Nothing, if that revaluation is based on some actual change in the facts, not just feelz and herd instinct.
“The valuation of a stock should reflect its current net assets plus a discounted value of its future net distributable profits offset by a risk factor. “
Yes, because all the Korean domestic investors who were buying memory stocks on 2,3 and 4 times leverage were REALLY doing a rational calculation as to the value of those future cash flows weren’t they?
Memory prices are the result of contracts. “I’ll buy x units at y/unit”.
So the fact that there won’t be as much sold doesn’t matter to the one who’s committed to doing that. At most they can decide not to renew the contract when they’ve finished buying.
The stock can move up and down, while the contracts mean the actual memory bought at higher prices is either selling at a loss or more likely being held back hoping for another rise.
“prices of memory haven’t dropped have they? They are still through the roof. Yet stock prices have cratered. If the stock was worth 10X yesterday, and memory prices have not so much as twitched lower why is the stock only worthy 4X today?”
Because the estimate of how long memory prices will stay high for has dropped.
Yes, you’re right it was always known that new production would come on board and prices would drop. But that will take a few years, giving several years in which to make out like bandits.
What’s hit the prices is the unexpected drop in the estimate of likely demand over the next few years. Now it’s becoming obvious that the current crop of ‘AI’ isn’t anything like as useful as it was hyped, that’s affected the likely medium-term demand for AI data centres and therefore of chips.
I’ll give you that the swing was probably exaggerated; that’s how humans work. But that there was a swing, up and then down, is rational.
That prices of memory stocks would rise on the reporting of or even possibility of massive windfall profits is rational. Bidding them to the moon is not, its based entirely on irrationality.
At any time a stock price comprises two elements, a rational one based on a rational analysis of the known information, and an irrational one, based on nothing but feelz. At no point can the price be said to be entirely rational, or based purely on all available information at the time. There is always an element of human perception involved, which can alter in a very short space of time despite no real world information changes at all. Sometimes a stock price is 99% human perception and no rationality whatsoever [cough]Spacex[cough] and sometimes its virtually all rationality. Sometimes the irrationality is negative, a decent stock can drop despite all the data being good, just because of a negative ‘vibe’ that people feel about it.
The idea that stock prices are always 100% rational and based on all the available current information is obviously nuts. There would never be mad bull markets and subsequent crashes otherwise.
Because there’s some knowledge out there that the supply constraints won’t last much longer.
Not all this knowledge is widespread. Rational markets don’t spread knowledge, the *incorporate* the knowledge of all the market participants.
Some decent amount of people out there think they have knowledge that either the demand is going to start dropping or the supply will be increased.
The market is like a voting machine in the short term and a weighing machine in the long” as Buffett (I think) said.
“Humans are notoriously prone to pack behaviour”
The thing I observe about the market is there’s a lot of people who don’t actually understand the business they’re buying or analysing. There was this whole thing about Claude Mythos finding security holes in software and it sent cybersecurity stocks down about 7%. But the truth is, the stuff Mythos could find was a tiny aspect of cybersecurity. It was unmanaged code, and most companies aren’t running much unmanaged code now. But you look at the average “tech analyst” CV, they never worked in a software business. They joined a bank as an analyst, read reports, spoke to people. The words “unmanaged code” mean nothing.
And when that’s you’re situation, what you’re going to do is opt for security in numbers. If you get it wrong, well, so did everyone else.
It’s why we now have this AI hardware reverse ferret going on. Everyone thought this would shit gold, and didn’t understand it. Now the facts are emerging, it’s melting down.
“Humans are notoriously prone to pack behaviour”
Or as Charles Mackay said in 1841
“Men, it has been well said, think in herds; it will be seen that they go mad in herds, while they only recover their senses slowly, one by one.”
He also explained the reason Murphy gets so many readers:
“During seasons of great pestilence men have often believed the prophecies of crazed fanatics, that the end of the world was come. Credulity is always greatest in times of calamity. Prophecies of all sorts are rife on such occasions, and are readily believed, whether for good or evil.”
No – his mentor, Ben Graham
And yet the best stock market investment you can make is an index fund, which is the aggregate of millions of investor decisions. Well second best. The much-maligned Tim Cook has boosted Apple’s share price ten times during his tenure. The S&P 500 only went up five times.
NO. It may be for someone who doesn’t know anything about the market and has no special knowledge/understanding of a particular company. But Investment Trusts significantly outperform index funds (my wife holds Law Debenture and Scottish Mortgage for reasons analogous to Warren Buffet wanting his widow to hold an index fund). The latest data shows that Law Debenture after expenses has outperformed the index fund *before* expenses by 33% over 10 years, SMT has outperformed by 43% over 10 years.
The FCA’s would-be hatchet job on active funds found that the money-weighted average of “active” funds (including a number of closet trackers which charged active fees for near-passive investment so the performance of genuine active funds was better) was better than the index whereas *every* index-tracker was guaranteed to underperform the index due to admin costs.
I don’t know who told you that index-trackers were good but you should ask for your money back uplifted by the average performance of a middle-of-the-road investment trust.
Active funds tend to have higher admin costs, which more than wipes out any extra gain.
You have to compare after costs. Since that’s what you the customer actually have in the end.
Possibly the active fund costs have been reduced in the decades since index funds were introduced. So perhaps they’re competitive now.
But they would only have been reduced because so many people found that index funds were giving them more money.
At the moment Apple is teasing the market about a lot of new hardware products that are good to go but just waiting IoS 27 to be ironed out. If those new products don’t match the hype, and Apple has form there especially with Siri, expect a severe market correction.
That said, I’m running the 2nd beta version of IoS 27 and they do seem to have finally cracked Siri.
Apple’s valuation not dropping with the rest of the big tech stocks is most likely because they chose not to pss all their cash and then some up the wall on AI. Add to that, AAPL doesn’t have as many leveraged positions (as a proportion of valuation) as most of its peers, so it has a much lower gamma.
They gave it a go but sensibly bought in third party tech when Apple Intelligence foundered (though that third party tech seems to like escaping its sandbox). Apple has done well selling bucket loads of Mac Minis and Studios to people who like to do AI development in house rather than buy time on servers. They’ve always been better at hardware + marketing than software.
Hey Siri, how come all my personal, private and work data ended up at Google and Anthropic?
Well no. VOO did very well for a while. But there are periods where the S&P 500 didn’t rise for a decade (e.g. dot-com bust). EWY which tracks the KOPSI is down 34% in a month
The benefit of an index fund is generally about avoiding terrible shocks. You buy ISF, you buy the FTSE 100. If Vodafone go down the pan you’ll still be ok.
VOO is actually a bad fund, like EWY is because they lack diversification. About half the KOPSI is Samsung and SK Hynix. About 1/3rd of the S&P 500 is the Mag 7.
Only I, Oracle of Ely, know the answers to all questions. And they’re all MOAR TAXXXX!!1!
Prof Potato’s knowledge of markets is literally immeasurable.
Why he is not using this knowledge so that he can live carefree on a gigantic model train or houseboat (or both) in Monaco is one of the great mysteries of our age.
I suppose that we mere mortals should just be grateful that he gives us the chance to “buy him a coffee” instead.
He doesn’t understand that markets run on Greater Fool theory. Which is ironic, I think.
A quick search indicates that they don’t currently make the memory used in data centres for AI. If only they could match his production rate of straw men.
There are a lot of people investing in tech at the moment who don’t know anything about it and they seem to be prone to panic.
I build enterprise software, have done for decades. I know more about this stuff than 95% of the population, and I avoided anything to do with AI because I just couldn’t get my head around the potential scale and profit.
I have bought some of the software companies that everyone thought would be destroyed by AI, like ServiceNow, Adobe, Salesforce, RelX, and they’ve gone up about 10% this week.
I bought some RELX shares after the Saaspocalyse crash, currently up 20% 🙂
Was that purchase rational at the time, or just based on feelz?
Feelz naturally. Do you think I did ten days of analysis, working out discounted cash flows and future returns on capital employed beforehand? Of course not. I had a gut feel the whole AI bollox was just that, we were being pushed by vested interests in a way that suited them (ie the owners of the LLMs were involved in heavy handed PR and market manipulation to make their products sound more important and threatening to existing businesses). And the herd had stampeded off a cliff and decent businesses were being marked way down as a result, in a way that would eventually turn itself around as reality reasserted itself. Very little data entered into it. I couldn’t even tell you what Relx do. It was all about gut feel of the psychology of the situation.
Another factor most people aren’t considering is that after the CHIPS act banned exports, of some Nvidia chips, China started getting nervous that the USA could cut them off, and they’re trying to replicate everything to be self-sufficient in computing. Design software, lithography machines, CPUs based on RISC-V, fabrication plants, and memory. We think of Samsung, SK Hynix and Micron, but there’s also CXMT and YMTC
in China.
I wouldn’t be surprised if we’re headed for a glut.
And we are told markets should be relied upon and that they know the answers to all questions.
Actually, what we’re told is that despite the volatility and trial and error inherent in allowing markets to operate, it still leads to better results than relying on the actions of fools who say they already know the answers to all questions.
You just have to compare HS2 with an average business project. Apart from the justification being bollocks, we then saw 3 years of season ticket decline (which wasn’t expected) and during Covid, that clown Boris pushed the button. Who in their right mind decides to go ahead when it was likely that demand was going to be affected?
There are lots of projects at Dyson, Apple, movie companies where they get shelved during development. Dyson spent a ton of money on an EV and one day decided it wasn’t going to work.
The Snowy 2 Pumped Storage Power Station was originally estimated to cost $2 billion. They’re now saying it’ll cost $42 billion!!
Of course it’s a government project so it goes on regardless.
There is this quote from Willie Brown, former mayor of San Francisco and Speaker of the California Assembly.
News that the Transbay Terminal is something like $300 million over budget should not come as a shock to anyone. We always knew the initial estimate was way under the real cost. Just like we never had a real cost for the [San Francisco] Central Subway or the [San Francisco-Oakland] Bay Bridge or any other massive construction project. So get off it. In the world of civic projects, the first budget is really just a down payment. If people knew the real cost from the start, nothing would ever be approved. The idea is to get going. Start digging a hole and make it so big, there’s no alternative to coming up with the money to fill it in.
Literally the sunk cost fallacy.
“If people knew the real cost from the start, nothing would ever be approved.”
What he means is that he wouldn’t be able to defend it.
The problem with both the USA and the UK is that we had an era of road and city rail infrastructure improvement in the 20th century, and most of it was bleedin’ obvious. You didn’t have to lie about how much the M25, M4, Severn Crossing, diesel trains were going to cost to build. The value was enormous. Everyone agreed with them.
The problem is that by the time you get to near the end of the century, we’d ticked off all the easy wins, and the things with a clear margin of value. You’re really down to vanity and political projects. HS1, HS2, trams, East-West Rail.
For a much simpler example, Bernie Sanders said that we don’t need 23 brands of deodorant or 18 brands of sneakers (trainers). Of course, implicit in that statement is the assumption that he knows just how many brands of each we do need, and if he ruled the world that’s what you’d get.
He’d probably also say that you don’t need 5 different gauges of model railway (or however many there are), and he’d ban the gauge Spud uses.
I wish I could remember how to spell moemry.
Surely, isn’t the proof of concept is that there is sufficient memory to do AI? End of.
Can his Spudness suggest a mechanism to achieve, given the starting condition is the use of memory in AIs is not a known?
Steady on. You’ll be asking him to walk and chew gum next.
AI started because of the availability of cheap memory. Cheap because market failure had created a surplus. With Spudanomics, supply of memory would have matched demand. The market creates, even when it fails. Works the other way. Market fails because it doesn’t match supply to demand.. Giving someone the opportunity to find an alternative.
A perfect system is static system, doesn’t give opportunities for change. ith Spudanomics you’d be lucky to get to the C17th.
I wonder how much of the availability of cheap memory was due to the crash in cryptocurrency, or rather the crash in video card prices due to Bitcoin and Ethereum (the two most popular ones) changing to proof of work. Which required hard drive space instead of graphics card processing power.
Off topic
cbsnews.com/live-updates/anthony-fauci-rand-paul-covid-origins-senate/
As predicted, Fauci has pled the Fifth Amendment.
Not unexpected at all. I thought I read (probably here) that the 5th isn’t available to him because he has already been pardoned (which seems reasonable) – but I’ll leave that to one of our constitutional experts.
It isn’t. But consequences will depend on what Congress chooses to do about it. And I think a lot of those Congressmen are looking at what they said and how it will look if they vote for consequences for Fauci now. And what he can testify to in return.
Breaking up gangs has many of the same problems, due to exactly the same incentives.
He doesn’t understand what is meant by ‘markets are rational’. He doesn’t understand when people talk about distributed knowledge in markets – he thinks that it means markets are acausal and can send information back through time.
Markets are able to *price* knowledge that only a few people have. Markets can move prices through time but they can’t move knowledge.
Its that simple but since he doesn’t know the definitions of the terms he’s using he screws it up again with an ignorant polemic. Its why he doesn’t see clear real-world examples contradicting his statements – he *can’t* see them or else he would have to stop and think about why that discrepancy exists. And that would come down to ‘I don’t actually know what I am talking about’.
Its sad, really. I don’t know about his personal life but he’s successfully raised at least one child so you could count that as a success but he’s been a complete and utter failure professionally. Because he’s ignorant and denies it. He could likely have been a good accountant. No one would have known he existed but he would have had a decent career.
Instead he chose the path of the lolcow.
I mean, how bad do you have to be when Jeremy Corbyn thinks you’re an idiot?
I thought it was John McDonnell who pushed Murphy off the bus?
Nah. It was a big, liquid holding for Situational Awareness, so as that hedge fund’s troubles deepened they dumped the stock. Now that the SA quoted holdings have been sold to Citadel – a stable holder, those stocks impacted have ripped higher.