Nvidia has current market cap equivalent to ~8 times ExxonMobil. If tomorrow ExxonMobil disappears from existence, you'd get half of the world paralysed. If Nvidia tomorrow gets replaced by a massive hole in the ground, you'd just shrug, go down the road a bit and buy AMD. Sure, they can't make cards as fast as Nvidia does, but they still work, the old cards won't suddenly stop working, and they don't even manufacture their own hardware.
The AI stocks nowadays are pumped up by pure hype, and it's inevitable that the market will recalibrate sooner or later
I'm not sure that's a fair comparison. The difference in the product lifecycle is too big. A GPU has a 3 year depreciation cycle and continues working for well over a decade after that, if I buy gas today I need new gas tomorrow. The market can react in the timespan of years, it can't react in the timespan of weeks, and having a product that isn't used up makes timelines more elastic.
If we eliminate both factors by imagining a world where GPUs just stop working every three years and where AMD doesn't have time to ramp up production we'd be pretty screwed without Nvidia, and everything depending on GPUs would quickly grind to a halt. AMD sells a tiny number of dedicated GPUs compared to Nvidia, and right now they have no spare capacity
The word "essential" is lifting your entire argument for you. If "essential" means whatever bdauvergne on Hacker News decides humans deserve to have in their lives, and nothing else, then sure, GPUs are non-essential. But that's not up to you. You don't get to pick and choose what other humans deserve, what they want, and what they are allowed to have. That's all "essential" has ever meant: whatever I, the author, whose Word is obviously Divine, think Other People deserve to have. Why even bother using that word when talking about the economy? It's meaningless. Get rid of it and your argument collapses. People want GPUs.
> If "essential" means whatever bdauvergne on Hacker News decides humans deserve to have in their lives, and nothing else, then sure, GPUs are non-essential.
You sure have a weird definition of it.
To make a quantitative claim, I'm not sure anyone would die immediately if Nvidia disappeared overnight, except maybe for a few traders. The potential long term casualties would likely be related to it possibly triggering a stock market crash, rather than first-order consequences of the company no longer delivering products.
Obviously, the disappearance of a company intimately related to logistics would be harder to mitigate.
> You don't get to pick and choose what other humans deserve
The crux of your confusion seems to be that you don't make a distinction between "deserve" and "need". Food and entertainment are both things everyone deserves, but only food is required for everyone to make it to the end of the month.
The category of "food" in economics is vast and absolutely includes things that humans don't need to live. Nobody dies if they can't buy clothing, except in very extreme cases, yet clothing is generally considered "essential". Meanwhile, people do die because they can't get jobs and become homeless, and you need an internet connection to get a job, but internet access is very rarely considered "essential" (although I suspect this is changing).
Besides, the usual definition of "essential" in economics is more about price elasticity, how consistent demand is, how spending on the category changes as income changes, etc. But whatever your parameters for that definition are, if you actually measure these things you'll see things that surprise you, and most of your results are going to be artifacts of how you categorize things. Lots of entertainment shows low price elasticity. Should dried beans and rice be in the same "food" category as foie gras? Is a Disney+ subscription essential to a working single mother of young children? Is heroin essential to a heroin addict? Are opiates essential to someone in chronic pain? Is alcohol essential to an alcoholic? Some would literally die if it were suddenly unavailable!
The category is murky, nobody can agree on what is or is not essential, nor even what its definition is: low price elasticity? necessary for life? necessary for a fulfilling life? able to be temporarily deferred in a crisis? All of these result in different lists.
> You sure have a weird definition of it.
As I feel like I've made quite clear: I do not have any definition of it, and neither do any of you. So let's not make policy decisions and economic predictions based on what is or is not "essential", please. People want GPUs, and you'll find lots of people who are more willing to give up their clothing and restaurant food than their GPUs.
Is this a kind of linguistical scorched-earth policy? I would like to know, because if we're going to be dishonest, there's plenty of other words we could start claiming have no meaning, until no meaningful conversation can happen.
You're right that I have a general issue with how people use words versus how they think they use words. Yes, there are plenty of words that have this problem. People reach for definitions as some sort of argument-ender, without realizing how much those definitions rely on what are essentially arbitrary categorizations by an arbitrary authority. Those are two other peeves of mine: bad categorization and arbitrary authority.
There is actually a theory behind all this, based largely on the critically important fact that all models are wrong, but some models are useful. But yes, I recognize the futility of trying to fight this war in comment-sized battles in tangentially related Hacker News threads.
> until no meaningful conversation can happen
I believe accepting "all models are wrong, but some models are useful" is in fact a prerequisite for meaningful conversation, because otherwise people simply aren't even arguing about the same thing. What appears to you as a "linguistical scorched-earth policy" is something I can trace logically from that statement. Most arguments are actually arguments about definitions and categories, and therefore useless. I am trying to get people to abandon the category of "essential vs. non-essential" because it, like so many others, is arbitrary.
Imagine for a second that the year was 1880. You would say that telephones aren't essential, wouldn't you? In the previous 25 centuries of recorded history we have lived without them. Nobody's going to die if they were to stop working.
And thus that the valuation of the Bell System must be based on pure hype. Right?
If they have stopped working today it would be disaster. If it stopped at that time, life would go on. It is not about technology but how society adapted around it.
"Disaster" is a very subjective term. One man's disaster is another's normality.
How many people died between 1812 and 1815 because there were no Trans-Atlantic telephone lines? About 30 thousand soldiers, wasn't it? Probably quite a few civilians as well.
I'd call the preventable death of 30 thousand men a disaster, wouldn't you? But in 1812 it was business as usual.
Would you say penicillin isn't essential, just because it's 1928 and people are accustomed to deaths from bacterial infections?
Or you could recognize that "essential" has a meaning in economic/financial terms, but that would entirely deflate the ad hominem attack you launched to avoid acknowledging that the answer to his question is: "Not really, with a few possible exceptions in some edge cases."
There's absolutely a reason to differentiate between essential and non-essential goods when talking about the economy. Why do you think the US runs a huge food production surplus? Why do you think publicly traded stock sectors include consumer staples (essential goods) and consumer discretionary (non-essential goods and services)?
> Or you could recognize that "essential" has a meaning in economic/financial terms
I do not recognize that. That is the point of my argument. A large portion of economics is rich people trying to justify their own greed as being moral. Classifying goods as "essential" vs. "non-essential" is a way of telling poor people what they're allowed to have, and always has been. A good goes from "non-essential" to "essential" only when rich people are worried they'll get guillotined if the poor don't have access to it.
I'm aware that it has a definition in terms of what people are able to stop purchasing when their income goes down, or how consumption relates to income levels in general, but the former is a problematic definition for many reasons, and the latter does not actually coincide particularly well with the categories of goods people list off when they think of "essential goods". Humans in real life just don't respond to changing conditions the same way the little econs in your head do; they way you've decided they "should".
Ever heard that humans don't "behave logically"? Yeah, that's economists with overly simplified models being annoyed that (mostly) poor people don't act the way that they've decided poor people should act. See the trend?
Ask four economists write out a list of "essential goods" and you'll get five different lists. That is not how definitions work. Ask four mathematicians whether something is a Commutative Ring or not and they'll all agree. That's a definition. "Essential" does not have a definition. Its meaning shifts depending on which group the author of the Wall Street Journal op-ed you're reading wants to villainize this time.
> Ask four mathematicians whether something is a Commutative Ring or not and they'll all agree.
Amusingly, there are at least 3 different definitions of commutative ring.
The principal issue is whether it must have a 1 (unity, ie a multiplicative inverse). Wikipedia https://en.wikipedia.org/wiki/Commutative_ring as well as most modern sources insist on this.
Finally, if you do have a 1, then sometimes people include the condition that 0 != 1, ie the trivial/zero ring is deemed not a [commutative] ring. This is somewhat hard to find, but is relatively common among people who specifically define the concept of "ring with identity" (eg Zariski+Samuel). I have also found it unqualified (ie, just in the definition of "commutative ring") in the wild, eg in "Handbook of Mathematical Logic" by Barwise or "The Math You Need" by Mack.
(I agree with people like Conrad and Poonen that rings should have a 1. And I guess that the zero ring is in fact a [commutative] ring.)
The moral of the story is: "be careful with simile and analogy". Otherwise you get your arse handed to you on a plate and your very reasonable argument gets lost in the weeds 8)
In these circles is is generally safer to stick with car analogies.
> I do not recognize that. That is the point of my argument.
And my point is that you're going to continue to be frustrated and disappointed by refusing to use the same terminology for a topic as everyone else.
> A large portion of economics is rich people trying to justify their own greed as being moral.
Nope, but that view explains most of your reasoning.
> Classifying goods as "essential" vs. "non-essential" is a way of telling poor people what they're allowed to have, and always has been. A good goes from "non-essential" to "essential" only when rich people are worried they'll get guillotined if the poor don't have access to it.
Good lord. Which of these is more essential for human life? Food, or a luxury car? Basic medicine, or a trip to a casino? No one is stopping "poor people" from buying things from either category, but people clearly prioritize one over the other when funds are limited.
> I'm aware that it has a definition in terms of what people are able to stop purchasing when their income goes down, or how consumption relates to income levels in general, but the former is a problematic definition for many reasons, and the latter does not actually coincide particularly well with the categories of goods people list off when they think of "essential goods". Humans in real life just don't respond to changing conditions the same way the little econs in your head do; they way you've decided they "should".
So you do acknowledge the actual definition, you just refuse to accept it because you'd rather rage against the machine? Have fun with that.
> Ever heard that humans don't "behave logically"? Yeah, that's economists with overly simplified models being annoyed that (mostly) poor people don't act the way that they've decided poor people should act. See the trend?
Rich people also don't behave logically, for the record. It's almost like this class war you're describing is a figment of your imagination.
> Ask four economists write out a list of "essential goods" and you'll get five different lists. That is not how definitions work. Ask four mathematicians whether something is a Commutative Ring or not and they'll all agree. That's a definition. "Essential" does not have a definition. Its meaning shifts depending on which group the author of the Wall Street Journal op-ed you're reading wants to villainize this time.
Congratulations on your discovery that economics is not a hard science.
You already said that essential does have a widely agreed upon definition above, so the rest of this rant seems odd.
> Which of these is more essential for human life? ... So you do acknowledge the actual definition ...
And you don't acknowledge the actual definition, or you would have been asking about price elasticity, not what is or is not essential for human life. Plenty of people will give up Chinese takeout before they give up GPUs.
> Food, or a luxury car?
The luxury car, because it's grandma's old classic that Tom has kept in good enough shape to drive to his job, and he doesn't know how he'll get there when it finally breaks down, but he doesn't really have a penchant for fancy food and gets by with some cheap staples he prepares at home.
> Basic medicine, or a trip to a casino?
The trip to the casino, because Judy is getting evicted, and doesn't have any friends or family she can stay with, and won't survive the winter on the street, and hitting it big at Blackjack is unlikely, but desperate times call for desperate measures. And although she does sometimes take Tylenol for headaches, she's otherwise in good health and doesn't have any ongoing medicinal needs.
What are we doing? You're deciding what other people need for their lives?
> Rich people also don't behave logically, for the record.
Of course they do, because however they behave, they have an army of op-ed writers, sycophants, and apologists who spew out post-hoc justification for their behavior, and viola, their behavior turns out to have been rational all along.
Meanwhile, if the poor (and we are always talking about the poor when we are talking about "essential") stubbornly show an unwillingness to stop purchasing some good that some economist has decided is "non-essential", they're villainized and called irrational. Many of them even have refrigerators!
> It's almost like this class war you're describing is a figment of your imagination.
You choose now, in 2025, to deny that there's a class war? That's certainly a take.
> And you don't acknowledge the actual definition, or you would have been asking about price elasticity, not what is or is not essential for human life. Plenty of people will give up Chinese takeout before they give up GPUs.
You just listed two examples of discretionary items (prepared food is generally not considered a consumer staple).
> The luxury car, because it's grandma's old classic that Tom has kept in good enough shape to drive to his job, and he doesn't know how he'll get there when it finally breaks down, but he doesn't really have a penchant for fancy food and gets by with some cheap staples he prepares at home.
You're 0 for 2. That's not a luxury car, unless Grandma had a love of rare exotics that he should probably sell to buy food and a reliable car.
> The trip to the casino, because Judy is getting evicted, and doesn't have any friends or family she can stay with, and won't survive the winter on the street, and hitting it big at Blackjack is unlikely, but desperate times call for desperate measures. And although she does sometimes take Tylenol for headaches, she's otherwise in good health and doesn't have any ongoing medicinal needs.
And now you're 0 for 3. Your earlier comments are making more sense.
> What are we doing? You're deciding what other people need for their lives?
Most humans understand what is essential to sustain human life and what is not. In fact, every functioning adult I've ever met does.
The implications of the fact that you don't can be left for you or other readers to decuce.
> Of course they do, because however they behave, they have an army of op-ed writers, sycophants, and apologists who spew out post-hoc justification for their behavior, and viola, their behavior turns out to have been rational all along.
> Meanwhile, if the poor (and we are always talking about the poor when we are talking about "essential") stubbornly show an unwillingness to stop purchasing some good that some economist has decided is "non-essential", they're villainized and called irrational. Many of them even have refrigerators!
Not remotely true.
> You choose now, in 2025, to deny that there's a class war? That's certainly a take.
Whether there is or isn't a class war is debatable, but the one you've concocted that is led by economists is certainly not happening.
So an "old classic" (I was thinking a Mercedes) stops being a luxury car as soon as a poor person inherits it? Do you see how nebulous these categories are? How easy it is to just move the line whenever it suits your argument?
> Whether there is or isn't a class war is debatable, but the one you've concocted that is led by economists is certainly not happening.
I'll give you an economist who led the charge: Donald Regan, Ronald Reagan's secretary of the treasury and chief of staff. This was back in a time when people felt they needed to come up with a plausible-sounding excuse to strip poor people of their rights and keep them stuck in a cycle poverty. The banner was "Reaganomics", or "Trickle-Down Economics". You're so incredulous that the discipline of economics has anything to do with a class war that was largely started and still fought under the name "Trickle-Down Economics"?
I think you're missing part of the story with stocks like NVDA. The value of a stock is also based on expectations, so it could be that all of NVDA's growth for the next decade has already been frontrun by the market, and it's essentially partly a prediction market on how valuable the infrastructure will be for AI, given that the chip requirements will only increase as AI systems are implemented in more and more hardware (robots / appliances / transportation / medicine, etc.). While the growth potential of carbon fuels really remains as it is with modest growths aligned with demand/population, but tempered by alternate energy taking greater and greater marketshare.
So it could simultaneously be hype (very optimistic predictions) and yet still valued appropriatey by the market with future expectations priced in, just with some additional premium due to that demand/hype.
> Nvidia has current market cap equivalent to ~8 times ExxonMobil.
$XOM's current revenues are known and no one will suddenly be throwing billions at them for CapEx purposes. People are throwing billions at the general direction of $NVDA. That's the difference: which company has a better change of (growing) more revenues and profits in the future?
> The AI stocks nowadays are pumped up by pure hype, and it's inevitable that the market will recalibrate sooner or later
And until that recalibration happens you can buy now and see your holdings go up; then, once you're happy with the ROI (10%? 20%? More?), you can sell and realize your capital gains and have a large number in your account. Or you can not buy now and potentially miss the ride up.
Just because The Market™ is (allegedly) irrational does not preclude the possibility you can make money.
> If tomorrow ExxonMobil disappears from existence, you'd get half of the world paralysed. If Nvidia tomorrow gets replaced by a massive hole in the ground, you'd just shrug, go down the road a bit and buy AMD.
That's the wrong way to think about it, because the stock price is about all future profit over time, not the current moment. Over the next 50 years, which one do you think will have made more profit?
Imagine you're in the year 1900, and you're comparing a light bulb company with a steam engine company. Industry needs steam engines, you say! Half the world would paralyze if they stopped working! Meanwhile, who cares about a light bulb company?
But you can understand why light bulbs actually turned out to make much more money moving forwards.
You should probably think of it with a different perspective. The long term outlook for oil majors is stagnating at best. They provide great logistics for fuel globally but haven't expanded past that.
If we continue to go to electricity and go via solar, energy storage, wind and nuclear - you end up in a spot where oil and gas are limited.
NVIDIA has blue sky ahead of it -- are valuations totally out of line? Most likely. That said it has a highly desirable product globally. Oil is a valuable commodity but there are many other providers that could snatch up exxonmobil share.
Also if you want a better example -- good look at any critical supplier in the food space. Thats way more important - we lose that we get in a world of hurt.
All you're really highlighting is the difference (from an economic perspective) between needs and wants. The world needs petroleum. But once that need is met, extra petroleum production is pointless.
Whereas the world (or, perhaps, a specific class of investors within a specific segment of the world) WANTS generative AI. The amount someone wants something (and, by extension, is willing to pay for it) is potentially unbounded, and can even be uncorrelated with real utility. (See: gemstones, trading cards, cryptocurrency...)
I wouldn't be so quick to pick on AI hype. Investors are always desperately looking for where to put their money. That it is AI suggests, perhaps, that all the alternatives look less promising right now? (And that is a bit tragic.)
I hope this doesn't come across as pedantic and negative, but there is a good reason Why a resource extraction company in a market that topped out a few years ago has got a lower than average PE ratio.
On the other hand, Nvidia is a result of the AI bubble. Oddly, though, there's a case to be made that Nvidia could come out of this, even after a correction, looking pretty good.
But what I really can't grok is how Tesla keeps an insane P/E ratio after several consecutive quarters of bad news. Or how Grok gets a high valuation without even anything close to OpenAI's money-losing revenue levels, while swallowing a decrepit old social media site. Or how that big rocket can keep blowing up without dinging the valuation.
The way it's written in English it has to refer to production speed. The context is also about economics.
"Make thing as fast as" = "make" is fast. Versus "Make thing that is as fast as" = now the thing is fast. Or use a word like performant which is less ambiguous and would obviously refer to the chips.
Can rephrase slightly and it's even more obvious: "I make chips faster than you". Or, "I make chips that are faster than yours".
The AI stocks nowadays are pumped up by pure hype, and it's inevitable that the market will recalibrate sooner or later