90% of the comments in this thread make it clear that knowing about technology does not in any way qualify someone to think correctly about markets and equity valuations.
He recently said that evil people can’t survive long as founders of tech companies because they need smart people to work for them and smart people can work anywhere. There are lots of other examples. Especially read his recent tweets/essays that aren’t about his area of expertise.
Look it up yourself. Paul Graham does not have a core competency regarding what smart people are or are not willing to do. That would be sociology or psychology or economics.
Just realised that this is a subtweet of PG's post.
So I think what he's saying here, with some brevity, is: startups require lots of smart people to succeed. If the boss isn't nice to them, then those people can easily find work elsewhere. Therefore, the boss will have to be nice to them. Therefore, the boss can't be a horrible person.
Therefore, the boss will have to be nice to them. Therefore, the boss can't be a horrible person.
You should study up on logic. You are saying that since person A has to be nice to people who work for startups they can’t be a horrible person. This means that if a person is horrible they can’t be nice to people who work for startups. This is egregiously wrong.
I think what PG is saying in this situation is, he's using "evil" as shorthand for "is not nice to their employees", whereas you're reading it as "is fundamentally bad in every way".
Returning to this: even if he were somehow "misrepresenting" the idea of wealth tax - and there are many variants - that does not make him wrong.
My original point was that when PG is writing outside of his core competencies, it's usually about subjective opinions. Such as, is wealth tax a good or bad thing? That is a subject about which you can have an opinion, but there is no absolute.
My concern is that when people say "he doesn't know anything when he talks about, say, economics or politics" what they really mean is "I disagree with him and therefore he is wrong." (I disagree with PG on many such things, but that's like, just my opinion, man)
I guess I'm invested in the debate because I want people to be more open-minded and charitable and not less.
Isn’t PG using that as an example and actually says it’s simpler to calculate that way?
This doesn’t feel like a generous reading.
I’m not against a wealth tax and I’m not aligned with PG, but I do think it’s important to be generous in your interpretation. I think PG is just arguing wealth taxes are unfair on founders.
$2m starting point is reasonable btw, that’s a good ballpark for a founders share at pre seed.
I think it's presented in a way that makes him sound very reasonable, and make a wealth tax sound very stupid.
- A small time founder losing 2/3rds of their stock - OH NO, wealth tax is terrible!
- A businessman worth $162 of stock ends up being worth billions as their stock appreciates, despite having to pay wealth taxes. Huh, maybe wealth tax is alright.
But every successful SV founder and or VC is not only a tech genius but also a geopolitical and socioeconomic expert! That’s why they make war companies, cozy up to politicians, and talk about how woke is ruining the world. /s
In fairness, 'geopolitical experts' may not really exist. There are a range of people who make up interesting stories to a greater or lesser extent but all seem to be serially misinformed. Some things are too complicated to have expertise in.
Indeed, while the existence of socioeconomic experts seems more likely we don't have any way of reliably identifying them. The people who actually end up making social or economic policy seem to be winging it by picking the policy that most benefits wealthy people and/or established asset owners. It is barely possible to blink twice without stumbling over a policy disaster.
>In fairness, 'geopolitical experts' may not really exist.
Except for, I don't know, the many thousands of people who work at various government agencies (diplomatic, intelligence) or even private sector policy circles whose job it is to literally be geopolitical experts in a given area.
There are thousands of gamblers whose job is to literally predict the tumbling random number generators in the slot machines they play, and will be rewarded with thousands of dollars if they do a good job.
They are not experts. As said above, some things are too complicated to have expertise in.
It's plausible that geopolitics may work the same way, with the ones who get lucky mistaken for actual experts.
Absolute rubbish. There's lots of factual information here you can know and use to make informed "guesses" (if you will).
People like Musk, who are often absolutely clueless about countries' political situations, their people, their makeup, their relationships and agreements with neighboring countries, as well as their history and geography, are obviously going to be terrible at predicting outcomes compared to someone who actually has deep knowledge of these things.
Also we seem to be using the term "geopolitics" a bit loosely in this thread. Maybe we could inform ourselves what the term we are using even means before we discount that anyone could have expertise in it[1]. I don't think people here meant to narrow it down to just that. What we really seem to concern ourselves with here is international relations theory and political sciences in general.
Now whether most politicians should also be considered experts in these areas is another matter. From my personal experience, I'd say most are not. People generally don't elect politicians for being experts - they elect politicians for representing their uninformed opinions. There seems to be only a weak overlap between being competent at the actual job and the ability to be elected into it.
> There's lots of factual information here you can know and use to make informed "guesses" (if you will).
The gambler who learned the entire observable history of a tumbling RNG will not be in a better position to take the jackpot than the gambler who models it as a simple distribution. You cannot become an expert on certain things.
Geopolitics may or may not be one of these things, but you've made no substantial argument either way.
Geopolitics is a complex system. Having lots of factual and historical information to inform your decision is not obviously an advantage over a guess based on a cursory read of the situation.
It is like economists - they have 0 predictive power vs. some random bit player with a taste for stats when operating at the level of a country's economy. They're doing well if they can even explain what actually happened. They tend to get the details right but the big picture is an entirely different kettle of fish.
Geopolitics is much harder to work with than economics, because it covers economics plus distance and cultural barriers even before the problem of leaders doing damn silly things for silly reasons. And unlike economics there is barely the most tenuous of anchors to check if the geopolitical "experts" get things right even with hindsight. I'd bet the people who sent the US into Afghanistan and Iraq are still patting themselves on the back for a job well done despite what I think most people could accept as the total failure of those particular expeditions.
I thought Peter Zeihan was a geopolitical expert until he started talking about things I lived through, with complete ignorance of the basics. It's not that his take was wrong, it's that his basic underlying assumptions were all wildly different from reality on the ground.
Any sort of geopolitical expert is generally going to be labeled as such because he works in the domain at a reasonably high level.
The problem with that is that when at such a level, political factors start to come into play.
The net effect is that in any conflict, the winning side will have competent and qualified expert geopolitical analyses, while the losing side will have propagandists.
So the geopolitical expert is, at best, a liminal species.
That's so wrong in so many levels, also cynical. If the world worked by what you described, we would have been already obliterated ourselves a long time ago, or mass-enslavery would have happened. It didn't.
Geopolitics can be studied and learned, and is something that diplomats heavily rely upon.
Of course, those geopolitical strategies can play in certain ways we don't foresee, as on the other side we also have an actor that is free to do what they want.
But for instance, if you give Mexico a very good trade agreement as a strong country like the US, it's very likely that they will work with you on your special requests.
They may exist, but the real expertise is mostly kept non-public. Regarding the Ukraine war, both pro-Russian and pro-American public pundits never mentioned economic and real strategic issues apart from NATO membership for almost 2.5 years.
Then Lindsey Graham outright mentioned the mineral wealth and it became a topic, though not a prominent one.
Access to the Caspian Sea via the Volga-Don canal and the Sea of Azov is never mentioned. Even though there are age old Rand corporation papers that demand more US influence in that region.
The best public pundits get personalities and some of the political history correct (and are entertaining), but it is always a game of omission on both sides.
So, you think the system is genuinely trying to identify expertise to achieve equitable outcomes, and just happening to fail at it? Rather than policy being shaped by personal networks and existing power structures that tend to benefit themselves?
I think the system has been carefully configured to benefit wealthy people and/or established asset owners. But the reason that there is no effective resistance to that is because identifying generalist socioeconomic experts is practically impossible.
With the crowdstrike outage earlier last year it was incredible how many hidden security and kernel "experts" came out crawling from the woodwork, questioning why anything needs to run in the kernel and predicting the company's demise.
They were correct that there is no need for it to run in the kernel. They were incorrect in thinking this would affect the company's future, because of course the sales of their product have nothing to do with its technical merit.
I think you've got it half correct: sales absolutely does have to do with the technical merit. Their platform works, it's just folks overestimated the impact of a single critical defect.
Nobody would pay crowdstrikes prices if it didn't stop attacks, or improve your detection chances (and I can assure you, it does, better than most platforms)
> Nobody would pay crowdstrikes prices if it didn't stop attacks, or improve your detection chances
In my experience people pay because they need to tick the audit box, and it's (marginally) less terrible than their competitors. Actually preventing or detecting an attack is not really a priority.
And yet crowdstrike's stock price is still 28% up on where it was 12 month ago, 46% up on 6 months ago after their crash.
Sibling is right, that type of product is nothing to do with actually preventing problems, its to do with outsourcing personal risk. Same as SAAS. Nobody got fired when office 365 was down for the second day in a year, but have a 5 minute outage on your on-prem kit after 5 years and there's nasty questions to answer.
The crash is absolutely rational; the cascading effect highlights the missing moat for companies like OpenAI. Without a moat, no investor will provide these companies with the billions that fueled most of the demand. This demand was essential for NVIDIA to squeeze such companies with incredible profit margins.
NVIDIA was overvalued before, and this correction is entirely justified. The larger impact of DeepSeek is more challenging to grasp. While companies like Google and Meta could benefit in the long term from this development, they still overpaid for an excessive number of GPUs. The rise in their stock prices was assumed to be driven by the moat they were expected to develop themselves.
I was always skeptical of those valuations. LLM inference was highly likely to become commoditized in the future anyway.
It has been clear for a while that one of two things is true.
1) AI stuff isn't really worth trillions, in which case Nvidia is overvalued.
2) AI stuff is really worth trillions, in which case there will be no moat, because you can cross any moat for that amount of money, e.g. you could recreate CUDA from scratch for far less than a trillion dollars and in fact Nvidia didn't spend anywhere near that much to create it to begin with. Someone else, or many someones, will spend the money to cross the moat and get their share.
So Nvidia is overvalued on the fundamentals. But is it overvalued on the hype cycle? Lots of people riding the bubble because number goes up until it doesn't, and you can lose money (opportunity cost) by selling too early just like you can lose money by selling too late.
Then events like this make some people skittish that they're going to sell too late, and number doesn't go up that day.
One thing you’re missing is that there’s nothing that says the value must correct. There are at least two very good reasons it might not: Nvidia now has huge amounts of money to invest in developing new technologies, exploring other ideas, and the other is that very little of the stock market is about the actual value of the company itself, but speculation. If people think it will go up, they buy it, reducing supply, and driving up the price. If people think it will go down, they sell it, increasing supply and driving down the price. It is a self-fulfilling prophecy on a large scale, and completely secondary to the actual business.
> Nvidia now has huge amounts of money to invest in developing new technologies
This is not actually a reason for investors to invest in a company, because it's caused by investors investing in the company. If the market would invest in some other company instead then that company would have huge amounts of money to invest in developing new technologies. Meanwhile the ones that tend to succeed in that are more often new, nimble companies breaking into or creating a new market rather than large established ones with bureaucracy, internal politics and fear of cannibalizing existing sales.
Example: If there is a popular new application for consumer GPUs that requires a lot of VRAM, a competitor could make a lot of money by developing consumer GPUs with a lot of VRAM, but Nvidia would have to worry about that eroding sales of enterprise GPUs. Then investing in the competitor could have a better return, both because of potentially higher growth (people invest $5B in developing the GPU and then it becomes a $100B+ company, huge ROI; very little chance of Nvidia going from $3T to $60T), and because when it happens it comes at the expense of the incumbent, who loses not just the consumer GPU sales but the enterprise ones to the competitor selling for consumer prices. Which means the incumbent still has a very significant risk of losing value, but without as much potential upside.
People often try to make this argument by pointing to Microsoft or Apple, but those are major outliers who got there through anti-trust violations. Meanwhile Kodak, Xerox, Yahoo, AOL, Sears, IBM, GM, GE, etc.
> very little of the stock market is about the actual value of the company itself, but speculation
That's the hype cycle. We know which section of the graph we're on right now.
Eventually people will sell their stock to invest in some business that is actually growing or giving proportional dividends.
Of course, that "eventually" there is holding a way too much load. And it's very likely this won't happen in a time the US government is printing lots of money and distributing it to rich investors. But that second one has to stop eventually too.
It's a lot of people holding the stock, you are expecting everybody to just not do it.
Private companies are different, but on publicly traded ones it tends to happen.
(Oh, you may mean that printing money part. It's a lot of people holding that money, eventually somebody will want to buy something real with it and inflation explodes.)
Yeah the printing money bit. Generously one might even say that that’s the reason for printing more money: make sure that the value of peoples investments decays over time so there’s no need for the market to crash to “get the money back out”.
Related to your #2. I mentioned this elsewhere yesterday, but NVDA's margins (55% last quarter!) are a gift and a curse. They look great for the stock in the short term, but they also encourage its customers to aggressively go after them. Second, their best customers are huge tech companies who have the capital and expertise (or can buy it) to go after NVDA. DeepSeek just laid out a path to put NVDAs margins under pressure, hence the pullback.
2) Seems the most plausible, but how to value the moat, or, how long / how many dollars will it cost to overcome the moat? The lead that CUDA currently has suggests that it's probably a lot of money, and it's not clear what the landscape will look like afterwards.
It seems likely that the technology / moat won't just melt away into nothing, it'll at least continue to be a major player 10 years from now. The question is if the market share will be 70%, 10% or 30% but still holding a lead over a market that becomes completely fractured....
I think the analysis of (2) is too simplistic because it ignores network effects. A community of developers and users around a specific toolset (e.g. CUDA) is hard to just "buy". Imagine trying to build a better programming language than python -- you could do it for a trillion dollars, but good luck getting the world to use it. For a real example, see Meta and Threads, or any other Twitter competitor.
You have a trillion dollars in incentive. You can use it for more than just creating the software, you can offer incentives to use it or directly contribute patches to the tools people are already using so they support your system. Moreover, third parties already have a large motivation to use any viable replacement because they'd avoid the premium Nvidia charges for hardware.
You could apply this analysis to any of the other big tech innovations like operating systems, search, social media, ...
MS threw a lot of money after Windows Phone. I worked for a company that not only got access to great resources, but also plain money, just to port our app. We took the money and made the port. Needless to say, it still didn't work out for MS.
Those markets have a much stronger network effect (especially social media), or were/are propped up by aggressive antitrust violations, or both.
To use your example, the problem with entering the phone market is that customers expect to buy one phone and then use it for everything. So then it needs to support everything out of the gate in order to get the first satisfied customer, meanwhile there are millions of third party apps.
Enterprise GPUs aren't like that. If one GPU supports 100% of code and another one supports 10% of code, but you're a research group where that 10% includes the thing you're doing (or you're in a position to port your own code), you can switch 100% of your GPUs. If you're a cloud provider buying a thousand GPUs to run the full gamut of applications, you can switch what proportion of your GPUs that run supported applications, instead of needing 100% coverage to switch a single one. Then lots of competing GPUs get made and fund the competition and soon put the competition's GPUs into the used market where they become obtainium and people start porting even more applications to them etc.
It also allows the competition to capture the head of the distribution first and go after the long tail after. There might be a million small projects that are tied to CUDA, but if you get the most popular models running on competing hardware, by volume that's most of the market. And once they're shipping in volume the small projects start to add support on their own.
Why can’t you just build something that’s CUDA-compatible? You won’t have to move anyone over then. Or is the actual CUDA api patented? And will Chinese companies care about that?
AFAIK, CUDA is protected. There are patents, and the terms of use of the compiler forbids using it on other devices.
Of course, most countries will stump over the terms of use thing (or worse, use it as evidence to go after Nvidia), and will probably ignore the patents because they are anticompetitive. It's not only China that will ignore them.
AMD is actively working to recreate CUDA. "Haven't succeeded yet" is very different from having failed, and they're certainly not giving up.
Intel's fab is in trouble, but that's not the relevant part of Intel for this. They get a CUDA competitor going with GPUs built on TSMC and they're off to the races. Also, Intel's fab might very well get bailed out by the government and in the process leave them with more resources to dedicate to this.
Then you have Apple, Google, Amazon, Microsoft, any one of which have the resources to do this and they all have a reason to try.
Which isn't even considering what happens if they team up. Suppose AMD is useless at software but Google isn't and then Google does the software and releases it to the public because they're tired of paying Nvidia's margins. Suppose the whole rest of the industry gets behind an open standard.
A lot of things can happen and there's a lot of money to make them happen.
While we can bet on "AMD are too sclerotic to fix their drivers even if it's an existential threat to the company", I don't think we can bet on "if we deny technology to China they won't try to copy it anyway".
You don't need external competition to have NVDA correct. All it takes is for one or more of the big customers to say they don't need as many GPUs for any reason. It could be their in house efforts are 'good enough', or that the new models are more efficient and take less compute, or their shareholders are done letting them spend like drunken sailors. NVDAs stock was/is priced for perfection and any sort of market or margin contraction will cause the party to stop.
The danger for NVDA is their margins are so large right now, there is a ton of money chasing them not just from their typical competition like AMD, but from their own customers.
The crash of NVIDIA is not about the moat of OpenAI.
But because DeepSeek was able to cut training costs from billions to millions (and with even better performance). This means cheaper training but it also proves that OpenAI was not at the cutting edge of what was possible in training algorithms and that there are still huge gaps and disruptions possible in this area. So there is a lot less need to scale by pumping more and more GPUs but instead to invest in research that can cut down the cost. More gaps mean more possibility to cut costs and less of a need to buy GPUs to scale in terms of model quality.
For NVIDIA that means that all the GPUs of today are good enough for a long time and people will invest a lot less in them and a lot more in research like this to cut costs. (But I am sure they will be fine)
This is partially why Apple is the one that stands to gain more, and it showed.
Their "small models, on device" approach can only be perfected with something like DeepSeek, and they're not exposed to NVIDIA pricing, nor have to prove investors that their approach is still valid.
I keep seeing this argument, but I don't buy it at all. I want a phone with an AGI, not a phone that is only AGI. Often it's just easier to press a button rather than talk to an AI, regardless how smart it is. I have no interest in natural language being the only interface to my device, that sounds awful. In public, I want to preserve my privacy. I do not want to have everyone listening in on what I'm doing.
If we can create an AGI that can literally read my mind, okay, maybe that's a better interface than the current one, but we are far away from that scenario.
Until then, I'm convinced users will prefer a phone with AI functionalities rather than the reverse. It's easier for a phone company to create such a phone than it is for an AI company.
Perennial reminder that we do not have any real evidence that we are anywhere close to AGI, or that "throwing more resources at LLMs" is even theoretically a possible way to get to an AGI.
"Lots of people with either a financial motivation to say so or a deep desire for AGI to be real Soon™ said they can do it" is not actual evidence.
We do not know how to make an AGI. We do not know how to define an AGI. It is hypothetically possible that we could accidentally stumble into one, but nothing has actually shown that, and counting on it is a fool's bet.
I don’t understand why this is not obvious to many people: tech and stock trading are totally two different things, why on earth a tech expert is expected to know trading at all? Imagining how ridiculous it would be if a computer science graduate will also automatically get a financial degree from college even though no financial class has been taken.
People developing statistical models that are excercising the financial market at scale are the quants. These people don't come from financial degree background.
I’ve noticed this phenomenon among IT & tech VC crowd. They will launch pod cast, offer expert opinion and what not on just about every topic under the Sun, from cold fusion to COVID vaccine to Ukraine war.
You wouldn’t see this in other folks, for example, a successful medical surgeon won’t offer much assertion about NVIDIA.
And the general tendency among audience is to assume that expertise can be carried across domains.
> You wouldn’t see this in other folks, for example, a successful medical surgeon won’t offer much assertion about NVIDIA.
Doctors are actually known for this phenomenon. Flight schools particularly watch out for them because their overconfidence gets them in trouble.
And, though humans everywhere do this, Americans are particularly known for it. There are many compilation videos where Americans are asked their opinion on whether Elbonia needs to be bombed or not, followed by enthusiastic agreement. That's highly abnormal in most other countries, where "I don't know" is seen as an acceptable response.
> Anti-intellectualism has been a constant thread winding its way through our political and cultural life, nurtured by the false notion that democracy means that 'my ignorance is just as good as your knowledge.
― Isaac Asimov
> a successful medical surgeon won’t offer much assertion about NVIDIA.
You haven't meet many surgeons have you? When I was working in medical imaging, the technicians all said we (the programmers) were almost as bad as the surgeons.
This is exacerbated by the tendency in popular media to depict a Scientist character, who can do all kinds of Science (which includes technology, all kinds of computing, and math).
Systems, it’s all about systems thinking. It is absolutely true that people in tech are often optimistic and/or delusional about the other expertise at their command. But it’s not like the basic assumption here is completely crazy.
Being a surgeon might require thinking about a few interacting systems, but mostly the number and nature of those systems involved stay the same. Talented programmers without even formal training in CS will eat and digest a dozen brand new systems before breakfast, and model interactions mentally with some degree of fidelity before lunch. And then, any formal training in CS kind of makes general systems just another type of object. This is not the same as how a surgeon is going to look at a heart, or even the body as a whole.
Not that this is the only way to acquire skills in systems thinking. But the other paths might require, IDK, a phd in history/geopolitics, or special studies or extensive work experience in physics or math. And not to rule out other kinds of science or engineering experts as systems thinkers, but a surprisingly large subset of them will specialize and so avoid it.
By the numbers.. there are probably just more people in software/IT, therefore more of us to look stupid if/when we get stuff wrong.
Obviously general systems expertise can’t automatically make you an expert on particle physics. But honestly it’s a good piece of background for lots of the wicked problems[1], and the wicked problems are what everyone always wants to talk about.
But even if we just look at the examples given by the parent, most of them are not about systems or models at all. Epidemiology and politics concern practical matters of life. In such matters, life experience will always trump abstract knowledge.
Epidemiology and politics do involve systems, I’m afraid. We can call it “practical” or “human” or “subjective” all we like, but human behaviors exhibit the same patterns when understood from a statistical instead of an individual standpoint.
Epidemiology and politics are pretty much the poster children of systems[0], next to their eldest sibling, economics. Life and experience may trump abstract knowledge dumbly applied, but alone it won't let you reason at larger scales (not that you could collect any actual experience on e.g. pandemics to fuel your intuition here anyway).
A part of learning how to model things as systems is understanding your model doesn't include all the components that affect the system - but it also means learning how to quantify those effects, or at least to estimate upper bounds on their sizes. It's knowing which effects average out at scale (like e.g. free will mostly does, and quite quickly), and which effects can't possibly be strong enough to influence outcome and thus can be excluded, and then to keep track of those that could occasionally spike.
Mathematics and systems-related fields downstream of it provide us with plenty of tools to correctly handle and reason about uncertainty, errors, and even "unknown unknowns". Yes, you can (and should) model your own ignorance as part of the system model.
--
[0] - In the most blatant example of this, around February 2020, i.e. in the early days of the COVID-19 pandemic going global, you could quite accurately predict the daily infection stats a week or two ahead by just drawing up an exponential function in Excel and lining it up with the already reported numbers. This relationship held pretty well until governments started messing with numbers and then lockdowns started. This was a simple case because at that stage, the exponential component was overwhelmingly stronger than any more nuanced factor - but identifying which parts of a phenomenon dominate and describing their dynamics is precisely the what learning about systems lets you do.
It's because software devs are smart and make a lot of money - a natural next step is to try and use their smarts to do something with that money. Hence stocks.
>It's because software devs are smart and make a lot of money
They just think they're smart BECAUSE they make a lot of money. Just because you can center divs for six figures a year at a F500 doesn't make you smart at everything.
I've never met a fellow software engineer who "centers divs" for 6 figures.
But then I work with engineers using FPGAs to trade in the markets with tick to trade times in double digit nanoseconds and processing streams of market data at ~10 million messages per second (80Gbps)
The truth is, a lot of P&L in trading these days is a technical feat of mathematics and engineering and not just one of fundamental analysis and punting on business plans
If you were really smart surely you would be able to see that there are more long-term valuable things for you to do with your time than just make yourself more money...
Tech people are allowed to quickly learn a domain enough to build the software that powers it, bringing in insights from other domains they've been across.
Just don't allow them to then comment on that domain with any degree of insight.
No, nvidia's demand and importance might reduce in the long term.
We are forgetting that China has a whole hardware ecosystem. Now we learn that building SOTA models does not need SOTA hardware in massive quanties from nvidia. So the crash in the market implicitly could mean that the (hardware) monopoly of American companies is not going to be more than a few years. The hardware moat is not as deep as the West thought.
Once China brings scale like it did to batteries, EVs, solar, infrastructure, drones (etc) they will be able to run and train their models on their own hardware. Probably some time away but less time than what Wall Street thought.
This is actually more about nvidia than about OpenAI. OpenAI owns the end interface and it will be generally safe (maybe at a smaller valuation). In the long term nvidia is more replaceable than you think it is. Inference is going to dominate the market -- its going to be cerebras, groq, amd, intel, nvidia, google TPUs, chinese TPUs etc.
On the training side, there will be less demand for nvidia GPUs as meta, google, microsoft etc. extract efficiencies with the GPUs they already have given the embarrasing success of DeepSeek. Now, China might have been another insatiable market for nvidia but the export controls have ensured that it wont be.
>On the training side, there will be less demand for nvidia GPUs as meta, google, microsoft etc. extract efficiencies with the GPUs they already have given the embarrasing success of DeepSeek. Now, China might have been another insatiable market for nvidia but the export controls have ensured that it wont be.
Why? If DeepSeek made training 10x more efficient, just train a 10x bigger model. The end goal is AGI.
You are assuming that a 10x bigger model will be 10x better or will bring us close to AGI. It might be too unweildy to do inference on. Or the gain in performance maybe minor and more scientific thought needs to go into the model before it can reap the reward with more training. Scientific breakthroughts sometimes take time.
I’m not assuming 10x bigger will yield 10x better. We have scaling laws that can tell you more.
But I find it bizarre that you made the conclusion that AI has stopped scaling because DeepSeek optimized the heck out of the sanctioned GPUs they had. Weird.
I have not said that. I simply said that you now know that you can get more juice for the amount you spend. If you’ve just learnt this you would now first ask your engineers to improve your model to scale it rather than place any further orders with nvidia to scale it. Only once you think you have got the most out of the existing GPUs you would buy more. DeepSeek have made people wonder if their engineers have missed some more stuff and maybe they should just pause spending to make sure before sinking in more billions. It breaks the hegemony of the spend more to dominate attitude that was gripping the industry e.g $500 billion planned spend by openAI consortium etc
It doesn’t break the attitude. The number one problem DeepSeek’s CEO stated in an interview is they don’t have access to more advanced GPUs. They’re GPU starved.
There’s no reason why American companies can’t use DeepSeek’s techniques to improve their efficiency but continue the GPU arms race to AGI.
Baader-Meinhof phenomenon, but also because everyone is writing about GPU demand and Jevon's paradox is an easy way to express the idea in a trite keyword.
I never knew there was an actual term for this, but I knew of the concept in my professional work because this situation often plays out when the government widens roads here in the States. Ostensibly the road widening is intended to lower congestion, but instead it often just causes more people to live there and use it, thereby increasing congestion.
Probably a decent amount of professions have some variation of this, so it probably is accurate to say most people know OF Jevon’s Paradox because it’s pretty easy to dig up examples of it. But probably much fewer know it’s actual name, or even that it has a name
IMHO it happens as long as you can find use cases that were previously unfeasible due cost or availability constraints.
At some point the thing no longer brings any benefits because other costs or limitations overtake. for example, even faster broadband is no longer that big of a deal because your experience on most websites is now limited by their servers ability to process your request. However maybe in the future the costs and speeds will be so amazing that all the user devices will become thin clients and no one will care about their devices processing power, therefore one more increase in demand can happen.
The increase in efficiency is usually accompanied with the process of commoditization as stuff get cheaper to develop, which is very bad news for nvidia.
If you dont need the super high end chips than Nvidia loses it's biggest moat and ability to monopolize the tech, CUDA isn't enough.
> Nvidia loses it's biggest moat and ability to monopolize the tech, CUDA isn't enough
CUDA is plenty for right now. AMD can't/won't get their act together with GPU software and drivers. Intel isn't in much better of a position than AMD and has a host of other problems. It's also unlikely the "let's just glue a thousand ARM cores together" hardware will work as planned and still needs the software layer.
CUDA won't be an Nvidia moat forever but it's a decent moat for the next five years. If a company wants to build GPU compute resources it will be hard to go wrong buying Nvidia kit. At least from a platform point of view.
CUDA will still be a moat for the near future and nobody is saying that Nvidia will die, but the thing is that Nvidia margins will drop like crazy and so will it's valuation. It will go back down to being a "medium tech" company.
Basically training got way cheaper, and for inference you don't really need nvidia, so even if there's an increase for cheaper chips there's no way the volume makes up for the loss of margin.
No, Nvidia's margins won't drop at all and the proof for this is Apple.
The units of AI accelerators will explode, the market will explode.
At the end of the day, Nvidia will have 20-30% of the unit share in AI HW and 70-80% of the profit share in the AI HW market. Just like Apple makes 3x the money compared to the rest of the smartphone market.
Jensen has considered Nvidia a premium vendor for 2 decades and track record of Nvidia's margins show this.
And while Nvidia remains a high premium AI infrastructure vendor, they will also add lots of great SW frameworks to make even more profit.
Omniverse has literally no competition. That digital world simulation combines all of Nvidia's expertise (AI HW, Graphics HW, Physics HW, Networking, SW) into one huge product. And it will be a revolution because it's the first time we will be able to finally digitalize the analog world. And Nvidia will earn tons of money because Omniverse itself is licensed, it needs OVX systems (visual part) and it needs DGX systems (AI part).
Don't worry, Nvidia's margins will be totally fine. I would even expect them to be higher in 10 years than they are today. Nobody believes that but that's Jensen's goal.
There is a reason why Nvidia has always been the company with the highest P/S ratio and anyone who understands why, will see the quality management immediately.
Why should they invest in Nvidia now instead of investing companies which can capitalize on the applications of AI.
Also, why not invest in AMD or Intel bur Nvidia till now: Because Nvidia had the moat and there was a race to buy as much GPU as possible at the moment. Now momentarily Nvidia sales would go down.
For long term investers who are investing in a future, not now, Nvidia was way overpriced. They will start buying when the price is right, but at the moment it's still way too high. Nvidia is worth 20-30 billion or so in reality.
Part of NVidias valuation was due to the perception that AI companies would need lots and lots of GPUs, which is still true. But I think the main problem causing the selloff was that another part of the popular perception was that NVidia was the only company who could make powerful enough GPUs. Now it has been shown that you might not need the latest and greatest to compete, who knows how many other companies might start to compete for some of that market. NVidia just went from a perceived monopolist to "merely" a leading player in the AI supplier market and the expected future profits have been adjusted accordingly.
I was under the impression too that this would bump the retail customers demand for the 50 series given the extra AI and cuda cores, add to that the relatively low cost of the hardware. But I know nothing of the sentiments around wallstreet.
I don't feel like upgrading my 4090 that said. Maybe wallstreet believes that the larger company deals that have driven the price up for so long might slow down?
Or I'm completely wrong on the impact of the hardware upgrades.
Output quantity consumed (almost) always increases with falling inputs (ie, costs, whether in dollars or GPUs). But for Jevon's paradox to hold, the slope of quantity-consumption-increase-per-falling-costs must exceed a certain threshold. Otherwise, the result is just that quantity consumed increases while quantity of inputs consumed decreases.
Applied to AI and NVIDIA, the result of an increase in the AI-per-GPU on demand for GPUs depends on the demand curve for AI. If the quantity of AI consumed is completely independent of its price, then the result of better efficiency is cheaper AI, no change in AI quantity consumed, and a decrease in the number of GPUs needed. Of course, that's not a realistic scenario.
(I'm using "consumed" as shorthand; we both know that training AIs does not consume GPUs and AIs are also not consumed like apples. I'm using "consumed" rather than the term "demand" because demand has multiple meanings, referring both to a quantity demanded and a bid price, and this would confuse the conversation).
But a scenario that is potentially realistic is that as the efficiency of training/serving AI drops by 90%, the quantity of AI consumed increases by a factor of 5, and the end result is the economy still only needs half as many GPUs as it needed before.
For Jevons paradox to hold, if the efficiency of converting GPUs to AI increases by X, resulting in a decrease in price by 1/X, the quantity of AI consumed must increase by a factor of more than X as a result of that price decrease. That's certainly possible, but it's not guaranteed; we basically have to wait to observe it empirically.
There's also another complication: as the efficiency of producing AI improves, substitutes for datacenter GPUs may become viable. It may be that the total amount of compute hardware required to train and run all this new AI does increase, but big-iron datacenter investments could still be obsoleted by this change because demand shifts to alternative providers that weren't viable when efficiency was low. For example, training or running AIs on smaller clusters or even on mobile devices.
If tech CEOs really believe in Jevons Paradox, it means that last month when they decided to invest $500 billion in GPUs, then this month after learning of DeepSeek, they now realize $500 billion is not enough and they'll need to buy even more GPUs, and pay even more each one. And, well, maybe that's the case. There's no doubt that demand for AI is going to keep growing. But at some point, investment in more GPUs trades off against other investments that are also needed, and the thing the economy is most urgently lacking ceases to be AI.
If you care to respond though, my first question would be what examples of falling input prices not subject to the Jevons Paradox are. Several of the more notorious ones involve energy, and that was Jevons's principle topic of study (The Coal Question most notably).
As might be pertinent to AI and LLM, whilst fuels and power applications seem to scale linearly against input (constant slope, if not 1:1 relation), information processing delivers far more variable returns, often with critical thresholds. Network effects and Metcalfe's Law are the best known of these (if highly inaccurate themselves, see Tilly-Odlyzko's refutation), but another is the limited returns of predictive and targeting applications.
For the latter, the 18 order of magnitude increase in computing power from 1965--2025 (60 years, about 20--30 Moore's Law cycles) has roughly doubled the length of accurate long-term weather forecasting from roughly 5 days to 10. It's made possible fully-resuable first-stage boosters for orbital spaceflight, which is visually impressive, but has only resulted in a five-fold reduction ($1,400/kg vs. $5,400/kg) in low-Earth orbit (LEO) launch costs (Falcon Heavy vs. Saturn V). SpaceX are looking for another factor of 2--4 reduction (to $250--600/kg), but that's still far less improvement than we've seen in raw compute. At some point orbital physics, the rocket equation, and fuel chemistry simply dominate other considerations.
Similarly, AdTech makes possible far more targeted advertising, but to heavily diminishing returns, the core result has been an abandonment of non-targetable media by advertisers, notably print and broadcast, as well as an arms-race between the browser (for a very small fraction of the market) and advertisers (the largest of which also has the largest browser marketshare), and a concentration of advertising revenue amongst two online entities, Google (a/k/a Alphabet) and Facebook (a/k/a Meta).
Which makes me wonder what applications AI LLMs might practically be put to. Advertising, manipulation, fraud, and propaganda certainly seem to be benefiting.
> I say DeepSeek should increase Nvidia’s demand due to Jevon’s Paradox.
If their claims were true, DeepSeek would increase the demand for GPU. It's so obvious that I don't know why we even need a name to describe this scenario (I guess Jeven's Paradox just sounds cool).
The only issue is that whether it would make a competitor to Nvidia viable. My bet is no, but the market seems to have betted yes.
> DeepSeek should increase Nvidia’s demand due to Jevon’s Paradox.
How exactly? From what I’ve read the full model can run on MacBook M1 sort of hardware just fine. And this is their first release, I’d expect it to get more efficient and maybe domain specific models can be run on much lower grade hardware sort of raspberry pi sort.
I agree but in the short/medium term, I think it will slow down because companies now will prefer to invest in research to optimize (training) costs rather than those very expensive GPUs. Only when the scientific community will reach the edge of what is possible in terms of optimization that it will be back at pumping GPUs like today. (Although small actors will continue to pump GPUs since they do not have the best talents to compete).
The other way is certainly also true. Your short piece is rational, but lacks insight into the inference and training dynamics of ML adoption unconstrained.
The rate of ML progress is spectacularly compute constrained today. Every step in today’s scaling program is setup to de-risked the next scale up, because the opportunity cost of compute is so high. If the opportunity cost of compute is not so high, you can skip the 1B to 8B scale ups and grid search data mixes and hyperparameters.
The market/concentration risk premium drove most of the volatility today. If it was truly value driven, then this should have happened 6 months ago when DeepSeek released V2 that had the vast majority of cost optimizations.
Cloud data center CapEx is backstopped by their growth outlook driven by the technology, not by GPU manufacturers. Dollars will shift just as quickly (like how Meta literally teared down a half built data center in 2023 to restart it to meet new designs).
I think it's entirely possible that one categorically can't think correctly about markets and equity valuations since they are vibes-based. Post hoc, sure, but not ahead of time.
Most people don't care about the fundamentals of equity valuations is the crux of it. If they can make money via derivatives, who cares about the underlying valuations? I mean just look at GME for one example, it's been mostly a squeeze driven play between speculators. And then you have the massive dispersion trade that's been happening on the SP500 over the last year+. And when most people invest in index funds, and index funds are weighted mostly by market cap, value investing has been essentially dead for a while now.
"Briefly stated, the Gell-Mann Amnesia effect is as follows. You open the newspaper to an article on some subject you know well. In Murray's case, physics. In mine, show business. You read the article and see the journalist has absolutely no understanding of either the facts or the issues. Often, the article is so wrong it actually presents the story backward—reversing cause and effect. I call these the "wet streets cause rain" stories. Paper's full of them.
In any case, you read with exasperation or amusement the multiple errors in a story, and then turn the page to national or international affairs, and read as if the rest of the newspaper was somehow more accurate about Palestine than the baloney you just read. You turn the page, and forget what you know."
Because your comment was posted 9 hours ago, I have no idea what view you think is wrong. Could you explain what the incorrect view is and — ideally — what’s wrong with it?
That means the remaining 10% are similarly disillusioned by the impression Apple or AMD could "just write" a CUDA alternative and compete on their merits. You don't want either of those people spending their money on datacenter bets.
10 years ago people said OpenCL would break CUDA's moat, 5 years ago people said ASICs would beat CUDA, and now we're arguing that older Nvidia GPUs will make CUDA obsolete. I have spent the past decade reading delusional eulogies for Nvidia, and I still find people adamant they're doomed despite being unable to name a real CUDA alternative.
Did ASICs beat CUDA out in crypto coin mining? Not the benchmark I really care about, but if things slow down in AI (they probably won’t) ASICs could probably take over some of it.
Equity valuations for AI hardware future earnings changed dramatically in the last day. The belief that NVIDIA demand for their product is insatiable for the near future had been dented and the concern that energy is the biggest bottle neck might not be the case.
Lots to figure out on this information but the playbook radically changed.
Let's not forget, that also doesn't make you an expert in the history and society evolution, as we can all agree that a part of HN public still believe in “meritocracy” only and think that DEI programs are useless.
