This is a known potential side-effect of these drugs. When I started them my doctor warned me about a possible lack of motivation. These drugs trigger your body's satiation mechanism (as I understand them–not a doctor) and that can apply to more things than just eating.
This is true if you take the ai market as equal to the market for labor discounted to 5-10% penetration.
It’s not a totally unreasonable assertion, it’s the implication of the assertion that we are uncomfortable with. There is no reason for the models to stop their improvements in the near future.
> There is no reason for the models to stop their improvements in the near future.
Sure there is.
1. The cost of each new generation of training runs appears to be rapidly rising
2. The Trump admin just told the leading model to stop making it available to non-Americans, which in practice meant stop providing it at all
3. The factories to make the hardware are hitting bottlenecks, and while they've currently been navigated around, there's never a guarantee the next one will be
Currently I'm wondering at what point the direct impact on the US energy supply gives the US a taste of Baumol's cost disease as AI companies continue to outbid everyone else for electricity.
There are some counters to this, especially in electricity. We'll see massive expansions of wind and solar in the US because of this. Both the speed of install and low costs will guarantee it.
> We'll see massive expansions of wind and solar in the US because of this. Both the speed of install and low costs will guarantee it.
Implausible while Trump remains in office. He hates renewables, shuts them down even when doing so actively costs money.
Between AI hallucinated content and the politicisation of the numbers, I'm not sure how much AI compute capacity is being planned right now; would you accept a claim of 300 GW? It's a number I heard recently.
Given the capacity factor of PV, even China would have to think carefully before supplying that much PV over the next few years (300 GW avg ~= 3TW nameplate).
(Not sure about wind, wind's CF seems to vary between years).
Unless it's something on federal land businesses are pretty much ignoring Trump on renewables.
And 300GW power planned doesn't seem too far out of bounds, there are a huge number of 'planned' data centers all over the US.
World wide over 800GW of solar and wind was installed in 2025 and 2026 numbers should be over 1TW of renewables. How much of that will the US install itself is a much smaller percent, but as power prices increase the pace to profit off of it will quicken. I know China installed over 300GW themselves last year.
> Unless it's something on federal land businesses are pretty much ignoring Trump on renewables.
Except for all of the tariffs etc.
And that's without Trump seeming to be actively choosing winners based on favour to him, as with supporting Grok despite the data centre pollution in another thread and *possibly* (I don't wish to overstate my case) the ban on Fable.
> World wide over 800GW of solar and wind was installed in 2025 and 2026 numbers should be over 1TW of renewables. How much of that will the US install itself is a much smaller percent, but as power prices increase the pace to profit off of it will quicken. I know China installed over 300GW themselves last year.
This is why I wrote:
Given the capacity factor of PV, even China would have to think carefully before supplying that much PV over the next few years (300 GW avg ~= 3TW nameplate).
When the recent good news is that "the world installs 800 GW of PV" (TBH, I thought this was closer to the last 12 months of just PV than the sum PV+wind), that's the nameplate capacity, not the actual year-long-average output, which is about a tenth of that.
The most recent PV capacity factor number on Wikipedia was 13%, which would make "800 GW" only 104 GW in real output; the figures I see for wind are that the CF is 25% (with much higher variability) but the nameplate capacity is lower, so they're pretty close as totals in real total currently installed output.
As long as Chinese companies keep pushing on, so must US companies too.
It would not surprise me at all if we suddenly start seeing top US AI companies lobby against Chinese models, or even the gov. making it illegal to use Chinese AI models.
But in this day and age, I just don't think it is possible. A distant third option would be that the big AI companies try to make hardware so expensive that people simply can't run their own models, while blocking access to foreign models.
> There is no reason for the models to stop their improvements in the near future.
You speak as if "improvements to models" is just function of time, and resources are infinite.
Models keep improving as long as there are resources to allow for larger and larger datacenters, if we hit a scientific breakthrough once LLM technology become the bottleneck, if the economy is infinite to allow infinite growth, and (geo)politics is not a thing to worry about. Or we discover ASI, machine improve themselves and we reach the technological singularity.
I know everybody is drinking the kool aid by the gallon, but can we maintain a little bit of objectivity?
The s curve won’t inflect until it becomes difficult to allocate additional resources due to economic limitations. There is no sign that training a model on 10x the compute won’t lead to at least an equivalent improvement as the last order of magnitude increase.
If we define the Pareto frontier’s input in terms of a magic “compute equivalent unit”. We get a free order of magnitude from nvidia hardware improvements every 2-3 years. We get another order of magnitude from capital expenditure every 6-12 months. Kernel improvements to the models themselves likely yield an order of magnitude gain at some periodicity.
> the market for AI is 2.5 larger than all the food sold on this planet.
It just shows how much the automation has impacted agriculture and the food industry. Sure, there're rural farms that apply 200 yo technology. But e.g. the grain production and farming are incredibly efficient at scale. So, it's not that costly for as a humanity to feed 8 billion people (at a varying level, of course).
Food is a solved problem. We can grow far more food than we need and we stop doing so simply because the low prices mean it's not economically viable.
In the places where famine remains a problem, it's due to political issues, not that we can't grow enough.
And growing all that food requires a tiny workforce compared to 400 years ago before the Agricultural Revolution. AI might extend such a massive reduction in labour requirements to many other industries.
> We can grow far more food than we need and we stop doing so simply because the low prices mean it's not economically viable.
Half. This depends on there being a reliable source of cheap fertiliser, which would be much more secure if not for the situations regarding Hormuz and Russia.
> We can grow far more food than we need and we stop doing so simply because the low prices mean it's not economically viable.
So, it's not a solved problem.
Last time I checked we have plenty of people in several parts of the world with difficulties to access the required level of food to be healthy.
Technologically yes, but this is a vast oversimplification.
You need lots of money to be able to buy the tech you need to do so. And you can't exactly earn that from not using the tech, since foreign (or even local) competition will slaughter you on prices. And if you do make it, you're stuck with a low-margin race to the bottom on price.
I don't think anyone is claiming AI and food have the same elasticity of demand, which is what this really talks to, but, after a claim the AI market is 26 trillion dollars... I wouldn't be surprised if someone did.
Ackshully, AI does have an upper bound in information theory, but since we're not anywhere close to writing data to the surface of a black hole I don't think it's a big issue yet.
We could remove 100% of world AI supply and humanity would not be worse off. It is still additive and in areas of generally indeterminate value except in hype.
Reasoning and RAG is amazing already and is a productivity gain but I'm yet to be convinced GenAI is anything but a slop machine.
You can't food maxx a trillion calories a day to generate a multi million dollar bill. You can token maxx it though.
I think the issue is the reality that most life is worth a lot less (in US Freedom units) than some software running doing absolutely nothing truly valuable for anyone.
What precisely is the moat surrounding AI that SpaceX is using to justify this kind of spending spree? I don't how SpaceX and other AI companies will be able to keep the weights of their AI models private in the face of interest by virtually everyone in the world. It would be absolutely trivial for a nation state to walk into a data center using a state issued security certificate to seize a few of the physical servers running the cloud services of OpenAI / Grok / Claude. Copying the weights is trivial. Infiltrating a company with spies as new hire coders to gain access to source code is also trivial.
This is really starting to feel like the pets.com era again.
This is an insane take. Of course there are other areas that are growing or could grow if there was investment.
The problem is an absolute lack of vision on the part of those holding the capital.
Meeting the challenges of climate change could hold huge opportunities. Look at China’s massive expansion in renewables, look at the expansion of renewables in the US despite political headwinds.
Have some imagination, break out of your echo chamber. AI ain’t the only game to be played.
It's not lack of vision. It's that capital demands gambling. Everyone knows you could plough money into big projects in the US and double your money reliably. But the powers that be do not want that. They want to gamble, and try to become trillionaire #2.
It's not insane. The GP is correctly describing a bubble economy.
The money chasing investments is orders of magnitude larger than the money people have on their pockets to spend. As a consequence, the only profitable thing to do is sell capital goods to make business and there is no profit on selling actually useful things.
China is in a different reality in large part because of their capital barriers that stop money from flowing in. Countries with bad reputation are also less affected.
What the GP gets wrong is that none of this makes AI a good business. Instead, it makes Nvidia a good business, but that's not news.
Looking at venture funding, it's definitely true. That doesn't mean other problems don't exist or aren't worth solving. But the concentration of (competing) capital and talent is insane.
Not to be a jerk or anything, but that only matters for the guys and gals left holding the bag.
I don't think that will be Musk. He'll probably pull out significant resource from all this financial engineering relatively quickly. Probably via more financial engineering.
Cursor is a harness that can be used with all kinds of models. It's a much better harness than anyone else's and takes the company out of just playing the model game.
Exactly this. I think valuations and the AI market could get stirred up if:
- We get an open source Opus 4.8 equivalent and pair it with an open source coding agent
- Running this OS stack becomes cheaper than what frontier model providers charge (see OS model prices on OpenRouter vs. frontier lab prices)
- This happens across verticals (i.e. not just software)
The first “DeepSeek moment” didn’t do much damage back in the days, but I wouldn’t be surprised if a similar moment becomes a lasting, effective, cheaper alternative.
OpenCode exists, it is your "open source coding agent" that is practically on par with Claude Code and Copilot in terms of being able to do the 80% of things that most people actually use.
DeepSeek v4 Flash/Pro also exist, they are open weight and on par with Sonnet, just a bit below Opus. Again: practically useful and sufficient for 80% of things most people actually do. And most of the remaining 20% are benchmarks designed to push the limits, not productive work.
Using these already is way cheaper than your typical Claude API prices. What's still missing is a) mindshare - everyone still thinks "claude = coding" and everyone thinks he/she really needs the very best models because he/she is doing such incredibly complex stuff - and b) someone pushing such a stack as a convenient solution for corporations to easily dump their token money into, complete with user management, enrollment, monitoring, all that enterprisey stuff you need if you want to sell to, well, enterprise customers.
Need to give that a try then. What's your experience so far? Also in terms of cost/benefit?
My current hypothesis is that the $150 Claude Max subscription - of which I barely hit the limits anyways, even though I used it non-stop at work - still is very cost effective.
If the price of Claude Max increased significantly (say 2-3x), and my business would balk at paying the subscription, then I'd look for an alternative.
They are a distant third at best, at least in trading companies. If you look at Chinese and other likely national actors, they are probably further down.
The thing with dotComs is that they didn't have THIS level of unsustainable financing burn, and a tangible issue of token processing cost that has no magic wand coming with the current practical limits of Moore's law.
To be fair, food is the smallest bucket of my monthly expenses. And there are many people here on hacker news who pay more for their AI tokens than for their food.
How does argrar industry and tech industry compare as share of gdp in the US?
We'll figure out how to make it much much cheaper to produce the compute needs we have today with tokens. The question is will how many new use cases arise that need much more - clearly we aren't meeting needs (price stays high) but how much more do we need, 100x, 10x, 10^6 x?
Look at electricity, the world of 1900 could not create enough electricity or even conceive of how to add enough to meet 1950s needs. But we made it incredibly cheaper to produce, but also created a lot more, and boy do we have so much more use of electricity now. And it's not that expensive for a human to pay for their needs (not free, its not cheap for poor people but it's still gotten cheaper).
Food is worth a lot more than that. If the alternative was starvation, we would pay approximately all the money for food. By that metric food is worth more than $100T. The difference between $100T and $10T is called the consumer surplus, one of the largest benefits of a free market economy.
AI might eventually provide $26T worth of value, but if it captures anywhere close to that amount of revenue that'll indicate a failure of the free market economy. Competition and open source will have failed and the oligarchy has won.
(Either that, or inflation will have made $26T a relatively smaller number).
Always fun to remember when calculating TAM - something like 85 - 90% of the world earns less than $1000 usd per month.
The math don’t math here, there literally aren’t enough people to afford this and businesses will go under the more people are displaced for gainful employment.
There have been a few recent stories about businesses finding themselves spending more on tokens than they were spending on the workers these AI Agents were supposed to have replaced.
You need to understand the definition of “total addressable market.” It’s a maximum theoretical number for the size of the market (not your company’s revenue) under ideal assumptions. A $26 trillion TAM is high but it’s not “unhinged.” For example, the logistics and transportation market is over $10 trillion and expected to double by 2035. Under ideal assumptions, if AI replaces everything from coders to lawyers, why is that “unhinged?”
I also saw a quote from Musk saying that he expects SpaceX to hit $1 trillion in revenue by 2031. Given his track record of predicting performance I think it's safe to ignore such future looking statements from him or companies he controls.
Another way of putting this: global GDP is ~$132Trillion from what I gather.
So this is saying AI products will increase global GDP by about 20%.
The Federal Reserve says AI is contributing about 1% GDP growth per year to the US [0].
So maybe you can get to $13 trillion over a decade just from that. If you assume some acceleration, 20% isn't out of the question.
It is an extremely rosy projection, but if AI can replicate large fractions of the workforce, leaving those humans with the ability to work on other things, it doesn't seem unhinged when you think of it through this lens, just very optimistic - not Elon Musk level optimistic, just "everything goes according to plan and a bunch of things in the causal chain are all slightly on the higher end."
> So this is saying AI products will increase global GDP by about 20%.
No business gets to capture 100% of the value it produces without physical coercion.
For infrastructure that requires high investment, it usually captures something around 5% of it. People tend to work really hard to replace or reduce any kind of infrastructure that gets near 10%. So we are talking about AIs increasing the global GDP by 200% at minimum, 400% more realistically.
With SPCX shares never going down in price, SpaceX can acquire all companies in the US in exchange for its stock, so SpaceX itself is worth at least as much as the US GDP! (/s)
And even if it _was_ related to AI, they would not admit it. First course of action is to blame user/programmer error and then QA process error. You shall not blame the golden calf. I am half serious and half not. But I do recommend reading the book "The Field Guide to Understanding 'Human Error'" in conjunction with my hyperbole.
I think it's reasonable to request the person making the assertion to back it up. It's not on the audience to either only debunk or accept the assertion. It can just be rejected.
Why is an anodyne factual claim an “extraordinary claim”? What makes that particular claim extraordinary? They didn’t claim to have discovered perpetual motion or something you can’t prove or disprove yourself, just shared a historical fact you can easily just check up on if you choose not to believe them.
>Why is an anodyne factual claim an “extraordinary claim”? What makes that particular claim extraordinary?
FWIW I tried to get AI to substantiate it and came up empty. Maybe it's not as "extraordinary" as "Obama was a reptilian alien" or whatever, but for everything else what counts as "extraordinary" depends on your prejudices, I suppose. Regardless of whether it's "extraordinary" or not, it's definitely not common knowledge and needs to be substantiated rather than asserted without evidence.
reply