AFAIK cost here counts only the manufacturing side. While your conclusion that in the long run economies of scale will prevail, the lifetime costs are probably more than 30%. For example I expect recycling costs to be significantly worse for the Li-Ion.
> For example I expect recycling costs to be significantly worse for the Li-Ion.
I think there's a good argument for the opposite.
Recycling costs for Li-Ion once we are doing it at scale should be significantly negative. There are valuable materials you get to extract, they aren't in that complex a blend to extract them from, and there's a lot of basically the same blend. The biggest risk in this claim is, I think, the implicit claim that we won't figure out how to extract the same materials from the earth much cheaper in the meantime cratering the end of life value of batteries - but in that event the CO2 battery technology is underwater anyways and the chemical batteries win on not wasting R&D costs.
By contrast while there's some value in the steel that goes into building tanks and pumps and so on, the material cost if a much lower fraction of the cost of the device. Most of the cost went into shaping it into those complex shapes. I don't know for sure what the cost breakdown of the CO2 plant looks like but if a lot of the cost is something else it's probably something like concrete or white paint that actually costs money to dispose of.
Grid scale LFP with once daily cycling lasts 30 years before the cells are degraded enough to think about recycling.
And those are very low maintenance over that time.
You're probably mostly going to swap voltage regulators and their fans, perhaps bypass the occasional bad cell by turning the current to zero, unscrewing the links from the adjacent cells to the bad cell, and screwing in a fresh link with the connect length to bridge across.
Also: From what I understand the LFP battery degradation is essentially corrosion that is removed by recycling and you can retrieve 99% of the essential LFP elements and make it into a new battery. So economically we only need to extract the LFP related materials for each human user almost once, instead of oil over and over again.
> But part of this is the latest incarnation of the prosperity gospel, an early (and now pervasive) idea that wealth is a demonstration that you are favored by God. Another version of this is the myth of meritocracy under capitalism. So if you're wealthy, it's because you're better than all "the poors", by definition.
There is a whole industry that caters to ultra wealthy and one of their core tenets is that customer is always right. This is problematic because negative feedback is an important mechanism of personal growth and self development. Thus we end up with ultra wealthy people being mostly disconnected from the society.
> The first thing that blew my mind was how stupid the whole idea is. Think for one second. One full second. Why do you ever want to add a chatbot to a snack vending machine?
This feels forced, there are obvious and good reasons for running that experiment. Namely, learning how it fails and to generate some potentially viral content for investor relationship. The second one seems like an extremely good business move. It is also a great business move from WSJ, get access to some of that investor money in an obviously sponsored content bit that could go viral.
Having said that, I do feels the overall premise of the blog - the world dynamics seems exceedingly irrational in recent times. The concerning fact is that irattionality seems to be accelerating, or perhaps it is keeping pace with the scale of civilization... hard to tell.
> This feels forced, there are obvious and good reasons for running that experiment. Namely, learning how it fails and to generate some potentially viral content for investor relationship. The second one seems like an extremely good business move. It is also a great business move from WSJ, get access to some of that investor money in an obviously sponsored content bit that could go viral.
That's... exactly what the author said in the post. But with the argument that those are cynical and terrible reasons. I think it's pretty clear the "you" in "why would you want an AI" vending machine is supposed to be "an actual user of a vending machine."
I think you’re overstating your own interpretation of what the author wrote. If we’re going to take your use of the word “exactly” (with emphasis) for real then I’d argue that the author offers no charitable reasons for why the experiment took place.
The closest that I think he even gets to one is:
> At first glance, it is funny and it looks like journalists doing their job criticising the AI industry.
Which arguably assumes that journalists ought to be critical of AI in the same way as him...
Are you sure? The entire post treats the event incredulously. I can’t pick out a single line that affords the issue with the level of consideration that the GP comment does.
The two reasons I believe you may be referring to from above are:
1) "learning how it fails"
2) "to generate some potentially viral content for investor relationship."
The whole of Ploum’s argument may be summarized in his own words as:
> But what appears to be journalism is, in fact, pure advertising. [...] What this video is really doing is normalising the fact that “even if it is completely stupid, AI will be everywhere, get used to it!” [...] So the whole thing is advertising a world where chatbots will be everywhere and where world-class workers will do long queue just to get a free soda. And the best advice about it is that you should probably prepare for that world.
I hate to be pedantic...but my growing disdain for modern blog posts compels me to do so in defense of literacy and clear arguments.
Whether the GP and the author offer the “exact same two reasons” is a matter of interpretation that becomes the duty of readers like us to figure out.
If we take Ploum’s words at their face...the most he does is presuppose (and I hope I’m using that word correctly) that the reader is already keen on the two reasons that `TrainedMonkey makes explicit and like the author, finds them to be stupid. While he does say that the video is not journalism and that it is advertising and that the video does show how the AI failed at the task it was assigned he does not give any credence as to why this is the case from a position other than his own.
Maybe I’m misunderstanding the concept of a “charitable interpretation” too. But I don’t think that there is one present in this post that we’re responding to. `TrainedMonkey’s comment leads off by telling us that this is what (I think) he’s about to offer in the remarks that follow when he says “there are obvious and good reasons for running that experiment”.
So my gripe is that you’re making it sound like there’s a clear counterargument entertained in this post when there isn’t. Because you overstated your interpretation of the GP comment in what looks like an attempt to make Ploum’s argument appear more appealing than it ought to be. Even though both `TrainedMonkey and myself have expressed agreement with the point he’s trying to make in general, perhaps we’re less inclined toward pugnaciousness without a well thought out warrant.
Anything that's an "extremely good business move" will most likely, in this day and age of late-stage capitalism and extreme enshittification, be negative for the consumer.
Good business moves can often be bad for humanity.
I can see this as a good business move for Anthropic in this case but for the vending machine manufacturers/ operators what would the advantage be besides saying that their machine is AI operated?
All journalist organizations, unfortunately, have an incentive to bias content towards maximum clickbait. The ones that don't end up being outcompeted.
This is not true. You just need to read the business papers, as they have incentives to be accurate. I like the FT, but apparently the non opinion WSJ is ok.
Currently helping with hiring and can't help but reflect on how it changed over past couple of years. We are now filtering for much stronger candidates across all experience levels, but junior side of the scale had been affected much more. Where previously we would take top 5% of junior applicants that made it past first phone screen, now it's below 2%.
AFAIK compute heavy datacenter in space don't work. But if you already have a vast fleet of laser connected LEO satellites throwing some efficient SSDs into them can make a lot of sense. A large portion of the traffic is fairly static, e.g. video based content or even model weights. Caching that will save you ground to space side of the transmission. This will let you put more user beams on satellites and use less ground stations.
> we need a hyper-productive world where someone working a few hours a year generates enough wealth to secure a comfortable lifestyle up there with the best of them.
This is naive, productivity increases had decoupled from compensation a long time ago. See https://www.epi.org/productivity-pay-gap/ for example. AI certainly can create wealth, and in fact already did (hey NVDA), but somehow that did not trickle down. I think more likely than not, AI will further stratify our society.
But humans aren't more productive for the most part. What had made people more productive is, for the most part, mechanization, computerization, and other tech tree improvements.
Even though everyone didn't get rich from the industrial revolution, ultimately people led easier lives, more stuff, and less work.
Peasants in the Middle Ages did hard, agrarian labor, had poor sanitation, high child mortality, and limited horizons. While nobles had wealth and comfort, the era was generally harsh, with most lacking modern amenities and facing short, physically demanding lives.
Why then, outside of Norway, Luxembourg, Switzerland, and the UAE (All of which are tiny countries, and at least two of them are Petrostates), does the United States have the world's highest median income with a population of over 342 million people?
The typical American is insanely wealthy by global standards.
Insanely wealthy when your comparison includes tin pot dictatorships, theocracies, and ex-soviet countries that still haven't gotten their shit together. Weird how that unimaginable wealth doesn't translate into financial security, access to high quality healthcare, or the ability to own a home.
Who cares? If higher wealth inequality produces a higher standard of living for the majority (note, median not mean), I’m all for it. Policy should not be driven by envy.
You should care because people vote and the social consequences are going to be devastating.
It is easy for me to take this perspective too because I never had much student debt or children.
The median though is getting crushed if they went to college and are paying for daycare.
If you are getting crushed for going to school and having children that is a pretty clear breakdown of the social contract.
The consequences are obvious. People are going to vote in socialist policies and the whole engine is going to get thrown in reverse.
The "let them eat cake" strategy is never the smart strategy.
It is not obvious at all our system is even compatible with the internet. If the starting conditions are 1999, it would seem like the system is imploding. It is easy to pretend like everything is working out economically when we borrowed 30 trillion dollars during that time from the future.
> The median though is getting crushed if they went to college and are paying for daycare.
> If you are getting crushed for going to school and having children that is a pretty clear breakdown of the social contract.
That isn't a factor in wealth inequality. Inequality is how much money they have relative to people like Musk and Bezos - or just local business owners. The poor side of that comparison always has such little wealth/income that their circumstances don't really matter. Someone poor will be sitting in the +-$100k band and not be particularly creditworthy. When compared to a millionaire the gap is still going to be about a million dollars whether they're on the crushed or non-crushed side of the band.
Part of the reason the economic situation gets so bad is because people keep trying to shift the conversation to inequality instead of talking about what actually matters - living standards and opportunities. And convincing people to value accumulating capital, we're been playing this game for centuries, inter-generational savings could have had a real impact if people focused on being effective about it.
The 1870-1900 period experienced the greatest expansion of U.S. industry, and the fastest rise in both U.S. wages and U.S. life expectancy, in history.
Right we had a functioning labor movement to thank for productivity gains being distributed to the working class. When that got undermined beginning late seventies early 80s with offshoring we see wealth just flowing to the top without significantly benefiting the working class.
I don't know why you think that graph is contradicting me, if wages and productivity aren't linked having people do make-work is even more stupid! They're already doing work that they aren't even being compensated for, fighting to preserve that when the work doesn't need to be done is legitimately crazy. It'd be fighting for the right to do work that isn't being compensated for and isn't useful. One of the rare situations that is even worse than just straight paying people to not do anything. I'm seeing a scenario where we have such high individual productivity that everyone can live a very comfortable life. If in practice the way it is working is a couple of people do all the work and the benefits are divvied up among everyone else then that hardly undermines the vision.
Although if we're talking the optimum way of organising society, y'know, re-linking wages and productivity is a probably a good path. This scheme of not rewarding productive people has seen the US make a transition from growth hub of the world to being out-competed by nominal Communists. They aren't exactly distinguishing themselves with that strategy.
I long had a similar idea for stocks. Analyze posts of people giving stock tips on WSB, Twitter, etc and rank by accuracy. I would be very surprised if this had not been done a thousand times by various trading firms and enterprising individuals.
Of course in the above example of stocks there are clear predictions (HNWS will go up) and an oracle who resolves it (stock market). This seems to be a way harder problem for generic free form comments. Who resolves what prediction a particular comment has made and whether it actually happened?
> Analyze posts of people giving stock tips on WSB, Twitter, etc and rank by accuracy.
Didn't somebody make an ETF once that went against the prediction of some famous CNBC stock picker, showing that it would have given you alpha in the past.
> seems to be a way harder problem for generic free form comments.
That's what prediction markets are for. People for whom truth and accuracy matters (often concentrated around the rationalist community) will often very explicitly make annual lists of concrete and quantifiable predictions, and then self-grade on them later.
long-term his choices don't do well, so the Inverse Cramer basically says "do the opposite of this goober" and has solid returns (sorta; depends a lot on methodology, and the sole hedgefund playing that strategy shutdown)
Out of curiosity, I built this. I extended karpathy's code and widened the date range to see what stocks these users would pick given their sentiments.
What came back were the usual suspects: GLP-1 companies and AI.
Back to the "boring but right" thesis. Not much alpha to be found
The academic consensus is that there is a housing shortage, not a surplus. Perhaps there is a local surplus in undesirable areas, but that isn’t true in cities or nationally.
Austin and Texas in general have less housing restrictions than other places wrt zoning and regulation, so they could plausibly have a surplus. You can only really speak locally about any of this because location determines the market
No, you just happened to build appropriately because a certain subfaction of the population weren't able to pass the typical laws that would stop building.
The housing shortage was created by regulation and it's foolish or selfish to pretend otherwise.
Austin is unique in that most of the harmful self-serving conservatism-as-in-block-and-deny-all-development that city people usually to do is constrained by the rest of the state, and as an obvious result arguably has the highest standard of living in the entire world.
regulation /tends/ to be introduced because builders are misbehaving (bad materials, bad workmanship, building in flood zones, etc), but the bigger problem is NIMBY who then use those laws to prevent other people building in "their" neighbourhood
it's the market dynamics that create under building, not regulations. if regulations are the problem, then how were builders able to build record number housing starts leading up the GFC? and after the crash, housing starts dramatically lowered. were there new regulations introduced during that time period?
highest standard of living unless you're a woman or a trans person, or a person of color, you mean. I'm sure living in a repressive dictatorship is awesome if you're a rich member of the in-group.
Fundamentally because demand and supply curves are lagged. This drives DRAM flood and drought cycles. Right now high DRAM prices are driving fab investment up and demand down. In 3-5 years new fabs will cause DRAM glut that will drive prices down. Lower prices will stimulate demand via things like doubling DRAM size in consumer electronics as companies compete on getting the number bigger. Eventually demand curve will eclipse the supply and we will end up in the DRAM drought again.
This time things are further complicated by the fact that the world is investing a sizable chunk of GDP into building RAM hungry data centers in hopes of building a god which will convert the rest of the world into data centers.
The biggest issue here is that it hurts smaller consumers. Scaling up a fab to produce ram takes 5+ years, people generally buy new hardware ~5 years so this lifecycle of hardware for some people is now locked out. Let's say you was due for an upgrade this year you now might be priced out of the market for the next 5 years and basically due to poor business practices
There is a similar issue happening in the manufacturing space where metal foundries are basically "full" up on allocation for other customers and will refuse to sell to you unless your purchase order is six digits otherwise you pay a hefty premium which once again drives capital towards larger corporations. Compounded by a stagnant jobs market the means that scarcity is just going up and up and nobody is re-investing to meet consumer demands because the market is poisoned by speculation to the absolute extreme
Do people really still upgrade so often? I mean it made sense pre 2015 for desktops, and pre 2020 for laptops... But since... Not much has changed from a performance standpoint.
Even gpus hardly advanced since 2022 (4090) and the next generation is at least 1++ years off. Likely 2-3... And it's unclear wherever it will actually be an upgrade or more of the AI shenanigans they released with the 50 generation.
Well, I just updated my 5950x with 128 GB to AMD Strix Halo with the same amount of much faster RAM. It is noticeable faster, but what's better: the whole computer is tiny and sips energy.
I'm very happy I ordered this in the summer, framework delivered it to me early this month. I wonder will these machine just be out of stock now or the price goes up a lot...
This isn't true. It used to be, as a new fab would appreciably add quantity. At 1M wspm in 2015, a new 100k fab at the most modern node would add effectively 20-30% capacity, and usually multiple.players at once, since all had cash.
Now, the relative shrink is tiny, so capacity adds are just wspm, in effect, and that gives 5%.
Put differently, you cannot invest your way out of the shortage, or into meaningful share....so you take profit.
guess I'm interpretting "1million wspm; add 10%; was effectively a 20-30% capacity increase" in 2015.
Not sure where 5% then comes from. Guessing "relative shrink" is referring to process size (5, 4, 2 nm, whatever) not linearly corresponding to density of transistors, etc.
reply