I don't actually think this interpretation is correct. ZIRP was the government printing money and buying bonds off the market. I think it had more to do with velocity of money, which ordinarily would have been corrected through a recession, but the government prevented that, which will create a much bigger recession at some point in the future.
No, the fed doesn't set rates. Not by much, not for long, not without inflation taking off. In theory they chase R*, in practice they chase the 2-year. Plot EFFR and US02Y on tradingview and tell me who leads whom, and tell me what happened when the fed tried to fight the market.
For the most part, congress decides to spend a certain amount in excess of receipts, congress decides the debt, the private sector bids on the debt, and the marginal bid sets the interest rate.
When the federal reserve steps in they can use their magic balance sheet to, at great expense, tug the interest rate around a little bit. However, artificially creating ZIRP in a non-ZIRP economy would not be a little tug, it would require buying most or all of the unattractive bonds. In 2020 it would have required $25T not $4T (GFDEBTN vs WALCL in 2020). Because the sum total of their intervention was small compared to the debt sold to that point, the low-intervention approximation is correct during the 2010 ZIRP era, the "P" in ZIRP is a misnomer, and "capital had more money than they knew what to do with" is the correct read on how ZIRP happened.
I suspect my IQ isn’t quite high enough to join Mensa but I’ve always flirted with the idea of trying to join and getting in just to see what a group of Mensa people are like.
We are basically random. 2% of population by top IQ test result has pretty much very similar distribution in all aspects to 100% of the population. It's not a high bar to be fastest processing one among 50 people.
If you calculate this properly, a Mensa member has about a cointoss chance to be the "smartest" among random group of 35 people (not 50).
Given that they are probably rarely in groups of purely random people, if you are Mensa member and you are in a group of dozen people (for example in professional setting) you'd probably have a 50/50 chance of not being the one with the highest IQ there.
What should I research? IQ follows normal distribution across entire population. I guess if you'd built your heterogenous population out of mental patients and university professors together you'd get bimodal distribution, but such heterogenous populations don't spontaneously pop up in your life very often.
In my experience, people who make fun of AI's ability to eliminate white-collar work are exactly the same people who are white-knuckling it, hoping they can make another $300k next year to buy that Rivian.
People who don't need a salary look at the situation more objectively and, in general, can see that a lot of white-collar work is in peril.
I think that is tied to the general trend that people see themselves as far less replaceable than others. And I’m not sure that people who don’t need a salary are exactly objective about it - unless you mean people who live off the state or something, those are people who need their capital reserves to continue to appreciate in order to sustain their lifestyles, and are thus relying on the destruction of white collar work to justify the valuations of the technology companies which are underpinning the strength of the markets. It’s like saying in the 19th century that only the propertied classes should be allowed to make decisions about labor policy, because the laborers themselves have their judgment clouded by their position.
But I take your point, and I only brought up the o3 example to emphasize that people on both sides of the booster/doubter debate can be prisoners of the moment. There are boosters eternally convinced that the utopia (or armageddon, for the doomer-inclined) is either already here or imminent, and there are doubters eternally convinced that we have reached the peak. One of them will eventually be right, but neither has been yet. You can’t fault one of them for calling their shot if you’re not willing to see it happening on the other side.
The flip side of that is that at least half of all white-collar work was already useless before AI. If it hasn't already been eliminated, there's no reason to assume a-priori that AI can eliminate it.
Not completely. It just means you can have an AI do the useless busywork instead of a human. Something can be economically useful but still have proponents in a firm.
Call your project Gas Town, write a blog about it, and investors will be swarming all over it thinking managing agents is even more impressive than managing people.
>In my experience, people who make fun of AI's ability to eliminate white-collar work are exactly the same people who are white-knuckling it, hoping they can make another $300k next year to buy that Rivian.
In my experience, people who believe that some white collar industry can easily be automated with AI usually don't work that same role or even that industry, and suffer from Dunning-Kruger bias. In my line of work, I'm desperate for Claude to actually do a better job at not producing a big mess, and it's failing terribly.
I don't work in law, but I feel like I barely need a lawyer if I can ask an LLM to interpret a contract for me. I don't work as a physician, but why can't I just cut and paste an MRI report and ask Claude what to do. I'm no plumber, but it I take a picture of some fixtures I am thinking about replacing, Claude will give me some items to add to the shopping cart from a plumbing supply store.
In all cases Claude was super confident and it wrote what felt really rational.
But in my experience it gets things right, in some amazing ways, but it gets things wrong, in shocking ways.
Everyone thinks it's someone else's job that it'll automate.
Sure. They’re in the same category of wrongness as Zitron. If someone said they follow the newsletter of a guy who claims we’ll never have a recession again, I’d quietly note that.
I don't know about Chegg, but the fundamental problem with Duolingo is their product does nothing to teach people a foreign language. The nasty reality of that product is if you go through their entire learning tree in a language, you might be at a CEFR level A1 for that language. And, you spent 10x the time and 10x the money you would have spent via something like Lingoda to get the same outcome.
This claim is just not true. And gets tiresome. You roughly end up where those courses claim to be in reading and listening - depending on language it can be over B1. (No course finishes B2, some do have B2 content). I ended up being able to watch some (not all) netflix series in foreign langue and I was in early B1 section. I clearly learned.
I speak 4 foreign languages, at (fully tested through exams like the DALF, DELE, Goethe Institute) levels of C2, C1, C1 and B2. Take an actual test proving your knowledge and get back to me.
I've spent a lot of time with Duolingo learners. They're all pre-A1. You're just lying to yourself to make yourself think your phone addiction isn't "that bad." Take a real language class.
It is simple. I do not want to and have absolutely no reason to take the language class. So, why would I? They costs money, time and are not something to look forward to. I got actual visible measurable results. Why would I changed?
Besides, language classes also fail. Going to language classes for years and not being able to use the language in any practical situation was always fairly common result.
> I've spent a lot of time with Duolingo learners. They're all pre-A1.
I know people who tested B2 after doing nothing but Duolingo German. And like I said, I can watch Netflix shows and read books in language I used Duolingo for. That is definitely something.
I don't think Zitron is wrong, exactly. He's just early. There are a lot of analysts who have faced the same criticism over the past 10-15 years. IMO the issue is that they fail to understand the staggering size and scope of the government fiscal + monetary interventions across those years. Essentially, the government has jammed all the risk into the future to repeatedly rescue near-term results.
I think something similar is happening with e.g. Amodei's predictions of mass unemployment due to AI. It won't happen in the immediate term, because deficit spending removes the economic incentive for firms to pare down their workforces.
All that said, I don't think these people are "wrong," per se. They're just early. When the sh*t hits the fan on all this, it's going to be a big problem. And, for example, companies whose primary business is collecting money for Internet ads will come face to face with the reality of how low value their products are. I have some insight into this, as I work for such a firm, and I know the true extent of the bot traffic out there.
- "artificial intelligence has three quarters to prove itself before the apocalypse comes" — Mar 2024
- "If OpenAI doesn’t either reduce their $8.5bn operating costs to $1bn or less and raise at least $5bn in the next year, they will die." — Jul 2024 2025 costs: $34B
- Generative AI “isn’t getting much more efficient” (Jul 2024 ). OpenAI’s frontier-model API price fell from $30/$60 per million input/output tokens for GPT-4 to $4/$20 for GPT-5.6 Sol, alongside major capability gains.
- “I would be shocked if [Musk’s] wealth doesn’t return to something more like he had in 2019 or 2020” (Dec 2022 ). Musk is now worth approximately $873 billion , several times his wealth when Zitron wrote this.
The actual quote and topic is far different than you’ve portrayed it. Here’s the actual quote: “ yet the company says that it expects to make $11.6 billion in 2025 and *$100 billion by 2029*, a statement so egregious that I am surprised it's not some kind of financial crime to say it out loud.” from https://www.wheresyoured.at/oai-business/
He goes on to specifically discuss how unrealistic $100B by 2029 is given that OAI is structurally unprofitable.
The fact that you’re skewing your misinterpretation of what he said so much shows your own bias I’m afraid.
This is another case of missing the forest for the trees. In the scope of what Zitron is talking about, the difference of >$2 billion of revenue doesn't make a difference of his overall point of it not being close to enough to cover their spending.
I follow Zitron but the way he talks has been grating and it is even more obvious how biased he is when he has guests on. As if he's trying to lead them into agreeing with his more extreme claims.
That being said, why does this blogger get a pass at this criticism of just being blanket "wrong" when the point of the quote is still very much valid in context? It seems to be the same in the next article, quoted from Ed's blog with the same point. Is this more evidence of being "wrong" or did he correct a previously wrong value and came to the same conclusion (which seems reasonable to me in context)?
It seems like this blogger is guilty of the same criticisms he has against Zitron. They seem eager to find where Zitron is wrong, exaggerating the value of when he misses the mark and without looking at big picture. Then they make sensationalist claims based on those findings.
I think it's fair to call his specific predictions "early." I'm saying his timelines are too short. Like almost all bearish market commentators, he doesn't understand the scope of the government stimulus. This is a systemic problem in market commentary. But, don't kid yourself. If the US stops massive deficit spending, a lot of what Zitron's saying moves forward on the timeline.
A prediction definitionally includes an outcome, and it _can_ include a timeline. (if you’re serious about your forecasting, then you generally include a timeline; forecasting is non-actionable if it does not include a timeline).
You might choose a different timeline for Ed’s outcomes. Okay: those are now your predictions, not his.
Zitron has done us the favor of including timelines with his predictions, so that Dan can invalidate almost all of them.
It's not government stimulus driving this. It's that when you start with a "successful" company, the financial consequences of bad decisions or bad leadership can take a long time to have a negative effect, if at all, because financial success has a momentum to it.
I left Microsoft during the windows 8 cycle in large part because I could tell nobody knew what the hell they were doing, which is an objectively true statement about the time and place. My dad happened to buy Microsoft stock at that same time and did very well with it.
That's the paradoxical problem that I think all big tech has. Poor decisions by incompetent people, met with inexplicable financial success.
Look at GDP growth (it's not great). Where are the sales coming from? The modal revenue boost to Mag-7 companies since Covid has been a profitless tech firm wasting venture money whose source was debt enabled by low interest rates.
The US can safely reduce its deficit spending by $1 trillion and absolutely nothing big will happen.
The US typically adds that much annual GDP every 16-24 months at this point. The notion that somehow the gigantic ~$31 trillion economy will fall apart if the outsized deficits don't continue, is very absurd.
The exact same things were said of the Bush deficits. The US economy was supposedly dead in 2009-2010. Here in 2026 the economy is 50% larger inflation adjusted and it has left most of Europe in the dust. The housing bubble contagion was much worse than anything we're sitting on now with AI spend.
Why do I say $1 trillion instead of $2 trillion? There are plausible scenarios where increased taxes bring down the deficits (the Dems will take the House + Senate + Presidency, we'll see how much taxes go up), there's no plausible scenario where the deficits go away completely.
I disagree. The massive divergence between US household income and European household income since 2009 tracks almost perfectly Fed interest-rate policy and government deficit spending.
The federal government has spent the time since 2009 lighting the furniture and the house on fire to support unsustainable increases in domestic standard of living by effectively mortgaging the future. The problem we have dwarfs anything that was happening during the Bush era.
Likely the difference between you and me is this: I'm 50 years old, already wealthy from tech, and planning to leave the US. You're probably still trying to earn your way. Sadly, I'm pretty sure we've pulled the ladder up, and folks like you are going to reap the whirlwind, as they say.
Would you say the 2012 apocalypse doomsayers were correct but just "early"? After all the sun exploding in a billion years will ultimately prove them correct.
> OpenAI’s frontier-model API price fell from $30/$60 per million input/output tokens for GPT-4 to $4/$20 for GPT-5.6 Sol, alongside major capability gains.
on this specific point: tokens aren't fungible between models right? Like the argument Zitron makes is that improvements in output are due to, glibly, using more tokens to get there. We see people turn on new models and instantly use up all their tokens.
Like the actual measure is more something like "for this specific task, did it cost less now to do it than it did 3 years ago with these AI pipelines" right? The token pricing isn't actually relevant in that discussion.
If you're saying that OpenAI forecast something and you're saying it's validation that the numbers match the forecast, I don't think you're really paying attention to the problems (with annualised run rate bullshit when these are proper companies who could be publishing proper numbers), with unclear financials that are designed to look good, etc.
> I don't think Zitron is wrong, exactly. He's just early.
By that metric, so was Nostradamus. The apocalypse is coming for sure, we're just quibbling about the timeline.
Realistically, timing is everything. You don't need to get it precisely right, but you also don't get a pass if you, for example, keep predicting an imminent recession through a decade of unprecedented growth.
"I believe we're reaching the upper limits about what generative AI can do and how accurate its outputs can be."
This was said in February 2024. This is months before GPT-4o released. GPT 4.5 was released a year later.
I don't know anyone holding on to models of GPT 4.5 caliber, yet know 4o.
ChatGPT 4o and 4.5 score 8 and 14 points on Artificial Analysis benchmarks. [1]
For perspective, Qwen 3.6 27B, which can be ran on single GPU setups, scores 3-5x that on modern benchmarks.
I don't know why anyone would even make a claim like that in the first place. It's like saying "Computers are never going to get faster". I feel like we could run out of sand and still have faster machines over time. Just a silly thing to say.
After looking at Nvidia's orderbook, it's a reasonable conclusion that nothing bad will happen for at least another year. In which case Ed Zitron is off by at least a year.
But there are enough signs of trouble that could happen soon: Oracle debt is junk. OpenAI might not be on a viable trajectory to IPO. One or both of those could collapse the lower quality data center companies. Zitron is probably overconfident about a crash in the short term. Probably.
> After looking at Nvidia's orderbook, it's a reasonable conclusion that nothing bad will happen for at least another year. In which case Ed Zitron is off by at least a year.
Zitron is fairly clear that nothing is going to happen for a year or so — he says himself that he thinks there's another round of funding possible for both OpenAI and Anthropic.
Right, but you know the thing. The market can stay irrational longer than you can stay solvent. Predicting the medium future is hardest.
Markets don't just pre-plan their behaviour three years earlier and act it out lock-step. Other circumstances can change. By now, the world's financial press has covered some of the scariest aspects of this, and the situation has evolved.
There are other moves (the OpenAI/Blackrock AI debt securitization idea for one) that could delay it even further.
Zitron, I get the impression, has moved on to talking about the horsemen of the bubble apocalypse — talking about banner events that would need to happen for his predictions to be true. This is safer ground for a forecaster, because every forecast affects the future.
His shorter term predictions have not all failed by any means: he described Oracle's woes before the ratings agency downgraded them specifically because of OpenAI.
But most of his predictions will be irrelevant if the insane securitization plan happens. Because it will stop being about an AI bubble then; the worst risk will be the collapse of the entire US economy. It will need a different kind of analyst.
Me, I don't really care either way. I'm not on the cloud AI hype train, I don't work for a YC company, I'm not an American taxpayer so I will not be directly on the hook, and as Americans like to point out, the UK economy is behind on the whole AI thing so (unlike Ireland, which the USA will 100% leave to fail) we are ironically insulated. Maybe a couple of small British investment banks will fail and a pension fund or two will default.
For the most part we'll just watch the flames.
I do enjoy watching a Brit — albeit an ex-pat — upset a bunch of po-faced AI evangelists. It’s like “Itanic” all over again.
I think he is directionally correct. My own impression is that the bubble will burst at the worst time, and so my assessment is that, given the way this is entangled with the functioning of the USA as an economic power and with the future of the US political hard right, it will therefore darkly but poetically burst sometime around Labor Day 2028, which is the worst possible time.
I agree he is directionally correct, or at least is a vessel to raise important points and valid criticisms.
But predictions need to be specific and falsifable. If not, its just rag-chewing over a beer (luv that shit, but i aint predicting on taco tuesday). If they arent falsifable, then its not a prediction.
I think a lot of the HN comments generally can be described as one camp which cares about and enforces the rigor of predictions and trying to direct limited ear-time to voices which tend to get predictions right, vs the other camp that puts more weight towards directional accuracy.
Why would one ever predict anything at all if it was always falsifiable? Do you mean like eventually falsifiable? Like, I don't know about you, but I tend to go ahead and do any falsifying of something first if possible, before I resort to predicting.
Can you give an example of what a good/proper prediction might be, even in a hypothetical universe, in this schema? Does one "predict" when they play blackjack? Or is there a different concept for that kind of thing?
Falsifiable means able to be correct or incorrect. So if i said that my stepmom was kinda whelp, that wouldnt be falsifable, because like what does that even mean. But if I said my stepmom is going to be on her third marriage by the end of the decade, that is, because in 2030 someone can yell at me either way.
This is a spectrum of course: a prediction that OAI will collapse is probably right as _eventually_ all companies come to an end, but under that interpretation, the prediction is useless. It's more signal / useful / falsifable to say OAI is going to collapse around/at <year> due to <thesis>.
Anyway thats my two cents. Its fine to outline forces and trends, but when you make predictions there are useful (better, falsifiable) and useless.
You can have a prediction be specific and falsifable while inside the system, its just then riskier (and less useful).
For example, that french dude that bet on a prediction market what the temperature would be, then broke into the weather station at the airport with a hair dryer. The prediction was still specific and falsifable!
I honestly don’t understand why you care so much about the “rigor of predictions”. Short-term predictions are almost always at least somewhat inaccurate, no matter who’s doing the predicting. Directional accuracy strikes me as far more important.
Take a look at the thread below yours, but basically if the prediction has to be wishy-washy or just interpreted as 'direction', theres way more noise and way less signal. Why not just state the underlying trend/forces instead? Why lean into the online culture of predictions and do it badly.
Bad (form) prediction: OAI is gonna be wobbly in a bit
Good (form) prediction (could be totally wrong): OAI as we know it today is going to collapse due to running out of money around/at 20xx.
And sure we can be pedantic about detail, but the litmus test is: is the prediction useful if you had a crystal ball and you could know if it was true/false a priori?
> OAI as we know it today is going to collapse due to running out of money around/at 20xx.
Now what happens if that prediction is published in a popular, widely-consumed way?
Pundits and analysts who are widely read, talking about a prediction that OpenAI will run out of money by a given date, will change when OpenAI runs out of money.
This is why directional accuracy is more valuable.
Who cares about predictions to this extent? They’re practically always incorrect or flawed in some way, and most people understand and accept that. No need to be so neurotic about it.
That’s the crux of it!! You don’t have to agree but people do care, and I think (ignoring the obvious partisans) that’s the two dominant camps of commenters on this thread.
One pony’s trash is another pony’s treasure, so I guess here one persons pedanticism is another persons hobby.
This is like a grade school-tier insult. The entire article is about predictions, of course I’m talking about them. I just think the obsession with the accuracy of short-term predictions is absurd and pedantic.
Oracle fits in Nvidia's pocket at this point: $67b sales, $22b op income.
Vs of course Nvidia: $302b sales, $197b op income.
Oracle was never as big of a deal as that brief market cap run implied. The herd pushed it up for no reason.
The comparison to Apple, Microsoft, Google and Amazon are pretty similar. Oracle fits in their pockets too. Who cares if their debt is junk. Ellison has nearly wrecked that ship on numerous occasions over the decades. He went on an elaborate acquisition binge in the previous epoch, buying his way to the next stage (preventing Oracle from being market-eliminated, or acquired), and that was an incredible mess that took a long time to sort. He's doing the same move now, trying to spend to stay on the board as the world rapidly changes under his feet.
Non-rhetorical answer to your rhetorical question, but:
The near future of the political right wing of the USA is dependent on the Ellisons staying afloat to create an impervious right-wing media sphere that would survive the end of Fox.
The Paramount-Skydance/Warner merger is now delayed until 2027. Trump/the GOP needs that merger to go ahead, but if Larry's debt position worsens it really might not.
So you can expect the executive branch to push for the USA to backstop Oracle's debt in some way, whether directly or indirectly (taking some sort of stake in OpenAI to allow it to guarantee Oracle gets most of its money, for example — anything to get the credit rating back up).
It will be the first stage of this becoming a problem for the American taxpayer.
You can make that claim for the general case that genAI capabilities will plateau or fishy financing in the sector will cause trouble at some point, but not for the specific talking heads that haven't constrained themselves to that sort of "broad future trajectory" prediction.
FTA:
> Note that I didn't attempt to catalogue statements that are nonsensical or were simply factually incorrect statements at the time, such as his December 2024 claim that “Generative AI's products have effectively been trapped in amber for over a year.” January 2026 claim that "[models are] basically the same as they were a year ago. They have the same efficacy". Zitron has not only made forward-looking statements that AI capabilities will not improve, he's also consistently made backwards-looking statements that capabilities have not improved which, while obviously false at the time, seem to play well to his base (along with his other false statements). If you connect all his statements together, it's implied that AI had the same capabilities in January 2026 as they did in December 2023 (and if you connect later statements, it's actually implied that capabilities in August 2026 are the same as in December 2023, though to be fair to Zitron he frequently contradicts himself and has also admitted to limited improvement at times).
Yes. I was specifically referring to Zitron's circular-financing complaints. I can't defend his separate claims that AI just isn't that useful. If he were involved in the tech industry, he'd know how preposterous those claims are.
The whole point of making predictions is timing. I can tell you the US dollar will continue to devalue (a 100% accurate prediction). But it's a worthless statement unless I can tell you when and how much.
What predictions would you say he is not wrong, but just early about?
If his issue is that he is accurate on something shady going on with the finances of AI companies but the effects are mitigated due to government intervention, he should say.
Your explanation gives him more credit than I think he deserves. He has been unambiguously incorrect many times over the past years in the time scales over which he makes predictions.
In general, claiming that a prediction is just early in an unfalsifiable claim. It allows no admission of being incorrect whether you are correct or incorrect in a given moment. If you are right, then good, you've made a correct claim. If you're wrong, just claim there are clandestine corrective forces keeping the disaster you are predicting at bay. Either way you make it out clean and still have some sense of legitimacy.
I recall a time when Google was struggling to monetize ads and then something changed and they started becoming profitable. The narrative was about improved targeting.
I think it might have helped a little but I never fully bought this argument. I don't think I've ever bought a product which was advertised to me online and I was using some of these platforms for years. I'm probably a liability to them; using up compute but not clicking on ads or buying anything.
Also when I ran social media ads many years back, I never got any users out of it. It literally seemed like mostly bot traffic back then; I can't imagine how bad the situation would be now with LLMs.
Not sure about being early when it comes to points about models reaching the peak of their capability. Honestly, I don't understand how he's qualified (in terms of knowledge) to make such claims.
Being wrong about when a financial bubble will burst is different than being wrong that there is a financial bubble. That's the question, because if AI is driving a bubble, then it will burst at some point.
"Early" works for an open-ended bubble thesis. It does not rescue dated claims that already failed.
Google's 500 million Gemini-user goal had an end-of-2025 deadline. Zitron called it so unrealistic that Sundar Pichai should be fired. Google reported more than 650 million monthly users by October.
"AI had already peaked" is also a claim about the state of the technology at that time. Agent and coding benchmarks moved sharply after it. The fact that every technology eventually peaks does not make a past claim that it already peaked correct.
The bubble may still burst. That would validate the bubble thesis. It would not retroactively fix the prediction record.
btw Zitron has the only fan base that has come after me with doxxing and death threats so far. Truly misery loves company!
Some numbers have a lot of squish to them. Gemini can't be called a failure. But Copilot is a flop and yet Microsoft can probably show you similar numbers to what Gemini has.
But we’re talking about Gemini, which is one of Dan’s point: the surface area is so large that any prediction is almost laughable. Zitron’s prediction here is laughable in fact because he claims both (1) Pichai is jamming AI everywhere and (2) the 500M number is impossible. Those statements are immediately contradictory to anyone who understands the scale of Google’s products.
> The bubble may still burst. That would validate the bubble thesis. It would not retroactively fix the prediction record.
If it does not burst it will be because specific efforts have been taken to deflate it in the face of concerns like those he is raising.
The bubble may not burst for example if the securitization of AI debt really happens. Then when it happens it won't be an "AI bubble" that bursts, it will be a full-on collapse of the US economy. Like when 2008 happened it wasn't really about mortgages anymore.
Macrofinance expert Nathan Tankus has an excellent post on why the financing situation around the AI boom is just not big enough to mess up the financial system.
I agree that the specific amounts invested in AI aren't enough to cause a calamity directly.
The problem is you have the vise of a stock-market decline on one side (something basically everyone thinks is impossible), and AI-induced unemployment on the other side (something a lot of commentators, including Zitron sometimes, seem to think won't happen). Those two things in tandem would be worse than 2009 by a multiple. I doubt the US government will be able to bail it out.
Did you read the post? The author cited very clearly a plethora of very specific predictions made by Zitron that were verifiably and factually wrong. False. Incorrect. Not “early”.
Isn't it super easy for the community to just replace HuggingFace? Isn't it pretty much just a repo/index for open-weight models? Why is it even worth anything?
> Isn't it pretty much just a repo/index for open-weight models?
No, its not just a repo/index (they also have training, inference hosting, and they develop/maintain a bunch of core AI infrastructure software), and even if it was just a repo/index, replacing a bug centralized repo/index that used by an large community isn’t trivial.
To me, the bigger issue with SpaceX is that there's no real argument for how space exploration will benefit humanity in any tangible way. Humans cannot live sustainably on Mars or the moon -- that's just a straight-up lie. Satellites add some value for sure, but it's pretty limited, which helps explain why most of the launches just support StarLink (which itself is probably unsustainable financially). This whole thing is just a bunch of space nerds pushing society down a terrible cul-de-sac which will lower quality of life in meaningful ways in the coming century.
How is Starlink financially unsustainable? In 2025, Starlink generated $4.42 billion in income from $11.39 billion in revenue. Last quarter Starlink generated $1.6 billion in income from $4.3 billion in revenue.[1] So it's already a profitable enterprise, and their costs will likely go down over time as their launch capabilities improve.
Starlink is almost certainly playing games with capex and depreciation to appear profitable when it is not. They're also using political patronage to get contracts from the government to do things that are probably worthless.
The income/profit numbers I used include depreciation, and even if you subtract all space capex from Starlink income, the program would still have a profit of $482 million last quarter. That's a very pessimistic estimate, as it includes costs for many things outside of what's necessary to build Starlink satellites and put them in orbit.
The post you linked to seems to be incorrect on matters of fact. For example, it claims that, "...a large portion of the Starlink constellation consists of V1 and V1.5 satellites...". But of the 9,200 operational Starlink satellites today, 8,000 are v2 minis. Over half of the older v1 and v1.5 satellites have been deorbited, so they don't seem critical to its continued operation. The post also claims that higher RAM and fuel prices will increase the cost of Starlink significantly, but neither are significant costs. It costs SpaceX around $1.4 million to construct a v2 mini and put it into orbit. Even if each satellite had 1TB of RAM, that's what, $50k at the most? And propellant is only a few percent of launch costs. The main expenses (besides satellite construction) are building second stages, refurbishing boosters, and vehicle integration.
If SpaceX is lying about Starlink's profitability, then it seems like the best way to show that is to focus on the numbers. They have 9,200 operational satellites and 12 million customers, meaning each satellite serves 1,300 customers on average. ARPU is $66/month, which means each satellite earns them $86k/month, or $5 million over a 5 year life. (And this is just consumer revenue, not B2B or government contracts.) Each Falcon 9 launch puts 24-29 v2 mini satellites in orbit, so each launch gets them $120-150 million in expected revenue. So the question is: is the cost to construct the satellites, launch them, and operate the network lower than that? SpaceX charges customers $76 million for a Falcon 9 launch, but their costs are much lower. (Probably around $20 million.) Even with pessimistic cost estimates, it's hard to find a way to make Starlink unprofitable without also making things like ISS missions or GPS satellite launches unprofitable. And if that's the case, then arguing over Starlink's profitability is academic, and the real issue is that SpaceX is blatantly lying in their SEC filings. I seriously doubt that is the case.
It doesn't require "lying," per se. It just requires hiding the truth, kind of like what Musk has done for dozens of his other ventures. You do know that SpaceX owns Twitter now, for example? You do know that he tricked his investors into giving his cousin a ton of money for his fake solar lease scheme? You do know that there's still no new roadster, something like 10 years after he accepted deposits for it?
Even if we assume the numbers you're citing are right and this business is insanely profitable, won't this just invite competition? Or are you assuming the government will make SpaceX a monopoly, as it did for Tesla via neverending subsidies?
I'm trying to address one very specific point: Whether Starlink is profitable. I'm not trying to litigate other business practices. The development timeline for a different's company's fancy car for the super-rich is not related to whether Starlink is profitable.
You can't hide the truth if you say you're following GAAP in your SEC filings, and then publish numbers saying that Starlink is making money. Doing so risks massive fines and even criminal penalties.
There are at least three other companies trying to build Starlink competitors (Amazon Leo, Guowang, & Qianfan). It's difficult for them to compete because unlike SpaceX, they don't have reusable boosters. Amazon is hoping that Blue Origin's New Glenn will be able to reduce their launch costs, but until their launch facility is rebuilt, Amazon has to rely on SpaceX, ULA, and Arianespace to get their satellites into orbit. The Chinese competitors also lack reusable spacecraft (and are also working to on that problem), so their launch costs are significantly higher. In the long term, whoever has the lowest launch costs will have the lowest prices, and SpaceX is betting that by the time others have reusable boosters (similar to Falcon 9), they'll have Starship operational. Time will tell if that works out for them.
When you say things like, "Or are you assuming the government will make SpaceX a monopoly, as it did for Tesla via neverending subsidies?" I'm not sure how to respond. First, SpaceX is practically a monopoly because their launch costs are so much lower than the competition. They were responsible for approximately 80% of all mass put in orbit in 2025 (over 2,400 tons). They needed subsidies and government contracts when the company started, but that is no longer their main source of revenue. Second, I don't see how Tesla is a monopoly (there are many competing car companies), nor how they're significantly subsidized compared to other US car manufacturers. How much in subsidies do you think Tesla has gotten compared to their revenue over the years, and how do you think that compares to say, Ford?
ChatGPT estimates Tesla has received $33B in government money directly and indirectly. My guess is it’s much more, since there are intangible benefits such as free marketing. For SpaceX the estimate is $35B+.
The reason no one else has lower launch costs is because no one else has a guaranteed customer willing to waste money on pointless rocket launches. SpaceX’s technology isn’t even particularly novel. If space nerds want Star Trek to actually come true, an actually novel propulsion system is needed, not just scaled up versions of the same thing that’s been done since 1957.
I am absolutely certain Musk has committed securities fraud, repeatedly. I hope that one day the US government returns to normal operations and he gets the Elizabeth Holmes treatment he deserves.
And this has nothing to do with DOGE and woke politics. That’s the one thing where I thought he was on the right side of things, more or less. The problem is straight-up low-interest-rate-enabled financial crime.
The Fed can still declare some kind of financial emergency and increase the money printing again. My guess is this is likely, given the sway that tech CEOs have over the White House, and the reality of overseas competition.
reply