But are they actually profitable, or do they employ creative accounting where only parts of overhead expenses are counted against all of inference revenue, similar to what Uber did?
OpenAI's numbers show that they definitely are not profitable on inference, and even worse, revenue growth scaled linearly with inference cost from 2024 to 2025, which means they can't outgrow this problem. See https://www.wheresyoured.at/oai_docs/
Does it matter if it’s creative accounting? Uber is a great example of a company that everyone was certain would fail because it was unprofitable and now it succeeded and is profitable.
If they shut down all training today they’d be absolutely printing money for the next couple quarters and then die with a bang once the other lab releases the next frontier to the public.
I'd be surprised if they're making money on inference just from that. There's no way someone paying $20 p/m and using it all day is not spending way more on even just the electricity for tokens, let alone the capex.
I don't really get the last bit. It's hard to imagine what a new fangled "frontier model" could do that would blow anyone out of the water. Like what does this look like? Really good benchmarks? Who cares about that anymore?
How? They're already burning $2 bills to make $1, court documents shown that Anthropic has already been lying around revenue (claimed to have made $19 billion when it's actually $5 billion to date [1]).
Not hard to believe they're lying about other things when they've been lying about the capability of their products since inception.
That is not what the article says, it says $19B ARR.
I don’t necessarily see a contradiction. $19B run rate, achieved very recently, is actually consistent with $5B lifetime earnings, because their growth curve is so sharp. Zitron is not good at math.
Didn't link to Zitron site but if you can't see how dishonest it is to say you have $19b ARR when the reality is you have only a total of $5b IDK what to tell you. Says more about how you think and why you think it's okay for corporations to be misleading.
This is not lying, that is just what run rate revenue means! It makes sense to use as a metric when a company’s user base is growing as fast as Anthropic’s is.
It makes sense to be extremely misleading about actual accounting figures? In what world is it okay to say you have $19b in ARR when you have only ever generated $5b for the entire duration of your company's existence?
Did Enron start a business school I'm unaware of something?
Well I happen to have a phd in that broader domain. It's not censorship, as you imply, but IQ is just way fuzzier a concept than people outside of this area of research think. The popular view is IQ is an objective thing, exactly measurable and so on (the metaphor of brains being computers, essentially). In reality you can put a 14 year old from a bad environment into an optimal environment and their IQ increases by up to 20 points over a view years.
IQ is highly heritable, so I don't think there is any environmental factor that has this big an impact, except in extreme cases like severe malnutrition. Also note that IQ increases with age roughly into early adulthood independently of environment, so the IQ of a 14 year old increasing is a perfectly normal part of the heritable parts of development.
> The results show that the heritability of IQ reaches an asymptote at about 0.80 at 18–20 years of age and continuing at that level well into adulthood.
If you don't know the people doing work in this field today, and instead are just Googling "heritability of IQ numbers" (which would explain the Wilson Effect paper you kind of inexplicably posted) you can find previous threads that I've discussed this in using the search bar below. Either way, "you probably just made this up" is not OK on HN.
Do you have any argument as to why what a social scientist says today is more valid than what one said 20 years ago?
I'd trust social scientists less today than 20 years ago due to the effect of social media on them. Social media creates much stronger social pressure on people to conform, and that isn't a good thing for science.
Ah ok, yeah genetics science has came a long way 20 years so there it makes more sense to listen to modern stuff, but why the scare quotes? Do you have a problem with social scientists?
Edit: Looked him up and he disagrees with you. "My sense is that heritability of IQ is in the range of 30-70% with very high confidence.", you said "It's likely somewhere between 15%-50%". There is a massive difference between 30-70 and 15-50, 30-70 sounds much more reasonable and matches most studies on the subject I have seen.
No, I don't. But you've lost track of the thread, because I didn't claim he said 15-50%. There's a reason I cited him where I did: to point out how silly the 0.80 estimate the previous commenter's cite was in current context. His antecedent in this thread is "even hereditarians...".
But now we have an authority figure saying 30-70, so I'll trust that over your 15-50. And yes I know why you didn't cite the 30-70, its because you disagree with it. You shouldn't say others use biased examples and then say you believe it could be as low as 15% without anything to back that up.
He didn't use a "biased example"; he used a prehistoric example based on premodern methology. The only thing we've established here is that he doesn't understand his own cite.
later
In other words, I deliberately cited someone on his side of the debate.
How much do I love that this person got promoted from woke social scientist to "authority figure" in the space of one Google query, though? Amazing.
Obviously scientists have thought about that and therefore administer age-adjusted IQ tests for different cohorts, precisely with the aim of IQ staying constant during aging of an individual. And yet here we are.
You really don't need a censorship conspiracy to explain these things. I'd recommend you try to challenge your assumptions about IQ and heritability by downloading a few textbooks about the topics. Many many papers are freely available, and the textbooks are let's say easy to find. You could try textbooks about a more accessible area like developmental psychology as it is more easily accessible and still covers these topics quite well.
The problem with the "I'm right, you are wrong, educate yourself" reply is that, because of the taboo, different books will say vastly different things about this topic. So you have to decide which books are the "good" ones and which are the "bad" ones.
Me too. Switching my home system from Ubuntu back to Debian was influenced a lot by snap. I don't get how they could fumble that one so hard. It goes against everything they used to stand for. If I want a bloated, slow, closed-source, proprietary app store with unclear security ramifications, I'll install MacOS or Windows. It also feels like app developers at least care a little bit about those stores. Mozilla for example still officially recommends installing their Debian package rather than through snap on Linux, despite shipping via snap by default on Ubuntu now.
Local LLM inference is all about memory bandwidth, and an M4 pro only has about the same as a Strix Halo or DGX Spark. That's why the older ultras are popular with the local LLM crowd.
Those are of almost zero use for people wishing to run Linux etc.
Yes, Asahi exists, and props to the developers, but I don't think I'm alone in being unwilling to buy hardware from a manufacturer who obviously is not interested in supporting open operating systems
the macs outperform it and I figure it's a better general purpose computer than strix halo. if budget is a problem, then a strix halo is a decent alternative.
Well a mac isn't really an alternative to a mac, or is it? ;)
Personally I'm not interested in having a mac as I work with linux. And yes, they outperform them, but only if you ignore the price. When comparing what you get for ~$2k, a Strix Halo is miles ahead.
Yeah I'm pretty happy with the M5 (beside the look). It's most probably the same SixUnited board most Strix Halo devices use (including the ones from HP and Lenovo).
Still lot to learn, but after a while you have something like Qwen3-Coder-Next-Q8_0 running and - at least for me - it works quite well, both as ChatGPT like chat-interface using llama.cpp and as coding agent
I'm not really using them for coding (only played a little bit with minimax2.1), which is probably the most common use case here.
I mainly use them for deep work with texts and deep research. My main criterion is privacy, both for legal reasons (I'm in the EU and can't and don't want to expose customer's data to non-gdpr-compliant services) and wouldn't use US services personally either, e.g. I would never explore health related topics chatgpt or gemini for obvious reasons.
Technically I've set it up in my office with llama.cpp and have exposed that (both chat interface and openai compatible api) with a simple wireguard tunnel behind nginx and http auth. Now I can use it everywhere. It's a small, quiet and pretty fast machine (compiling llama.cpp is around 20 seconds?), I quite like it.
The part you are leaving out is treating them solely as "illegals" means these people will avoid going to physicians to get their shots, because they would risk deportation.
The obvious solution for better health for all would be providing public and freely accessible locations for getting these shots, or mobile teams providing them at schools etc.
reply