I think it is easy to underestimate how big of a deal this is. Matt has a stranglehold on Automattic and Wordpress, it is not possible to remove his control over the ecosystem and he is volatile. Ousting him as CEO is almost certain to cause a major reaction from Matt, so, for Automattic to make this decision is a very big deal. The right decision for the long term but the hard decision. I expect we will see Matt try to punish Automattic’s board, it will get worse for Automattic before it gets better.
Firing the board is the least Matt-like option. He has absolute control and ownership over the WordPress project, I think it is more likely that he starts his own Automattic splinter company and removes Automattic from the WordPress project. I’d guess he is already filing incorporation docs for Antimatter Inc. or some other Matt themed name.
Automattic has exclusive commercial rights to the WordPress trademark. That was the self-dealing deal he made when he "gave" the trademark to the WordPress Foundation. He can start a new commercial WP company, but it won't have that trademark so would be limited in the same ways he was trying to force on WP Engine etc.
> He has absolute control and ownership over the WordPress project
The only thing he has that can't be moved or recreated are the trademarks. In an extremely short span of time everyone went from Xfree for their X server to Xorg, and Xfree limped along for a bit before falling over. This is the beauty of open source.
I've always wondered about these two. Dunwoody is a retired Army general. Decker insofar as I can tell is one of the suits who ran Yahoo into the ground. As a platform technology company why would you want these people on your board? I could see having the right sort of government suit on it, maybe. I couldn't see "turned Yahoo into an irrelevant hellhole of ads" as relevant experience.
If Mullenweg attempts to fire the board that just suspended him, he would be weaponizing proxy votes against the very investors who originally granted them. While founder proxies are often structured to be irrevocable, using them to purge the board would almost certainly trigger a massive legal battle, attempts by investors to breach or revoke the proxy agreements, and severe financial instability for the company.
While he technically holds the votes required to disband the board, executing that move would ignite a corporate civil war rather than a clean transition of power.
> While he technically holds the votes required to disband the board
This is very unlikely. True Ventures almost certainly controls at least one seat. (Until Phil Black left last year they actually held two seats, but that doesn't mean they had the right to. Either way, at the moment it's just Toni Schneider.)
“there’s a once in a lifetime discount happening at the OpenAI Intelligence Depot, and I brought 15 shopping carts.”
We have to understand current usage with this behavior in mind, there are tens of thousands of people just like this author who are intentionally generating as much usage as possible on their subsidized plans because they feel compelled by some need to get free intelligence.
The only reason these expensive models are generating so much usage is because they are so heavily subsidized. Pragmatic users will shift to cheaper models which will hurt per token revenue, yes, but huge volumes of usage is going to just disappear because there isn’t the demand when it isn’t being subsidized. Less revenue per token and less tokens.
https://tokscale.ai/leaderboard a small sample of just 2k users have generated over $100m of API-equivalent usage while paying closer to just $1m.
It goes a layer up, too. Harvey (legal AI) recently tweeted that a single user query cost that firm 26k USD. This, while they have thousands of users and reportedly whole law firms paying zero. That difference is being subsidized by billions of VC dollars.
A bit of a collective action problem, but if everyone leverages this services as hard as they can without cost concern, better capital decisions will eventually be made. Like the Federal Reserve draining the M2 money supply, AI users must drain the capital supply of subsidized tokens. This pulls forward the future of "Does this tech have value at actual unsubsidized costs?" If it does, tremendous, if it doesn't, also a reasonable outcome to the grand experiment. The current pain comes from the valley of uncertainty we find ourselves in at the moment.
> That difference is being subsidized by billions of VC dollars.
"History never repeats itself, but it rhymes." -- Twain
> I cut this deal with my neighborhood Italian restaurant!
I texted the owner about being miffed they hadn’t told me they were on DoorDash. He replied. They aren’t. We compared pricing, and found the prices advertised are way off from what the restaurant charges. So I placed a $5,000 order to the neighbourhood homeless shelter. DoorDash paid him over $20,000, and I get free pasta for the rest of the year. (My neighbours have also partaken.) Glad to know it’s scaling. SoftBank has assembled a unique concentration of stupidity for itself.
Yes, there are good examples of this working, anti-competitive as it may be.
What if, in the DoorDash example, the pizza shop was itself venture backed and selling pizza at a loss to win customers and using this 20k revenue as a basis to raise money?
Why do everyone assume they are subsidized? When we seemingly have no idea what it costs? Maybe average subscription is breaking even and token spend is pretty much pure profit?
Case in point, claude code seems hell bent on increasing usage at all cost. Which makes sense in the growing phase (get people hooked) but it does not make sense given the hardware shortage. So, which is it?
Anthropic admitted last year to losing money on inference, it had negative margins. The margins have improved and are now positive but there’s still a significant cost. If plans aren’t being subsidized it would mean that the margin on inference is ~99%+ which would mean OpenAI and Anthropic should be wildly profitable but both are still losing money. So, it’s mathematically impossible that they’re not subsidizing plans.
The most widely accepted estimates (though I disagree with them) are that Anthropic’s margin on API inference is ~70% from which people extrapolate what their token usage would cost via the API and compare that to what their plan costs.
re: increasing usage with resets, it’s because they’ve overblown usage and need to show that usage is growing ahead of the IPO. They’re increasing usage on fixed price plans without increasing the cost, the only plausible explanation is they have unused capacity. If they were capacity constrained then the last thing they would do is give away more usage for free.
The last I saw with Claude was that the plans are subsidized somewhere around 10x. The usage of Claude and similar products (for people not paying the actual API token costs) would be lot less if they were paying 10x more per account/seat.
A lot of people are using Claude and ChatGPT for all kinds of minor things at work, and they probably wouldn't be if they were paying the true cost of the product. And this is all the while their work product is suffering because AI is not a great fit for a lot of use cases.
Depends on whether the costs they have are inflated ten-fold due to hardware shortages.
Ignoring current state of shortages, the interesting part (for me at least) is how much a SOTA token costs on current hardware if you exclude the costs for paying for current TSMC shortage and without factoring in the cost of new datacenters being rushed and renting hardware from your competitors (who are also supply limited).
The "actual" cost is of course also relevant and interesting but it is another thing entirely.
Have they actually announced that they're charging less than the marginal cost of inference, or that the revenue they're taking in doesn't make up for the cost of training each newer and better model?
The former does not pass the smell test when there are random providers selling tokens for competitive open weight models.
The problem with judging Alphabet and Meta on aggregate performance is that their advertising businesses print so much money that they can invest everything in a boondoggle and coast when it blows up. Zuckerberg burned $100,000,000,000 on the metaverse and it didn’t make a dent.
That said, Alphabet and Meta are borrowing against their future advertising profits to fund their AI boondoggles. If the advertising businesses keep growing then they’re somewhat insulated from their own missteps but the reliability of the advertising business depends on companies having money to spend on advertising.
The metaverse was mostly self-contained and the failure had zero consequence for the broader economy. AI on the other hand, every major fund is investing everything into AI companies. Every advertiser is using subsidized AI tools to generate hyper specific adverts. If this all goes south, who knows how advertising spend will be impacted.
Google specifically are backstopping billions in data centre build out costs. They’ve committed to spending, like, all of their cash to data centres. If the market gets nervous and funding disappears, Google are deep in the hole.
Anthropic “invested” $50bn in data centers to be built by Fluidstack who raised a billion dollars from Situational Awareness who used their Anthropic ownership to raise money to fund these investments. Google are backstopping much of the Fluidstack build out, i.e: if Fluidstack goes out of business then Google is on the hook for $10bn+ in costs. And Google’s Anthropic investment makes up like $100bn on the balance sheet. So, Anthropic fails to live up to expectations and fails to pay Fluidstack who can’t pay their suppliers which puts Google on the hook to hand over tens of billions in cash while at the same time their biggest investment is going down the pan.
The top line numbers don’t really do justice to the scale of the risk. You need to look at the long term commitments they’re making.
A startup raises $50 million from OpenAI and Anthropic to finance API calls to OpenAI and Anthropic that they are using at a loss who in turn spend that money on compute with Microsoft and Google who in turn invest in Anthropic and OpenAI who then invest the startup using the startup’s revenue to value it… the cycle repeats.
There are multi-billion dollar valued startups invested in by OpenAI and Anthropic with hundreds of millions in ARR that are spending 90% of their revenue with Anthropic and OpenAI.
Situational Awareness, the fund that recently imploded, invested tens of billions into AI companies using their holdings in Anthropic to help finance the investments…
This could all work out fine in the long term, we’re all just speculating at this point, but the circular financing is absolutely making it to revenue because capital invested into startups is used to fund growth which is achieved by subsidizing costs incurred with OpenAI and Anthropic.
The vast majority of startups are not funded by OpenAI or Anthropic. They are not a significant source of venture capital. Meanwhile, OpenAI is pulling in $40B+ per year and Anthropic $65B+ per year.
You are mixing up valuations with liquid cash and you're also making sweeping statements about how those startups are spending their cash. A majority of a raise is not spent on AI compute.
Situational Awareness blew up because they used leverage to invest, and leverage is a great way to blow up any fund even if they were directionally correct about AI.
You’re applying pre-AI investing to a post-AI world. Yes, a decade ago, a startup raised money and spent 90% of it on people. The people built software which had incredible margins. Build it and then print money for ever more. That’s not the case any more, these startups no longer have incredible margins, they’re not collecting $100/m per user and banking $99 of it. They’re collecting $1000 and sending $999 of it to Anthropic and OpenAI.
Revenue numbers are vastly inflated compared to pre-AI but these startups aren’t keeping the money. Profits are worse than ever before. Startups with 30 employees that reach $100m ARR in 6 months are not banking $90m or $80m or… they’re just passing that money straight through to OpenAI and Anthropic.
If startups aren’t just funnelling all their funds raised straight through to OpenAI and Anthropic, where is this combined $100bn in revenue coming from? Who is paying for it? My spend on software certainly hasn’t gone up in a post-AI world. My company is spending less on software now.
OpenAI have stopped being so reckless with their cash investments which is why they appear to have slowed down but they’re still investing millions in huge numbers of startups through token allowances. They invest $2 million in every YC startup (or did a few months ago). There’s an entire market of reselling these tokens!
$800M a year is less than 1% of their quoted revenues at $100B+ / year. Claiming credits as revenue would be tax fraud. Credits to clients for services are counted as debits against
There are zero serious companies collecting $1000 on revenue and sending $999 as a cost of goods sold to Anthropic/AI. It would be unprofitable to even run a proxy to Anthropic on such thin margins. But I digress.
No company was banking $100 and keeping $99 in the "before times" either. These are fantasy numbers not even the most highly optimized software company produced. As an example, Slack famously went public in 2019 and it had revenue of $401M with a gross margin of ~79%, meaning they were pulling in $316M in gross profit. That is the figure before labor, administration, R&D, sales & marketing, etc. They actually operated on a net loss after factoring for those expenses, despite their high gross margin, which is common in high growth startups (Amazon famously ran losses or marginal profits until decades after their founding because they continuously reinvested in expansion).
Credits reduce revenue by all basic accounting standards. You can accuse these companies of fraud, it is within the realm of possibility, but it would also be <1% of their total quoted revenue, so not really worth the heat at the same time.
You are making conflicting arguments at the same time. There exist startups that are able to generate gross profit with some consumption of AI services, they are also able to invest nearly 100% of their capital into AI to generate those profits without needing to spend on traditional labor, and yet AI is not sustainable. By your own circular logic it is of course sustainable, but by grounded logic, you have to understand any business that goes from zero 4 years ago to $100B+ in annual revenue today with double digit growth rates is offering the world something of value. Anyone who has tried AI sees some value in it. There is some revenue and profit to be made here. Betting against that in the long term will just lose you money and sanity.
> No company was banking $100 and keeping $99 in the "before times" either.
Yes they were and are. The marginal cost of software as a service or data as a service is near zero. Slack is a good example. A new Slack customer costs Slack nothing. Free money! Slack had a high valuation because of the margins. Slack and other traditional high-growth technology companies were valued highly despite being loss making because there was an understanding that paying for growth early returns a lot more later on. Slack (pre-acquisition) could turn off their expensive growth engine and start making money hand over fist.
(Look at what Bending Spoons are doing now, they're picking up "zombie" technology companies that have incredible margins but no growth. Bending Spoons are cutting these companies to the bone, giving up on growth, running on a skeleton staff, and making money hand over fist, cashing out on the incredible margins of software.)
Someone shared up thread an example of Harvey, a legal AI company, who regularly post about their token consumption. They're consuming trillions of tokens per month for their product. Harvey's investors include OpenAI. Harvey has raised more than $1bn and is currently valued at $11bn (and raising again at $15bn apparently). As of last month, Harvey's revenue was reported to be $30m/month on 13 trillion tokens per month.
Let's be conservative and assume their average spend per million tokens with OpenAI is $2. That's $26 million in token spend per month, on $30 million per month revenue. $2 is lowballing it of course, they're surely using one of the frontier models. That's pretty close to every dollar coming in going straight out to OpenAI. Considering the capital they're raising and burning (seems like $50m a month) while relatively small (<1k employees) I would guess they're spending at least double their revenue with OpenAI.
Of course, long term, this is fine for Harvey, as model costs come down and businesses mature they are going to be spending a lot less. Maybe they'll start running their own hardware, offloading certain workloads to cheap models, using scripts for routine tasks where AI is overkill. Great for Harvey and Harvey's investors, an absolute disaster for OpenAI.
> By your own circular logic it is of course sustainable, but by grounded logic, you have to understand any business that goes from zero 4 years ago to $100B+ in annual revenue today with double digit growth rates is offering the world something of value. Anyone who has tried AI sees some value in it. There is some revenue and profit to be made here. Betting against that in the long term will just lose you money and sanity.
You're making a leap from "useful" to "profitable". Yes, there is absolutely revenue and profit to be made for companies building products, for the companies providing technology, not for the companies providing inference. There are not software margins in inference, it's a commodity, the only reason OpenAI and Anthropic went "from zero 4 years ago to $100B+ in annual revenue" is because nobody cares about the money today.
Right now, we're in a gold rush, we're in the growth-engine phase, we're in the "spend a billion to make a million as long as you're growing" phase. Right now, people at Harvey aren't worried that every dollar in is at least a dollar out to OpenAI, who cares, investors are funding it, they're growing, they're taking over the legal world, that's all that matters, they can balance the books later... and when they do start to balance the books, when they convert that growth-at-all-costs into profit (as every company eventually does) OpenAI are going to get absolutely eviscerated.
The circular financing problem doesn't mean that startups building on AI aren't generating revenue from normal companies, it means that the money going into Anthropic and OpenAI is coming from investment (whether OpenAI directly or indirectly (see: funds like Situational Awareness raising money off the back of their Anthropic investment)) and being immediately spent on inference. If the economic environment changes, OpenAI don't have a growth engine they can turn off to turn their revenue into profit... because it is their customers who are going to be collecting the profit.
We can see this already with OpenAI starting to try and bill based on solutions through ChatGPT (because they realise selling tokens is a god awful business to be in) and their partnerships like The OpenAI Deployment Company. OpenAI and Anthropic are triple screwed no matter whether AI is a wild success beyond your imagination or a disaster.
This is so completely wrong and deluded I’m not sure where to start.
I work with AI startups and scale ups on a regular basis as well as plenty of more old school companies, all of whom are spending money on AI models, because they are getting insane value from them.
This idea of the revenue for OAI and Anthropic coming from “circular financing” is just bizarre wishful thinking coming from AI doomers with zero financial literacy.
The revenue numbers reported by AI companies (not just OAI and Anthropic) isn’t being driven by Nvidia at all, in fact, the numbers wouldn’t add up if you thought that was the case. The revenue being brought in by AI companies is far, far higher than the sum of any investments from Nvidia.
The AI doomers just can’t handle the idea that AI is actually incredibly valuable and every company is using it and increasing their use of it every month.
And yes, I see this every day in my job and with every company I work with.
I also contract in the startup space, and many of these startups have pretty much their entire infra bill covered by AWS/Azure/GCP credits, and all of their AI spend covered by Anthropic/OpenAI credits.
Theoretically they'll spend real money on those things down the line, assuming they find product-market fit, but who knows how many of the current crop of startups will reach that point
Wow, thanks for your perspective, it’s lucky to find someone on Hacker News who works with technology every day!
You presume to know my position but you do not. AI is an innovative new technology that is radically changing how we build and use technology and will continue to do so. That doesn’t mean that trillions of dollars is going to be spent on it. Despite the penetration all technology has in our lives, most companies are barely using technology from 20 years ago because implementation is a nightmare. Businesses are risk and cost averse, better the line item you know. And so, most companies could be radically improved not by human-level intelligence, or even dog level intelligence, most companies just need macros that are easy to implement. Most companies could 10x their productivity without AI! After all that’s what startups have been doing for the 20 years pre-AI, that’s been the YC investment thesis (which has worked very well).
My position is that AI is a radical step forward in technology that pragmatic businesses will benefit from handsomely by using cost effective models. A middle of the road local model that can trigger tools is more than most companies need. The frontier models by the frontier labs are a complete waste of money outside of the most extreme edge cases.
Conflating “the technology is incredible” with “companies will spend trillions per year on the technology” is ridiculous. Your argument about usage says absolutely nothing about the financials yet you’re dismissing the AI “doomers” (people who are pessimistic about the financials, not the technology) on that basis.
If you look at what we know of the financials of OpenAI and Anthropic it is impossible to come up with a financial case to justify the trillions of dollars in revenue needed for the AI booster’s vision of the future.
How much money does The JavaScript Company make? How much money did Docker make? It’s like the AI boosters who argue for the financial case have forgotten the last 20 years. The world of technology is built on open source, it’s built on companies that made a huge impact and failed financially. Docker led the way with containerization, one of the most influential technologies of the last 20 years, and the company almost went under multiple times. We constantly gripe about how unsustainable open source is. Why is all this suddenly different? Why is making an innovative new technology suddenly guaranteeing trillions in revenue? How many trillions of dollars were invested in data centres to build Docker containers?
If you think I lack financial literacy, please explain where the money is going to come from. Please make the financial case for trillions of dollars being spent on AI over the next few years. Keep in mind that the reason technology has been so profitable over the last 20 years is because of the margins, software is basically free money. AI is not free money. AI is very expensive money. Also keep in mind that the current (rumored) revenue of Anthropic is primarily made up of the most expensive use case (generating millions of lines of code) being paid for by rich tech companies which does not represent the wider economy. A factory could revolutionize their operations with a middle of the road model they could run on local hardware. Hell, they could revolutionise their operations by hiring a single competent software engineer who understood their business. AI is so compelling because we, technologists, have failed to deliver for most businesses, not because businesses need frontier AI.
Bets are meaningless but feel free to stake a claim here to how you think things will be 4 years from now and we can return to review. I’ll stake my claim: AI will be more impactful than ever while Anthropic + OpenAI will have less revenue than today. And we will all be thinking “wtf were we thinking building all these data centres?”
It's coming from regular companies spending their own money on using AI. I.e. Profitable companies deciding they want to use AI for various reasons and spending their own revenue on said AI, whether it be Claude Cowork, OpenAI ChatGPT Work, OpenRouter, Nebius Tokenfactory, models hosted on BaseTen, Fireworks, etc, tools like Lovable. Or every piece of cyber software which are ALL using AI heavily these days.
It's really not as complicated as AI-doomers like to make out, they seem so confused somehow that existing profitable, successful companies are spending larger and larger amounts of their revenue on AI. It's not circular by any definition.
So anyway, this whole worry about "where the money comes from", is kind of funny. Where does the money come from to hire employees? Where does the money come from to pay for Cloud bills? The money for AI will come from the same place, it's not some big mystery. Total cloud/compute spend in the world is well over 5T per year, including all cloud and colo spend.
AI is basically both taking up software spend, dev salary spend, white collar officer worker spend, hardware spend, general IT spend, etc. And if you sum up all the budget associated with all company software, personnel, white collar workers, etc, you end up with a much bigger number (probably 10-20T or more most likely).
So the idea of total AI revenue being in the trillions, it's pretty simple, and will happen over the next couple of years, just like happened with regular servers and cloud.
People like yourself downplaying AI reminds me of people both downplaying the internet ("it will never make money") and also the original launch of cloud providers (AWS originally), "no one will ever trust the cloud not to lose your data, why would any pay for AWS" etc etc).
Both sets of folk were radically wrong, and the AI-doomers will be wrong this time too. Capabilities will increase across the board, amazing applications will be built (already happening), and people will want to pay for these products. People are ALREADY paying huge amounts of money for these products.
Who do you think is paying for Lovable? They are probably the fastest growing startup ever from 0 to 1B valuation because less technical people LOVE using it and have zero issue paying for it. But AI doomers will somehow dismiss Lovable as somehow getting "circular financing", when loads of small business owners I know love the product and spend 100s of dollars a month on it!
It's going to be funny watching the doomers over the next two years when none of the big AI companies goes bankrupt and keep growing in revenue. But but but the circular revenue!
> existing profitable, successful companies are spending larger and larger amounts of their revenue on AI
Apart from the brief "tokenmaxxing" craze among the big tech firms a while back, is there any evidence that profitable companies are cutting their own margins in order to spend on AI?
Lovable's success is the perfect example. Lovable has a large number of users who do not pay for the platform who have been subsidized by investors while inference costs were high. Lovable know that long term, sending all their revenue to Anthropic and OpenAI and Google is very bad for business, especially if that revenue is subsidized by investors, which is why they have trained their own model. Lovable's long term success is in conflict with OpenAI and Anthropic! Lovable succeeds when it stops sending $0.50 of every $1.00 to OpenAI and Anthropic and Google, Lovable succeeds when it drives down the costs of inference to as little as possible.
Regarding the cloud infrastructure comparison, it is not at all comparable. During the time I spend writing this comment, my device will make thousands of requests and connections to different servers for all sorts of reasons. During the time I spend writing this comment, my device has interacted with an LLM exactly zero times. The throughput of internet infrastructure is not even in the same universe as the throughput of LLMs at their most wildly successful. How many times does Lovable's AI run per month for their average customer? A few times? The repeated, continued value Lovable delivers to their customers is in the interactions that occur between their customer's customers and their customer's apps. A Lovable customer can love Lovable and have huge success with their Lovable app while using zero tokens per month.
You should be comparing AI to a product that eventually became commoditized, not comparing it to an entire category, e.g: shared website hosting. Shared website hosting was very expensive to set up 30 years ago. Over time, it got cheaper and cheaper, now today it is commoditized, the major brands have all consolidated, companies have gone under, and technological innovations have completely reshaped how websites are hosted. Who still uses shared website hosting today? Websites are bigger than ever, web servers underpin the economy, Stripe alone has web servers that process trillions of dollars... how much money is there in web servers?
> Capabilities will increase across the board, amazing applications will be built (already happening), and people will want to pay for these products. People are ALREADY paying huge amounts of money for these products.
You're so caught up in the technology that you're oblivious to the economic reality. The capabilities, the amazingness, the excitement, that isn't how money is made. The most cheap and boring technology (like web servers) are fundamental to our economy. AI can be all of these things, it can have incredible capabilities and be amazing and have so much excitement and radically reshape our economy and be fundamental to every business... and make no money.
You, like so many nerds, cannot seem to separate technology from business. Business is boring and simple and based on principles that have stood the test of time. Business doesn't run on excitement, it runs on numbers. Shopify powers most ecommerce, Shopify is one of the most important companies in ecommerce, Shopify is wildly successful, Shopify's revenue... $12bn. Stripe's revenue, on trillions of dollars in payments... less than $10bn. Shopify and Stripe are wildly successful and very important companies that are involved in trillions of dollars flowing through the economy and you're suggesting that AI is going to do 100x more revenue than them?
Tailwind CSS is used on probably half of all major websites today. The creators of Tailwind recently announced they had to lay off everyone because the company was struggling to make any money despite usage growing every single day. WordPress, which (supposedly) powers half of all websites is operated by a company that is struggling too. Google and Meta, some of the most profitable companies in the world, almost all of their revenue is still from advertising that has barely changed in 25 years. Google make hundreds of billions of dollars from... showing videos, technology that existed 25 years ago.
And so, that brings us back to the circular revenue argument. Right now, Anthropic and OpenAI have ~$50bn revenue each because of circular revenue, because of investors subsidizing, because in this experimental period, everyone is throwing shit at the wall to see what sticks, nobody wants to be left behind, they're digging for gold.
If you want to make a compelling argument for why AI will impact the economy, that's one thing, but to argue that Anthropic and OpenAI are going to generate trillions of dollars in revenue from it is an entirely different argument altogether. They're completely independent. One of them is a reasonable argument (of which people can debate the extent) but the other is absolutely batshit and indefensible.
> It's going to be funny watching the doomers over the next two years when none of the big AI companies goes bankrupt and keep growing in revenue. But but but the circular revenue!
I bet that Anthropic and OpenAI's revenue will be less than $100bn each in September 2028. At their current rate of growth based on the ai boomer takes, it should be well over $250bn each by the end of 2027.
Can you provide any examples of any of the megascalers publishing any detailed financials that touch on their AI spend or revenue or profit? The only one I’m aware of that comes close is Microsoft and they have still buried it in barely related line items which still leave us making assumptions.
There’s speculation on both sides and certainly Zitron is on the extreme end of the anti-AI side with the most cynical speculation but it is indisputable that none of the megascalers are open about their AI financials. Hence, we are all speculating endlessly. If only there were published financials then the speculation could end!
The obfuscation of financials doesn’t necessarily mean something bad is happening, it could be a competitive advantage for Google to be secretive about how cost effective their TPUs are or for Microsoft to hide how much revenue uplift they’ve experienced by adding AI to 365.
They don't publish their AI spending, revenue, and profit - but they do tend to break out other non-AI segments of their companies which can demonstrate that at least part of their growth isn't relevant to AI.
> They don't publish their AI spending, revenue, and profit - but they do tend to break out other non-AI segments of their companies which can demonstrate that at least part of their growth isn't relevant to AI.
So the previous statement that "As public companies, the megascalers publish pretty detailed financial reports" is incorrect and irrelevant to the question that was asked.
I'll expand the section of the article that it quotes:
> Although this wouldn't be in the spirit of Zitron's statement, one could argue that Meta is actually dying, it just hasn't died yet. However, the reasoning in Zitron's argument is incorrect here—the Meta, Google, and Microsoft ecosystems are not dying. Given how fast these companies are growing (in terms of revenue and profit), it doesn't seem that AI is, as Zitron implied, some kind of desperation move they're reaching for because "they don't know how to grow" and are all out of ideas.
So the argument here is that Ed says those companies are dying, but Dan Luu points out that their economic figures show that they are not.
The counter-argument is "Growth isn't a valid rebuttal, unless we can also sus out how much of that growth is tied up in circular financing of AI projects"
My point is that the public reports of these companies, while not helping us unwind the circular financing, do at least show us that their non-AI businesses are growing at a healthy pace. Which supports Dan's argument that these companies are not dying.
I don’t think this line of thinking is particularly robust because it ignores how AI is being used and where the resource usage is coming from.
Right now there are a small number of very very resource intensive use cases that are being subsidized by OpenAI and Anthropic. There are people generating millions of lines of code because it’s basically free at the point of use, despite the code producing very little value. Anthropic and OpenAI frequently “reset” customer limits to allow them to use even more resources at no additional cost.
The majority of use cases across business are not generating millions of lines of code per employee. The majority of businesses need just a little bit of automation to radically improve the way they operate. A software engineer making endless projects because it’s free to do so might use hundreds of billions of tokens per year, but an entire manufacturing business could be revolutionized with a few million tokens per year.
I think 2 things can be true:
1. There is very little penetration of AI across the economy and huge room to grow in the number of businesses deriving economic value from AI
2. The compute usage today is vastly overrepresented by usage outliers who are not paying the cost of their usage and will stop when forced to pay the cost
We could see AI usage 10x while seeing compute decrease 10x if the type of usage shifts. Most businesses just need smarter macros.
> Anthropic and OpenAI frequently “reset” customer limits to allow them to use even more resources at no additional cost.
Surely this applies to fixed-price subscriptions, not per-token spend? Large enterprises (the "very very resource intensive" large-scale users) have to pay per token.
A lot of companies avoid paying for usage by encouraging their employees to use individual subscriptions. Outside of the short lived tokenmaxxing fever dream, enterprises are conscious of their usage with companies like Uber and Amazon reigning in their usage massively and companies like Ramp building their own routers for cost minimization.
Facebook is reportedly the company that spent $500 million in a single month on tokens. There are individual non-enterprise users rotating multiple subscriptions incurring $10k+ in tokens per subscription. Facebook’s $500 million month… is equivalent to ~10k individual subscriptions which could be as little as a few thousand of the heaviest users. That’s $500 million when billed on usage, or ~$2 million on plans.
The reason resets are such a big deal (people have set up websites to track them, tweets announcing them get millions of impressions) is because there are huge numbers of users pushing their plan limits every single day. If there was huge demand from usage-based customers (the large enterprises) that OpenAI and Anthropic couldn’t meet, they wouldn’t be handing out resets like candy.
I think a realistic belief is that Anthropic and OpenAI have vastly overstated demand and are using resets as a way to keep usage artificially inflated at a substantial financial cost. I’d guess fixed price plan users make up at least 95% of usage.
Risklytics are taking the boring, sensible approach of being a broker, modernizing the application process to broaden the type of clients that can be underwritten.
Corgi are taking the batshit insane approach of insuring clients directly through an esoteric insurance structure that dodges regulation.
Choosing Corgi over a typical broker is playing with fire.
I'm finding Corgi so far to be really difficult to understand. The sales reps (agents?) you work with don't seem to understand common insurance industry terms, like me asking "is that actual cash value or replacement cost?" when e.g. stating a value for replacing our computers.
I also have open questions on what paying out a claim is actually like, or what it will be like if they have to mount a defence if I were to actually get sued. Since no Corgi customers have been sued yet (as far as I can tell), the answer is "We have no idea".
I'm not trying to be anti-Corgi, but I'm having a hard time understanding how exactly their business model is going to work long term.
Yep, being insured by Corgi is a huge risk. The reason why Corgi is especially terrible is that the inexperienced “growth” “interns” aren’t just selling low quality insurance they don’t understand that isn’t fit for purpose, that’s bad but whatever, the problem is that every new client adds more risk to pre-existing clients! Every Corgi client is, unknowingly, taking on more and more risk every day. The insurance is, basically, decaying, and entirely reliant on Corgi being able to raise money to bail it out when things implode.
The business model of spending hundreds of millions on marketing insurance with cafes is silly but maybe it’ll work, gotta swing for the fences, but selling insurance that could implode at a moment’s notice is grossly irresponsible. And when it implodes, it’s going to hurt thousands of YC startups.
Risklytics will be much more expensive than Corgi but it’ll be actual reliable insurance. Anyone going with Corgi is out of their damn mind.
My biggest question is how I would open a claim. I can’t find a hotline number to call to open a claim (or to give to someone else who thinks they have a claim against me).
The degree to which this lacks empathy for the struggle to survive that people have the world over is harder to believe. Sincerely, though, all fishing is pretty terrible, we (the rich west) have been overfishing and pillaging the oceans for decades. Even now in peak concern for the environment, we really don’t care nearly enough. If you personally want to avoid contributing to ravaging of the planet, switch to a low impact diet (e.g: only eat chicken, or go vegan).
Not to do poverty Olympics, but I once was poor, living on an island, and reliant on fish for a lot of meals. Don’t assume I can’t empathize.
Why so quick to make excuses for selfish and mind-bendingly stupid people? Why are you assigning guilt on the “rich west” for a local practice that is for local consumption? Not every indigenous practice is good. Some are the reason those communities can’t escape poverty. It’s not colonialism or oppression, their problem is stupidity.
I think some people belong to the very popular new religion that holds as dogma that anything bad done by someone in the developing world must be the fault of evil Westerners somehow.
> It’s not colonialism or oppression, their problem is stupidity.
The $200 max plan is for individuals. The individual plans are heavily subsidized. Employers should be using either the Team plan (which has much lower limits than max) or the Enterprise plan (which is entirely billed on usage).
Anthropic know that lots of people are doing all sorts of “bad” things like employers paying for Individual plans, (and using multiple accounts to get more usage) and aren’t yet enforcing the rules… but by the letter of the Anthropic terms, your employer should be paying Anthropic a whole lot more (and that’s one of the reasons why AI usage is going to get very very expensive as soon as the subsidies stop, you and a lot of other people are already paying a lot less than you should)
AFAIK there is nothing in the ToS that forbids an employee paying for the 20x, $200/month plan. I could be wrong about this in which case it'd be useful to have a link the clause.
I think multiple plans are against the ToS, but I'm not doing that.
The Teams plans are more convenient for a number of reasons, but yes, they top out at the 6x plan, not the 20x plan.
I've re-read it and I'm pretty sure there is nothing that forbids a business paying for a 20x account. Notably they say this in the consumer ToS:
> If you use an email address owned by your employer or another organization, your Account may be linked to the organization's Anthropic enterprise account, and the organization’s administrator may be able to monitor and control the Account, including having access to Materials (defined below). We will provide notice to you before linking your Account to an organization's enterprise account.
which goes at least moderately close to indicating using it in a work environment is allowed.
Not this again. There is not a single shred of evidence for this. In fact multiple times this year alone, people from Anthropic have said that inference and deployed models have positive margins. The big bucks are always being spent on training the next model.
It's a simple fact that paying enterprise token rates would cost many times what the individual plans cost. It may well be that the enterprise income outweighs the cheaper tokens on the individual plans for net profit, but using the individual accounts as subsidy is a common tech industry tactic and what you said doesn't prove they're not doing it, either.
It's a little hard to believe they wouldn't be trying to slow the burn rate for an IPO if the inference were truly so profitable. And i find anything they say a little hard to believe all the time anyway (though admittedly they're a lot more trustworthy than OAI)
I strongly disagree. GitHub, a year ago, acknowledged the fundamental problems and began work on them. We all agree with the diagnosis and strategy: stop building new things, bring stability. Why would whether the CTO codes have any bearing on the correctness of this strategy? GitHub’s problem isn’t that leadership don’t understand the product, or that they don’t know what they should be doing, it’s that they’re battling unprecedented demand. If it was a disconnect between users and leadership on what matters, sure, a CTO who doesn’t use the product would be notable, but that isn’t the problem. And that’s all assuming he doesn’t actually use the product, maybe his privacy settings hide private commits.
Interesting - There's a couple of things that I disagree with here, but I do think this resonates: "We all agree with the diagnosis and strategy: stop building new things, bring stability."
Where I have a problem with the positioning of their GitHub profile is, he's the CTO of GitHub, arguably the defacto standard for open-source version control systems. His GitHub profile is linked to as the author for the post, and his GitHub profile simply tells me: "this guy doesn't code."
I don't care if this guy doesn't work on GitHub itself, I hardly would expect that, but IMO, any CTO of a company like GitHub should eat, breathe, and sleep code. He might, but his profile, which is being published as if it means anything, tells me he doesn't.
Here is Matt talking about his control: https://news.ycombinator.com/item?id=42657150
reply