Rendered at 22:52:09 GMT+0000 (Coordinated Universal Time) with Cloudflare Workers.
simonw 2 days ago [-]
> Since our Series G in February, adoption has continued to grow across global enterprise customers, and our run-rate revenue crossed $47 billion earlier this month.
OK, so their self-reported run-rate revenue hit $47bn in early May.
And $9bn in December (according to the above April 6th link.)
mgfist 2 days ago [-]
Pretty unfathomable growth. I'm pretty sure I listened to Dario saying something along the lines of keeping Anthropic on track for 10x ARR growth (in December) and thinking that that was a bonkers idea for a $9B run-rate company, and now it's looking like that might be an underestimate ...
Zafira 2 days ago [-]
Personally, I’m just exhausted. They are not obligated to be truthful here, so the entire thing is interpreted based on vibes. If you are an AI booster, this is proof that the demand is there; if you aren’t, you’re baffled at where these numbers are coming from.
reilly3000 2 days ago [-]
I think the level of panic in CFO offices every where about AI cost explosion directionally validates that the spending is real.
ricardobayes 2 days ago [-]
There is no panic, it's misreporting and bad journalism. 2025 AI budgets were based on 2025 AI capabilities, and let's face it, LLMs only got acceptable in around November 2025. So it's natural usage went up and budgets didn't account for that.
aprilthird2021 2 days ago [-]
There's absolutely someone in every big company except the biggest tech companies in the world looking at spend these days because of exactly what you said. The models are good now which means people use them a lot more which means money is flying out the door more than ever before (and the impact on the businesses hasn't shown up yet as you might notice in the earnings of any business that isn't a frontier AI company)
ethbr1 1 days ago [-]
The root problem: At every company, there is always more work that could be done, but there is not always more work that would increase profits.
Existing corporate command and control has optimized for people control, because people cost money and performed work. Control their assignments, and you control costs and what's worked on.
Widespread unmetered AI turns this on its head, because suddenly each employee is directing their own work and the AI spend that comes with it.
F.ex. Bob in accounting may think it's a brilliant idea to rebuild Lotus 1-2-3.
That may help Bob, but 10x'ing Bob's spreadsheet output doesn't change the company's profitability, because it wasn't a limiting factor. It was to Bob, but not to the company's revenue generators.
Increasing AI spend without profitability improvements is a symptom that C2 is failing (or was insufficient to begin with).
Seen through a charitable "CEOs know what the fuck they're doing" lens, the preemptive layoffs are about forcing AI efficiency gains in areas CEOs expect them: instead of allowing those departments' remaining employees to build their own apps, they're forced to deploy AI to cover for their missing 3 team members.
Unfortunately, the layoffs were executed before there were solid results about which departments could benefit from AI use (and without a plan for continuity of institutional knowledge), so... we'll see.
aprilthird2021 1 days ago [-]
I mean, the layoffs we are going through right now are patently due to the high cost of AI buildouts and transitions. Meta is taking on significant debt for the first time in a long time to continue their build out plans as AI components have spiked in price.
I get what you're saying about how every employee becomes a liability when they are let loose with an AI, and totally agree. But I think the layoffs have a much more financial root because they are so widespread across so many companies and even industries (not just tech)
ethbr1 1 days ago [-]
My point about layoffs is that with widespread AI adoption and without department-targeted layoffs, a company risks their AI spend disappearing into the void.
Layoffs + access to AI forces most of that AI use to make up for the layoffs.
Which is a pretty shitty way to burn your employees out... but it has a method.
(Also, I'd say AI infra companies vs everyone else are apples to oranges. This point would be more applicable to a non-AI infra company that's paying for AI use rather than AI infra capital)
solumunus 19 hours ago [-]
I really can’t fathom the Meta strategy at all here.
ethbr1 5 hours ago [-]
The same as Google's: use overwhelming funding drown any burgeoning product category with the potential to threaten core profit center.
Turn a potential existential competition into (a) maybe I win or (b) I commoditize the market and you can't make money at it either.
conartist6 1 days ago [-]
Right but that kinda like just happened. The market hasn't fully reacted to it yet, so it's hard to say that now is the moment when it's safe to draw a linear extrapolation.
jimnotgym 2 days ago [-]
As that sort of person myself, albeit not in a software company right now, my thought process would be this.
1) I have a new extra cost
2) How does that make me more profit, and improved cashflow
I know I'll be bombarded with metrics about productivity, feature completion, bug fixes etc. But someone is going to have to tell me how that equates to more sales. And who is going to do that? I'll be worried that the feature wishlist will just creep up now we can handle more throughput, and yet everyone will be telling me that I don't need to worry about the lack of new customers this quarter, one more new feature and we will catch up in Q4. You can see how that could make one grumpy. Then when the sales don't come, I'll tell the CTO that they need to balance the books, if they want to use expensive tools to make each developer more productive then they will need fewer developers...
...but I don't want that, I want more great features that people will pay for, and faster. But they have to pay for themselves.
trhway 2 days ago [-]
As Anthropic with all its revenue still needs a large outside capital, it suggests that those AI costs those CFOs pay probably don't cover the actual Anthropic's costs.
c0rruptbytes 2 days ago [-]
i would assume investors giving billions of dollars have done their due diligence and not base it off vibes...
vishnugupta 2 days ago [-]
You’d be surprised how much vibe is there in VC investing.
When one is under the grip of irrational exuberance and FOMO due diligence goes out the window.
For example look at Sequoia’s investment into FTX and article about Sam that accompanied.
Nevermark 2 days ago [-]
True, but this is not what might be considered an unexamined company. You can be sure it's under all its investors' microscopes.
If for no other reason than they would rather spend their days procrastinating on everything else, so they can obsess over some of the most interesting numbers they will ever see.
casey2 2 days ago [-]
FTX was examined, investors thought it was impressive that Sam was playing Bronze ELO LOL while in meetings.
GoblinSlayer 2 days ago [-]
Soros too believes that traders are retarded.
c0rruptbytes 1 days ago [-]
yes but Sequoia invested like 200m, we're talking like Google investing 30b...very different scales and bureaucracies
fuzzfactor 1 days ago [-]
Well if you've got billions of dollars just sitting around doing nothing, and many many more billions of dollars that are performing wonderfully, you should be able to afford throwing a few billion away if that's what ends up happening.
sandeepkd 2 days ago [-]
I seriously doubt that anyone is doing much of a due diligence here. They are either counting on others to do it and just follow along OR investing some more money to keep this thing alive enough to go through the IPO.
ARR seems like a interesting concept which hides away the concrete details and is counting on a futuristic possible commitment especially for companies at this early stages. The crucial detail is probably the money being spent, that is what is fueling this sale of equity at this point.
Nevermark 2 days ago [-]
> ARR seems like a interesting concept
Yes, it is a meaningful financial measure of actual progress, exactly where progress is most needed for a new company.
> which hides away the concrete details
A normal metric isn't a magic trick. It's just the number it is.
> The crucial detail is probably the money being spent
Accelerating (not just fast) revenue growth at an astounding rate, with absolute numbers that are enormous for a new company, bonkers for a 2500 employee startup [0], is the crucial "detail".
There are lots of companies with deep pockets and compute, making great AI efforts with teams of smart people, that would love to be doing, but are not doing, what Anthropic is doing and doing well. That might be the second crucial detail.
> A normal metric isn't a magic trick. It's just the number it is.
ARR as a number for steady business or public company is much simpler, ARR for companies at this early stage fueled by the motivation for IPO is more than just a number. Its an attempt to convince the market for a certain outcome without revealing the books
> Accelerating (not just fast) revenue growth at an astounding rate
This acceleration is fueled by somewhat artificial demand (AI mandates by executives across the board, still figuring out the realistic use cases) and subsidized pricing. The expenses matter cause they are the ones which are going to dictate if the business is sustainable or not.
Nevermark 1 days ago [-]
> ARR as a number for steady business or public company is much simpler, ARR for companies at this early stage fueled by the motivation for IPO is more than just a number. Its an attempt to convince the market for a certain outcome without revealing the books
Of course ARR is simpler for less dynamic companies. Tautology.
Of course a PR is a communication to the market. Also a tautology.
You are framing functional numbers and behavior as fraudulent because... a newsy PR isn't a formal disclosure?
Cynicism is fine. However: accusations should reference some evidence, not just distrust.
2 days ago [-]
bnagh 2 days ago [-]
They get in now because they want a high IPO exit. They care about the perception of the numbers, not about reality.
If you think investment banks will catch them during the IPO: They have launched many overpriced IPOs in 1999/2000 and also want a high IPO price.
sberens 2 days ago [-]
Investors are usually prevented from selling until 180 days after IPO
throwaway85825 2 days ago [-]
The rules can be changed, they did for spacex.
stingraycharles 2 days ago [-]
Yeah but SpaceX has undergone some “changes” in the past few months that make it a dumpster fire, rather than Anthropic’s explosive growth. Also, I don’t think the CEOs of both companies operate in the same way.
throwaway85825 2 days ago [-]
I dont see why changes at spacex would necessitate any changes at nasdaq.
0Ggr3g 2 days ago [-]
From what I heard, nasdaq changed the rules so that Spacex can be added sooner to the index. Then pension will essentially buy SpaceX (via index), bringing the necessary liquidity for SpaceX exec to exit (very fast thanks to SpaceX rule change)
aussiegreenie 23 hours ago [-]
The US capital markets are closer to Putin's Russia than to free markets.
Historically, listing rules asked: "How much money did you make last year, and do you fit our standard corporate governance box?"
Today, NASDAQ's rules ask: "Do you have the massive market capitalisation, sufficient institutional public float, and transparent liquidity to ensure fair and orderly trading?"
Here are the obvious ones:
1. Free float - Every company that intended to list was required to have at least 20% free float.
2. Index weights were based on the free float.
3. Time before inclusion into the Indices (min 12 months) now 15 trading days
4. Lockup period - minimum 12 months up to 24 months - not 180 days.
I wish people would understand that if America had a functioning criminal justice system, no one would have heard Elon Musk nor Donald Trump.
coldtea 2 days ago [-]
They have a dozen tricks to get around that... e.g.
"The passive funds holding trillions of dollars of 401(k)s and other investments are rushing to change their rules as the IPOs of SpaceX, OpenAI and Anthropic draw closer."
Those index providers are the same interest class with VCs. With such moves they inflate demand post-IPO (hoping it holds for 180+ days), but also allows them to lure buyers in private secondary market and offload that shit pre-IPO.
If the investors are VCs, they can sell their holdings to a syndicate of underwriter banks in advance of the IPO, and let the banks shoulder the risk of finding a bigger fool in the secondary markets.
lurk2 2 days ago [-]
WeWork, Theranos, FTX
ajyoon 2 days ago [-]
Investors made bad decisions all the time. The proof is in revenue.
vdfs 2 days ago [-]
That's basically how 2008 happened
dogwalker5000 2 days ago [-]
After cases like Madoff, Theranos, and FTX … I don’t know about that.
2 days ago [-]
iancmceachern 1 days ago [-]
I used to think this too. Now I know different.
AlexandrB 2 days ago [-]
There are plenty of examples of investors going off of vibes: Theranos, Juicero, WeWork. Though Anthropic would be a particularly egregious example if it does end up failing.
coldtea 2 days ago [-]
Market crashes like the dotcom bust, and countless companies stock rising to high heavens to crash a few months after IPO to shit say otherwise. VALinux was a poster child for investment... lol
Not to mention even a total shitshow from an obvious crackpot like Theranos got $1.2 billion total funding, and a 9B valuation. Or FTX.
duped 2 days ago [-]
Investors would never give billions of dollars to liars
sandeepkd 2 days ago [-]
Idk if anyone cares about lying at this point anymore, this is just placing bets on the AI, and its a lot easier to bet when its someone else's money.
mgfist 1 days ago [-]
I started a new job recently after 9 months off. Last Sept I was coding almost entirely by hand, using AI for q/a, debugging and such.
In my new job I haven't written a single line by hand. I now almost entirely work in claude code / codex and in github PRs. Occasionally I open vscode to read code, but very rarely. My company probably spends ~$100 a day for my token use. I'm not even going crazy with parallelization or subagents and such.
I 100% believe the demand growth based on my personal experience.
Traubenfuchs 1 days ago [-]
Talking to dev friends, dev colleagues and qa colleagues in a space where everyone is doing high volume CRUD and cloud slop with many microservices and constantly evolving business requirements: Every single one is primarily coding with claude, a few with cursor. Handwriting code is dead.
semiquaver 2 days ago [-]
> They are not obligated to be truthful here
What are you talking about? It’s securities fraud for a CEO to lie about the financials of a company regardless of whether it’s public or private.
33vss 2 days ago [-]
more than likely what he meant is their metrics may be non-GAAP.
rchaud 2 days ago [-]
And the punishment for securities fraud is a slap on the wrist, as Tesla proved with its "going private at $420/share, funding secured" tweet. Who cares about a $40m fine from an increasingly toothless regulator when your investors have already fronted $1.5 billion to settle copyright infringement claims?
A lot of companies are mandating token use, there's certainly a smell going on here but I have no idea why all the CEO's got onboard unless they all get some of that stock. AI psychosis is very real.
simonw 2 days ago [-]
I think lying about their numbers now is risky if they're planning on an IPO this year - the real figures will come out in the S-1, and investors don't like untrustworthy companies.
> (Also in Axios today is an anonymously sourced note that "An AI consultant tells Axios one of their clients recently spent half a billion dollars in a single month after failing to put usage limits on Claude licenses for employees" - times that by 12 and you get an extra $6 billion in annualized run-rate!)
This one client, then, is 12.8% of Anthropic's run-rate revenue? That does not exactly fill me with confidence that it's a meaningful number. Doesn't this suggest run-rate revenue will fall off a cliff if Anthropic customers start applying cost controls?
vmbm 2 days ago [-]
Yeah, I think it is pretty obvious at this point that users have gotten over their skis a bit on the fixed rate plans. I suspect a lot of folks were playing around with agentic workflows on a $10-200/month plan and then started implementing them at their companies under enterprise accounts not realizing that API based billing would result in easily 10-100x the costs. Running hundreds of agents 24/7 is all fun and games when you can do it for beer money. Not as fun when it is super yacht money.
As an anecdote, Github is changing their copilot plan to usage based billing next month. They released a tool that allows their users to estimate what their bills will look like under the new plan based on their past usage. There are some screenshots online from users showing their plans will go from $40/month to $3-5k/month. I imagine this is happening everywhere. These tools absolutely can do more than they were capable of just six months ago. But if the true costs are as high as it appears, folks are going to be much more judicious with how they use them moving forward.
256BitChris 2 days ago [-]
GitHub doesn't have access to its own models so it has to pay the prices of the model it uses.
People confuse price with cost all the time. The price of Opus has dropped from $75/1M output tokens to $25. That's the price. The cost is much lower and according to Dario, about a month ago, they had about a 73% margin.
I don't understand how anyone would use GitHub copilot...it's basically running a custom harness and using close to API pricing for Opus. This is why Microsoft is cooked in this game.
But yeah, I don't understand why people switch from subscription to API prices for Claude. They're way higher, but again that's price and not necessarily the cost to Anthropic to serve.
vmbm 1 days ago [-]
For the longest time Copilot was the best deal in town. For $10 a month you would get ~1300 requests. A single prompt was counted as a request, and it didn't matter how many tokens were used or how many loops the agent did. It was a spectacular deal. Of course, now that they are moving to API based billing the plan is a not really a deal at all. The monthly plan is essentially pre-paying for API credits, which are use it or lose it and they are not discounted in any way.
And FYI, most enterprise accounts were forced to switch to a hybrid monthly seat license plus API based usage earlier this year. So that is why we are probably seeing so much alarm over Ai bills at the enterprise level. Companies went whole hog on agentic workflows not fully appreciating the costs structures of their new plans. Didn't help that pretty much every VC and board was probably breathing down their neck that if they didn't jump on the AI bandwagon they would get left behind.
riffraff 2 days ago [-]
> according to Dario, about a month ago, they had about a 73% margin.
Do you have a source for this? I missed this and it does not match my impression.
SatvikBeri 2 days ago [-]
Not OP, but Dario said on Dwarkesh's podcast 3 months ago that their gross margins are "significantly higher than 50%"
Incipient 2 days ago [-]
I'm really annoyed at github for only allowing export of a single month of usage.
There were a lot of people (here included) that were absolutely abusing github - getting Opus to generate its own subagents for days on end with a single premium request. If THAT'S $3k/mo, I'm honestly not worried.
I was just asking sonnet, 5.4, opus, in single agent session to fix a problem for 20 minutes...if THAT'S $3k/mo, then AI is truly cooked.
Yeah, I don't doubt that there was some abuse going on. But keep in mind there are plenty of reports out there that companies were setting up tokenmaxing leaderboards essentially encouraging this type of behavior from their employees. So I think the broader point here is that the insane growth that Anthropic has experienced over the last six months might be a temporary blip due to companies scrambling to take advantage of these more powerful agentic capabilities while not fully understanding the resulting costs.
I think for your use case, the most likely outcome is that you are going to need to be on a $100-200 month plan if you want access to the cutting edge models. But on the other end of the spectrum, you could probably get closer to $10-20 month using Chinese models. My usage was closer to yours, with the occasional tokenmaxxing sprint just to experiment a bit and test out the limits of the plan. I am not quite sure what I will be doing next month once it moves to API billing but I suspect I will move to openrouter and one of the open source CLI harnesses.
simonw 2 days ago [-]
Maybe? We don't know if that Axios anonymous note is accurate, and we don't know how Anthropic are actually calculating annualized run-rate - I'm just guessing that they might be taking that monthly number x12.
If they really did count that one customer as $6B it means they've gone from $30B in April to a mere $41B in May.
3ff 2 days ago [-]
lmao got him.
im really not sure why he keeps parroting on about this. its all irrelevant frankly. companies play games. non-gaap revenue recognition, adjusted operating income.... boo-ya.
wait for the somewhat official doc's to come out, then its worth talking about.
shimman 2 days ago [-]
Yeah, until they release their S1 all numbers are highly suspect.
2 days ago [-]
xyzzyz 2 days ago [-]
How are they “not obligated to be truthful”? Lying to investors is literally a crime.
e9 2 days ago [-]
Private companies are not obligated to report GAAP numbers which means they can creatively twist numbers while still be truthful until they go public
mullingitover 2 days ago [-]
In 2026 what really even is financial crime, other than failing to grease the right palms?
Zafira 2 days ago [-]
You need to think about this more creatively.
What we are talking about is entirely based on this one term in the announcement: run-rate revenue. This is a meaningless term just like how “clinically proven” for vitamins is a way for the companies to use weasel words to imply something without truly being able to back it up, but also not actually lying. There is no legal definition of “clinically proven” so, what exactly would you sue them for? The same thing is true for run-rate revenue. They can cherry pick numbers to use to generate the run-rate revenue value and they are not lying, but this isn’t exactly honest either. We have no transparency and run-rate revenue is not an accounting term.
This goes back to my entire point, this is a vibe. This announcement does not provide much in the way of substance, so a reader will take it to say whatever they want.
shimman 2 days ago [-]
You need to give these people actual examples because rarely can they see the line between two dots. A great example of cooking the books is to do circular investments, adopt a bunch of users that are likely to churn, or to take your single "best" week of revenue then multiple it by 52. All things to over inflate the perceived value of an IPO.
IDK if these tricks would work anymore but then again fraud is legal now so who knows.
foobiekr 2 days ago [-]
Are you familiar with Elon Musk's endless, repeated and obvious lies for the last decade?
cookiengineer 2 days ago [-]
> How are they “not obligated to be truthful”? Lying to investors is literally a crime.
In the US?
Hahaha, that was a good one, buddy.
crowcroft 2 days ago [-]
I mean, they're not a public company, but there are still laws that exist...
cyanydeez 2 days ago [-]
irresponsible growth
tailscaler2026 2 days ago [-]
not really. just a capacity planning issue on that side. much harder challenge is continuing to train better models.
cyanydeez 2 days ago [-]
having a trillion dollars that need 5x payment un 5 years is insane promise by anything but government.
like when you owe a bank a billion dollars, thats the banks problem.
mNovak 2 days ago [-]
The wild thing to me, is that they're serving $47B run rate worth of requests on maybe 2-3 GW of compute currently [1], of which only a fraction goes to inference, vs R&D and training. Obviously there have been complaints on token limits and such so they're stretched a bit thin, but nonetheless.
Hard to imagine what a world with 100GW of compute looks like.
It gets better; most of their incoming requests don't actually require a frontier model to handle. There's a huge potential for future optimization in this space. Anthropic, OpenAI, Google and a few other companies are going to be well positioned to scale in the few years. A 65$ billion round to finance operations over the next few years isn't that controversial if you look at the growth and profit potential.
I think token counts and GW are a gross over simplification here. Not all tokens are the same in the amount of GPU time they consume or the size of the GPUs they require or the amount of energy they consume. There's a huge optimization potential here once these companies get serious about consolidating the business they have and executing much more efficiently. Given enough time, these companies can heavily optimize their operations. Short term growth and not slamming the brakes on that is their primary concern.
0xDEAFBEAD 2 days ago [-]
Where's the moat though? What prevents a race to the bottom with competing AI providers, everyone trying to undercut one another?
epolanski 2 days ago [-]
I'm also thinking the same.
I have been trying Claude Code with DeepSeek 4 apis, and the experience is barely different. In fact the margin of error is so small that harness and prompting account for the most impact in output quality.
But, here's the catch: I spend barely more than a handful of dollars per day of regular usage. In fact DS4 via api is cheaper than Claude 100$ subscription.
I really think that very soon many will start realizing that the alternatives are extremely close in performance but dramatically different in pricing.
port11 1 days ago [-]
Claude includes or at least promises ZDR in some situations, whereas DeepSeek is explicitly using output to train models. The subsidising might be done with your data.
dzonga 1 days ago [-]
ARR vs realized revenue are two different things - in software we like to conflate those 2.
funny enough - those 2 might not meet. then what happens ?
Instead of ARR they should report actual revenue for once.
namenotrequired 2 days ago [-]
Nitpick: earlier this month ≠ in early May. “Earlier this month” includes yesterday
reactordev 2 days ago [-]
Shhh, you’ll make the numbers not make sense…
mountainriver 2 days ago [-]
Huh I thought AI didn’t have any use cases /s
ratakh 2 days ago [-]
[flagged]
Game_Ender 2 days ago [-]
It’s becoming like the iPhone, once the software has access to “Claude”, everyone in the org wants it. Finance wants it for excel, marketing and design for image generation, compliance for working with documents. Sure it’s not software engineering rates, but it increases the user base beyond software developers.
mikeyouse 2 days ago [-]
Rings true in my org -- finance is a big consumer because all the software shops who do integration work with CRMs and ERPs are garbage and take forever to implement changes and via some clever prompting, you can obviate most of that work.
limflick 1 days ago [-]
I'm curious, do you have an example?
mikeyouse 20 hours ago [-]
One regular workflow is a reconciliation we do for events that we put on — a number of costs that are expensed, a number of costs that are prepaid until the event happens, individual registration revenue that is recognized immediately and then the corporate sponsorships that are often paid in advance but their recognition is deferred until the event happens. Previously since that involved both balance sheet, income statement and CRM reporting, we relied on an integration vendor to write custom scripts to bring all the info (poorly) into our ERP. Since then, we’ve found a tool leveraging LLMs to ‘join’ those various sources and our events people generally described the report they wanted with a template in excel and it readily created that report with “export to sheets or excel” functionality.
A report that previously took ~4 hours per month for a very expensive resource now takes 30s to validate and can be run completely ad hoc by the events managers.
HDBaseT 2 days ago [-]
Tangential, but I'm partially surprised Anthropic hasn't released an image model, even something basic to complement their offering.
hgoel 2 days ago [-]
It doesn't really feel like an image model fits with the "theme" of Anthropic's products.
I see Claude as much more of a productivity tool than say, ChatGPT, so if they were to release an image model, I would expect it to be useful for presentations etc. But, then the images need to be more like infographics, which are difficult for an image model to get right. On the other hand, they are much easier for a strong coding model to implement, while also allowing users to make controlled edits.
simonw 2 days ago [-]
It's interesting. Until quite recently image models felt like toys, and Anthropic opting out seemed like a good decision to me. I feel like the "trained on stolen data" thing feels a lot more real with image models, since they so clearly compete with the artists whose work was used - so it felt aligned with Anthropic's image as the "more ethical" of the labs.
But then Nano Banana happened, and ChatGPT Images 2.0, and now image models aren't a toy any more - you can get real, useful work done with them.
Which is a problem for Anthropic, because companies that are buying a suite of AI tools for their employees may well value the ability to create usable posters, leaflets, and most importantly presentation slides.
Anthropic don't have anything to offer there, where Google and OpenAI have two of the best models anywhere.
hgoel 2 days ago [-]
I've seen plenty of people at my job that have tried to spice up their research seminar slides with some chatGPT-generated images, the only time they don't suck is when someone uses the images as a joke. There's always something blatantly off, such that the output is only usable for informal seminars. On the other hand, the approach of having an LLM produce a python script, or write out a PPTX directly with placeholders and "vector" flowcharts etc, is now a very common approach for us.
michimagdesign 2 days ago [-]
Focusing on improving the inline visualization feature and Claude design, which were both recently introduced, will likely make more money and be more helpful than building an image model and competing with the very good ones Google and OpenAI have.
conception 2 days ago [-]
400/mo is about what we are seeing a full time dev use on average, coincidentally.
SOLAR_FIELDS 2 days ago [-]
Indeed if I look at the usage of my cohorts the power users typically spend almost exactly this much
BoiledCabbage 2 days ago [-]
Using API or subscription?
simonw 2 days ago [-]
I spent $200/month on subscriptions last month ($100 each to Anthropic and OpenAI) and the API cost version of my token spend was $2,100 - and if I'd been on the "Enterprise" plans my company would have had to pay the full price, not the subscription discount: https://simonwillison.net/2026/May/27/product-market-fit/#en...
I agree that $400/month is a HUGE amount, but there might be a path to that.
theteapot 2 days ago [-]
I spend $0/month.
hombre_fatal 2 days ago [-]
I spend $200 for Claude, $100 for ChatGPT, and I still hit hourly and weekly limits mostly working on a couple projects with no "claude -p" style automation.
Seems like a lot of money until I consider how much time it would have cost me to get this far in any of them a few years ago. So long as the thing you're doing is worth doing (business, pleasure, curiosity, utility) then it's a bargain.
keepupnow 2 days ago [-]
I do hope they improve in time. Weekly limits are criminal.
hombre_fatal 2 days ago [-]
It's mutually beneficial: the load is smoothed for them, and it ensures I won't blow through a monthly limit.
hakimg 2 days ago [-]
There's about 20 million software engineers in the west and almost 50 million globally. Now add to the equation non technical paying subscribers who migrated from ChatGPT and enterprise API spend from internal AI applications and the figure seems pretty reasonable.
jeremyjh 2 days ago [-]
Why is the number of software engineers a limiting factor? I think what matters is the number of people who want custom software. Or cheap customer service. Or cheap financial analysis. Or cheap...
ericmay 2 days ago [-]
> Then everyone of these would need to spend $400 per month on tokens.
That's not that much money in the grand scheme of things, especially for software engineers - and you also have to include folks who pay for Claude or other AI tools on their own or for their own mundane purposes. More startups, &c. I don't use these tools much for work but I pay for a subscription and find it very valuable for my own personal uses.
> I don't know how much killing girls in Minab pays, but it looks like there is a lot of fake revenue reported here.
Well, the Iranians have us beat on that. They just use assault rifles to mow down 30,000+ of their own people: little girls, medium girls, and big girls too and we use a fancy bomb on accident to blow up a school they are launching missiles from to kill even more little girls.
newaccountman2 2 days ago [-]
> Then everyone of these would need to spend $400 per month on tokens.
ez
(not sarcastic)
himata4113 2 days ago [-]
that;s basically nothing, companies are spending upwards of $4,000 to $30,000 PER employee. Is it worth that much? probably not.
zarzavat 2 days ago [-]
Is that truly a recurring expense though or just the cost of doing things now? It remains to be seen whether any of this is sustainable.
himata4113 2 days ago [-]
afaik it's "hey, see how far you can push it if we don't limit you and how much you can achieve and we will compare against baseline"
VirusNewbie 2 days ago [-]
that seems pretty reasonable. If it makes software engineers 10% more productive that's a bargain.
photonair 2 days ago [-]
If these are fake revenues, I would think that the sophisticated investors would see through that? Or everyone is just in some ponzi scheme raising valuation and letting the last investors hold the bag or eventually pass it on to retail investors buying at the peak?
techblueberry 2 days ago [-]
Anthropic is private. They talk all
The time about their numbers being non-gaap run-rate revenues mean extrapolating a month out to a year. All they have to do is book a big deal in May and multiply by 12.
It’s probably still impressive growth, but probably not as impressive as it looks.
> An AI consultant tells Axios one of their clients recently spent half a billion dollars in a single month after failing to put usage limits on Claude licenses for employees
Times that by 12 to annualize it and you get to add an extra $6 billion to the number!
Escafati 2 days ago [-]
[dead]
fsuts 2 days ago [-]
Yes
Everyone is in the same bag, get it IPO’d and then let the retail investors and tracker funds worry long term
waterTanuki 2 days ago [-]
You're not factoring in enterprise contracts which will probably be more per dev per month
bnagh 2 days ago [-]
That is more than offset by the fact that most software devs have no accounts, half have maybe OpenAI or Deepseek or Gemini accounts.
pertymcpert 2 days ago [-]
$400 a month is a lot to you?
bnagh 2 days ago [-]
For something of the equivalent of surprise Pokemon cards, anything above $0 is too much.
topherPedersen 2 days ago [-]
Anthropic has a great product, but what's going on in the stock market is astonishing. Companies waiting to be valued at a trillion dollars before going public? (I'm writing this comment with the assumption that they will go public soon and the valuation will be higher than this $965 billion dollar private valuation) The stock market used to be a place for companies to raise money from investors. But that isn't what it is anymore, it's a dumping ground. Venture capitalists & private investors are sucking all of the possible growth and future upside from these companies and then dumping them on retail investors when there's nothing left. There is no growth or upside left by the time these companies go public. If you invest in these IPOs you are buying the absolute peak with all potential future profits baked into the price, with nowhere left to go but down.
taude 2 days ago [-]
Yeup, no shortage of tech IPOs over the past five years that are now valued at like 5% of what they were after being dumped onto the market: ZoomInfo, Bumble, Gemini
And many more that are 50% of what they were: Snowflake, Coinbase
And many more that went back to private companies and then were sold off: Carbon Black, etc...
I'm actually too lazy to go list out all of them.
But employees, beware, of those gnarly lockup periods post IPO where all the better classed options than yours get to exit.
bravoetch 2 days ago [-]
Coinbase wasn't an IPO, they didn't create any new shares to sell as part of going public. They did a DPO, Direct Public Offering where they listed the existing private shares publicly and allowed most shareholders to sell immediately from day one. It was a great way to make the founders rich, VC to cash out their initial investment, and... well mostly just that.
tnolet 2 days ago [-]
Hashicorp comes to mind. From ~$90 at IPO, cratered to $25 and then taken private again by IBM
jLaForest 2 days ago [-]
Maybe snowflake was a bad example considering that stock is up 36 percent today haha
taude 1 days ago [-]
yeah, wild!
pas 2 days ago [-]
... still, "on average" IPOs tend to make money, no? that's why people (fight to be able) to buy them.
this gives a nice comfy exit to many late-stage investors, etc.
and, of course, it's hard to say that it's great that these companies are mere shadows of themselves post-IPO, but also it's impossible to non-misleadingly assess each IPO as if they were in a vacuum.
obviously Coinbase is/was a stupid venture, but at the same time it was a pretty good bet at the time. and the same stands for a lot of these.
executivevice 2 days ago [-]
> Approximately 56% to 60% of U.S. initial public offerings (IPOs) lose absolute value over a five-year period. Historically, the median IPO stock has lost roughly 41% of its value five years after its first day of trading.
Ipos are somewhat notoriously risky investments.
DanielHB 2 days ago [-]
I remember seeing a video about this subject, since large IPOs are automatically included in index funds it is kinda of a way to extract value from passive investors. Insiders cash out before it hits the indexes, index crashes by a fraction for a %, all pensions in the country (and many overseas) pay for it.
But with OpenAI and SpaceX IPOing roughly at the same time it will likely be more than fraction of a % in this/next year.
pas 1 days ago [-]
but also, getting IPOs included captures the upside.
of course, public markets nowadays are definitely paying a pretty serious "agent-principal premium". (since public exits are usually very good for the C-suite and for all those vested stocks.)
so yeah, it seems it would make sense to buy the post-IPO dip, but then you would need to have some kind of formula for that, and ... that seems ripe for gaming by speculators ... so all in all, it's just more efficient to do what the rule of the index says. (and of course there's already speculation at the discontinuity.)
pas 1 days ago [-]
sure, but does that risk have good returns to go along? if IPOs are known to be very bad bets why do institutions (supposedly savvy professional investors) participate?
executivevice 1 days ago [-]
Because they (should) have sophisticated risk models that account for the long tail. If even a few ipos become Google or Facebook, the risk is worth it. But for average retail investors ipo participation will be bag holding exercise. That said betting your conviction is one of the only ways to beat the market, even if it comes with additional risk (emotional+intellectual attachments). If you really believe in ai or space exploration, the upcoming ipos represent an opportunity to bet on your beliefs and predictive capability
pas 19 hours ago [-]
you mean that if average Retail Ronnie directly buys the new hot stock at IPO versus getting exposure to it through whatever ETF they have?
yes, directly buying a stock at IPO sounds really strange for me. (because either you know it's undervalued, but then it's insider trading. if not, then why compete with irrational fanatics?)
2 days ago [-]
jeberle 2 days ago [-]
Is there any sense in buying (out-of-the money) put options, to make sure you can taste some of that sweet lucre that your overlords are getting?
peab 2 days ago [-]
If you like gambling, sure. Market can stay irrational longer than you can stay solvent, or something like that
2 days ago [-]
bugufu8f83 2 days ago [-]
You're normally not allowed to trade derivatives during the IPO lockup period. Otherwise it would defeat the whole point.
stingraycharles 2 days ago [-]
I think index funds are a big reason for this change, as many of these stocks are now guaranteed to be bought by a huge chunk of the market, making it much easier for them to become bag holders.
Lots of professional investors are passing on the SpaceX IPO for example, which is why they had to increase the share of the retail investors.
dataviz1000 2 days ago [-]
> dumping them on retail investors
They are dumping them on your 401k -- especially SpaceX.
w10-1 2 days ago [-]
Often only minimal shares are floated on the public market - 5-10% now is not unusual. Also, founders keep priority shares to keep the company.
So IPO is not particularly a liquidity event for investors as much as a valuation/pricing event. Indeed, the tech IPO's that have done the worst were the ones where shareholders wanted liquidity.
Clearly none of the multi-trillion dollar companies could find a buyer now if they really needed to sell themselves, so they're not really "worth" that much. (Nor are their founders, who can't sell their shares without tanking the stock.)
So these stocks are more like derivatives: a way to bet on the future where betting volume is huge relative to the underlying asset.
tailscaler2026 2 days ago [-]
dumping ground because they spent the last 20 years convincing everyone with a 401k it belonged in index funds.
index funds will be the cause of the next catastrophic collapse
2 days ago [-]
hibgymnb 2 days ago [-]
elaborate please
BoneShard 2 days ago [-]
easy:
1. Shit goes into S&P 500 (pump phase)
2. Shit goes to 0. Your 401(k) invested into S&P 500 takes a dive (dump phase)
3. Retail holding bags (full of shit) phase.
Case study: Tesla, with a P/E ratio in the hundreds along with declining sales and TAM, is a part of the S&P 500 and, consequently, of many people's 401(k)s.
stocknoob 2 days ago [-]
Index funds are market cap weighted. As companies fail, as they always do (median lifespan of S&P companies is about 15 years), you have less of it.
0xDEAFBEAD 2 days ago [-]
Buy high, sell low in other words?
jasongi 1 days ago [-]
Well... in normal times they would be entering at the bottom of the index due to the company beginning to grow, the purchase of which is being funded by a firm exiting the index due to shrinking, so assuming you have bought and held units in the fund, most of the time an index fund is buying low and selling low.
And then when you sell your units, hopefully in aggregate the index is worth more than it was when you entered...
BoneShard 1 days ago [-]
"That's the point: just pump the shit and offload it to retail. Yeah, it's less than 1% of the S&P 500, so who would even notice?"
refurb 1 days ago [-]
Tesla is 1.75% of Vanguard SP500 index. It could go to zero and most investors wouldn’t notice.
BoneShard 1 days ago [-]
Isn't it a nice bussiness, pulling 2% from sp500, I wish I could join.
onlyrealcuzzo 2 days ago [-]
> But that isn't what it is anymore, it's a dumping ground.
We got "dumped" Google and Facebook, so... Those probably made up for all the other "dumps".
We also got "dumped" TSLA, which is meme-ing in the trillions at the moment.
You can short Anthropic at IPO if you want...
halamadrid 2 days ago [-]
Looking back it feels like GOOG, FB, TSLA etc. all went IPO at reasonable valuations. Retail & public investors did benefit long term and continuing to get higher valuations in public is not a small feat compared to a VC valuation.
A trillion dollar valuation seemed so hard back in the day and now there are so many companies in that list. What's the next level?
Is this just signs that $ is no longer the inflating at the same rate over time and its the realistic inflation that is reflecting in the stock market?
Prices of all goods surely has to follow to make up for the revenue needed to sustain these valuations and also the salaries to sustain the prices.
Unfortunately, those who are not in the loop is not going to have a good time.
senko 2 days ago [-]
> Looking back it feels like GOOG, FB, TSLA etc. all went IPO at reasonable valuations
Yeah, looking back. At the time, I distinctly remember people were going batshit over the insane FB valuation. It wasn't at all obvious it was justified.
Hindsight is 20/20.
ml_basics 2 days ago [-]
maybe we will look back and also think that the current slate of IPOs were reasonable valuations?
lumost 2 days ago [-]
The critical view is that these IPOs are bumping up against physics. How many trillion dollar companies can the economy support? The US GDP is roughly 32 Trillion. A company with 100 billion dollars in revenue and 10x annual growth would be expected to increase the size of the economy by 3.2% in its first year, and about 30% in it's second year post IPO existing.
While we could claim that such a company can grow by consuming a larger share of the GDP ... this would not bode well for future political stability, and nationalization would be a major topic.
So your left with a fast take off scenario, a job apocalypse, or a massively reduced growth rate.
If Anthropic has anything line that level of success, it will also be at the expense of huge unemployment, because Claude is really competing with knowledge work
HDThoreaun 2 days ago [-]
I remember a ton of people talking about how FB ipo'd at way too high a valuation. Just checked and it opened at $38 and was $19 3 months later
DanielHB 2 days ago [-]
I think this is usually the case for most IPOs, insiders cash out, index-funds buy in. Basically it is a one-time technique for investors to extract money from pensions.
If you try to configure the index-fund to avoid this problem it is not longer passively managed as each new stock needs to be evaluated in a (at least) semi-subjective manner.
epolanski 2 days ago [-]
Facebook has been reasonable only in hindsight.
IPO was in the 50B $ valuation range, and at the time, there wasn't any hint it made any financial sense.
Of course, hindsight is hindsight, but for every Facebook there's been countless IPOs of tech companies shrinking their IPO valuations by 90% in the following years.
NewJazz 2 days ago [-]
Google IPO 20 years ago, Tesla 15 years ago, facebook almost 15 years ago.
Situations change.
icedchai 2 days ago [-]
"This time it's different", right?
onlyrealcuzzo 2 days ago [-]
When did they change? 3 years and 4 months ago? 1 year ago? 8 years?
Because when Facebook IPO'd everyone was saying the stock market was a dumping ground...
Same with Google...
Same with Pets.com and WebVan...
pixelatedindex 2 days ago [-]
Sometimes when things change in an insidious manner, it’s hard to point out _when_ the decline started.
Npovview 2 days ago [-]
Few Pheonix(s) rise from the ashes of many Unicorns.
signatoremo 2 days ago [-]
Maybe but can you elaborate what the changes are?
2 days ago [-]
surgical_fire 2 days ago [-]
Or he can just steer clear of the eventual Anthropic stock. Shorting is not the only strategy available to avoid losing money.
But you, of course, can buy on their IPO. They need every bagholder they can get :)
mcoliver 2 days ago [-]
Bingo. The bigger story is that the float is not there so the companies are "public" in that they sell a small number of shares at IPO to get crazy market caps that then force the ETFs to buy stimulating demand. It's genius and infuriating at the same time.
DanielHB 2 days ago [-]
It is a one-time technique for private investors to extract money from pensions.
I moved all my money outside US index and global index funds because of SpaceX and OpenAI. At least until these IPOs have passed I will not move any money back. The sheer size of these IPOs might trigger a market crash.
eudamoniac 1 days ago [-]
Shouldn't you have waited to move your money after the IPO but before the holding period for insiders expires?
DanielHB 1 days ago [-]
I moved more than a year ago, the US stock market is basically just gambling by now.
surgical_fire 2 days ago [-]
If they could have gone public, they probably would have. I hope they do, their S1 might be good meme material.
Companies that reached a level of maturity where going public make sense don't keep doing funding rounds to cover the rate at which they bleed money.
qeternity 2 days ago [-]
> Venture capitalists & private investors are sucking all of the possible growth and future upside from these companies and then dumping them on retail investors when there's nothing left.
A lot of the money that is deployed by VCs comes from pension funds and asset managers that ultimately manage money for the average Joe.
andriy_koval 2 days ago [-]
Is there any evidence of what is the share/volume of such assets involved?
jordemort 2 days ago [-]
I’m sorry, H? How much farther down the alphabet do they plan to go before anyone gets any of that money back?
jedberg 2 days ago [-]
Databricks raised an L last year.
jordemort 1 days ago [-]
sounds like an L to me!
crossroadsguy 1 days ago [-]
There's Greek alphabet.. etc.
senordevnyc 2 days ago [-]
God, I wish I was waiting to get some of my money back from an earlier round in Anthropic.
jordemort 2 days ago [-]
The person who put the first log on the bonfire gets exactly the same amount of wood back as the person that put the last
Lionga 2 days ago [-]
Boys we got more subsidy for Claude Code Plans! Let the VC financed spending of 1000$ of datacenter cost for 200$ sales price continue!
itsyonas 2 days ago [-]
$1,000? According to ccusage, I used around 3,200 USD worth of API credits last month, but I'm on a plan that only costs around 100 USD per month, and I'm not even a heavy user. At the end of each week, I have typically used about a third of my weekly Claude limit. So either their APIs are heavily overpriced, which seems rather unlikely, or they subsidise subscriptions to the tune of 100x or even more.
Well, I will enjoy it while it lasts.
HDThoreaun 1 days ago [-]
Why do you think the API being overpriced is unlikely? Seems pretty likely to me that sub sells at cost and api is the massive markup they force on enterprises.
great point. i haven't gotten over losing the Uber subsidy!
2 days ago [-]
eutropia 2 days ago [-]
What is run-rate revenue and how is it different than revenue (classic)?
jmalicki 2 days ago [-]
Revenue is an integral over the past year or past quarter.
Run-rate is taking a recent measurement and multiplying it out so it will span a year. Basically, it assumes they keep all of their current contracts, and don't gain any new ones.
Forward-looking revenue is an estimate of what the integral will be from now to a year from now.
For a growing company, run rate is between past revenue and estimates of future revenue.
Forward-looking revenue estimates are often made up from whole cloth so are highly untrustworthy. But a run-rate is saying "we've already been making this much money, we just need to maintain where we are and that's how much we'll bring in". Backward-looking revenue for something like Anthropic is meaningless because almost all of their customer base is recently acquired - they're growing like crazy, not 20% per year.
mrandish 2 days ago [-]
Sometimes it can be year-to-date extrapolated to full year (which might be calendar or fiscal). The tricky part is the timing of the cutoff period can yield true but misleading numbers, especially in the case of seasonal sales, a major order or just prior to some revenue's recognition period ending.
So, it's saying something but without more details it's also vague and always partially forward looking. I prefer the TTM metric (Trailing Twelve Month revenue).
parliament32 2 days ago [-]
Is it qualified at all? Or can you measure anything and multiply it through to get whatever number you want at the end?
Along the lines of "in the 4.75 hours starting at 02:35 yesterday we collected invoices worth $X so our run-rate revenue is now $Y"?
jmalicki 2 days ago [-]
I'm not an accountant, but afaik run rate is not a GAAP recognized metric. Presumably investors who care want it to be more precisely defined. In practice usually I've seen it be extrapolating the previous month. E.g. if you have per month revenue, that's what you want to extrapolate - I've only really seen run rate with SaaS where you have recurring revenue.
Occasionally it can be a snapshot if you've just completed a big contract - but it's what you expect to get per month if you're not growing or shrinking for the typical SaaS that charges per month (and assuming yearly pre-paid contracts renew etc.)
sophiebits 2 days ago [-]
Collecting invoices is cash accounting, whereas revenue is realized only over the length of the contract and doesn’t care when the customer pays. (Of course sometimes you have a short-term contract including for professional services and such, but not to the point where a single day would likely be particularly inflated.)
jmalicki 2 days ago [-]
For Anthropic in particular, that we're talking about, API token costs are revenue that is earned in real-time not on a contract - so hypothetically a giant spike in token use on an API use contract could spike revenue. But I don't think most expect that to randomly fluctuate enough to be material.
sophiebits 2 days ago [-]
People building applications that use AI do have longer contracts. Not sure how the split is vs Claude Code etc.
jmalicki 1 days ago [-]
Sure, but the moment they use the token the revenue is earned.
anomaly_ 2 days ago [-]
Are you trying to be deliberately obtuse? Obviously, you can fudge the number with assumptions around churn rates/etc, but of course an investor would want a view of the rough 12m state of the business.
parliament32 1 days ago [-]
...which is why we have GAAP-recognized metrics, right? To prevent fudge-ability? And those metrics.. they're deliberately not publishing? Makes you think.
jmalicki 7 hours ago [-]
GAAP is basically a standards body to recognize practices.
When there are interesting stories that can't be told with GAAP metrics, accountants derive new metrics. Just because they haven't gone through the standardization process yet doesn't mean they're bullshit - investors in Anthropic can hire auditors to ensure the Anthropic metrics are still meaningful. There are a very small number of deep pocketed investors in Anthropic - they're not a public company like Enron trying to sell to the WSB crowd, or like 2007 CDOs being sold to dentists.
And run rate has been a widely recognized metric for SaaS as long as it has existed - it has meaning and can be audited.
Hendrikto 1 days ago [-]
So it is essentially a made up number. Like everything Dario claims.
SatvikBeri 2 days ago [-]
Run rate is annualized revenue based on some recent period, e.g. taking the last month of revenue and multiplying by 12. Revenue (classic) is a historical measure, e.g. revenue in 2025.
merlindru 2 days ago [-]
it's just extrapolation.
it's
(revenue of this month) * 12
in other words, "if every month was as good as this one, here's how much we'd make in a year"
that means, of course, if you make $1000 in january, your RRR is $12000.
...even if you end up making $0 every other month and thus only $1000 total that year.
thats why RRR is perhaps harmful. especially when it's not growing. it can be much bigger than the actual revenue. in anthropics case it's rapidly climbing, though, so it underestimates revenue if that growth keeps up
rhubarbtree 2 days ago [-]
Yes, the temptation is that you bring all revenue forward at any cost so you can pump value at an IPO etc
sirlantis 2 days ago [-]
It's extrapolated annual revenue based on a recent period (e.g. 12 x revenue_of_may).
petterroea 2 days ago [-]
Insane evaluation - Anthropic being around 50% of the Norwegian oil fund.
Haven880 1 days ago [-]
Deepseek MiMo and Qwen are now dirt cheap and give out free as well with quality about 95% against the very best so called fable mythos. And all that is on Huawei 7nm Ascend. All those companies added up nowhere near 1T and they affect EVERY SINGLE American lives producing parts that Americans used either via patents or parts. And we throw money to Anthropic with almost no moats at best 6 mths ahead. While Chinese companies holding patents more than Google Microsoft combined and with market bigger than entire USA economy. I think maybe investors too are hallucinating like AI. Sound like mega Lehman Groupon in the making!
What’s your point with the comparison? By your interpretation, investors should’ve never put money into any startup/scaleup/large business in other markets, since the Chinese market had more/bigger competitors.
Investors are betting real money on a payout. It seems disingenuous to think that they’re all idiots.
wg0 2 days ago [-]
That you all have to pay of course. With interest. Directly or indirectly. Through subscriptions or through pension funds and such.
cmiles8 2 days ago [-]
Probably the bigger headline here is that they’ve blown past OpenAI in revenue and valuation, with OpenAI looking increasingly shaky and vulnerable.
Aurornis 2 days ago [-]
Their valuations differ by about 13%. That's close enough that I wouldn't call it "blown past".
Things change fast in this space. Anthropic had a big boost from having the premier coding model for a while, but GPT-5.5 has closed that gap at a time when a lot of Anthropic customers are looking for cheaper alternatives.
Anthropic is coming off of a recent change to their enterprise billing that substantially changed the pricing for many users. They were smart to do the fundraising before the effects of that change could fully propagate.
cmiles8 2 days ago [-]
The acceleration rate has been extraordinary… they went from mostly unknown outside AI circles to the number one player almost overnight. If that’s not “blown past” I don’t know what is.
jmathai 2 days ago [-]
The branding of Claude is so much stronger than ChatGPT. Even Anthropic is such better branding than OpenAI (especially considering they're not open at all).
My wife knows about Claude because that's what I use and we pay for. She uses it also as a result. And inevitably she will talk about Claude to her friends.
bbg2401 2 days ago [-]
OpenAI is as open as Anthropic is anthropic.
boringg 2 days ago [-]
Great line. The irony of these names can't be overstated.
Marciplan 2 days ago [-]
for normies it is the exact opposite.
mediaman 2 days ago [-]
If "normie" means a noncorporate knowledge worker who uses the free version, yes.
For enterprise, Anthropic is crushing it. In the manufacturing sector I anecdotally hear a 2:1 ratio of Claude to ChatGPT for teams who are settling on a platform.
toraway 2 days ago [-]
At my company the grassroots advocacy from devs has certainly been for Claude Code.
Unfortunately even though we have a degree or two of seperation from most federal contracts the punitive DoD blacklisting had enough of a chilling effect on our legal team to make them drag their feet on approving any contract involving Anthropic.
So I pitched OpenAI Business with Codex so we could drop our Github Copilot Business subscription before the billing change takes effect June 1st which was approved without pushback.
I felt some responsibility for finding an immediate solution to dump Copilot since I was the one who recommended adopting Copilot in the first place, ugh... Our prices would have quadrupled based on the single month Microsoft in their beneficence allowed previewing with their tool to simulate what the post-rug pull pricing would have looked like.
Codex becoming more or less a 1:1 replacement for CC made that a no brainer given our options and the exploitative value proposition of Copilot under the new pricing model (which Microsoft evidently hoped companies like us would just accept despite being a third tier option in the dev space these days).
nozzlegear 2 days ago [-]
What's a normie?
lstodd 2 days ago [-]
Homo Sapiens Normie, an extinct subspecies distinguished by its low susceptibility to hype.
tredre3 2 days ago [-]
> The branding of Claude is so much stronger than ChatGPT.
Absolutely not, you live in a bubble. Everybody knows about ChatGPT.
Few non-programmers have heard of anthropic or claude, nor do they care. But they all know what ChatGPT is.
andrewl-hn 2 days ago [-]
ChatGPT is a word now. People may use Perplexity, or Google, or Grok to ask questions online. And later they tell you "ChatGPT told me this". It's a new "I googled in Yahoo".
famouswaffles 2 days ago [-]
ChatGPT is the 5th most visited site (as well as has nearly a billion weekly active users) and none of the competitors are even close. In the consumer space, Gemini is doing well but Claude is not even in the same galaxy. OpenAI is undoubtedly the leader in consumer LLMs and by a large margin. I'm sure there are mixups, but if someone is telling you they're using chatGPT, they almost certainly mean they're using chatGPT.
HDThoreaun 2 days ago [-]
The consumer market is worthless though. Consumers will never pay, so the only revenue option is ads which barely, if even at all, pay for inference costs.
famouswaffles 2 days ago [-]
Ads implemented remotely competently would be worth a lot of money and more than pay for inference. Inference is cheap, especially outside token expensive ordeals like agentic coding.
HDThoreaun 1 days ago [-]
Maybe. To really make money on ads they would need to embed them directly into the chat I think. Banner ads arent worth enough I think and google is able to make so much off them largely because people are already looking to click a link when they search something. People would just ignore them with genAI.
Maybe Im projecting my distaste for being psychologically manipulated, but I dont think users would continue using a genAI that embeds ads directly into the response when they can just switch to gemini where they only see banners.
what 2 days ago [-]
I’m pretty sure Gemini would be the leader in consumer LLMs considering it’s on every single search result. Every single google search is also usage of gemini.
famouswaffles 2 days ago [-]
Google stuffing things in the search results of existing users does not mean active participation or usage. (Not that I'm saying it's not getting used but it's just a feature of google search, and only a fraction of the kind of queries llms get anyway)
okdood64 2 days ago [-]
Why not? If it prevents people from getting their customized LLM answers from ChatGPT or Claude?
famouswaffles 1 days ago [-]
People use LLMs for a lot of things. Different kind of search is only one of them. AI mode is not stopping people from using ChatGPT because it's just a subset of consumer LLM queries.
Many people I know have started simply calling it "Chat".
"I had Chat help me write this" (I didn't, I promise)
They own "chat.com", I think OpenAI should pull the trigger and move to the shorter domain and finish the rebrand.
jollyllama 2 days ago [-]
What are streamers going to call their audience now?
solfox 2 days ago [-]
Perhaps it's just regional, but I've been noticing more and more people saying "chat" to describe ANY ai chat interface including ChatGPT. They might have a Kleenex problem on their hands.
zozbot234 2 days ago [-]
[dead]
basch 2 days ago [-]
>Few non-programmers have heard of anthropic or claude
They ran a super bowl ad. It's all over the construction industry. Claude is still not quite the Kleenex that ChatGPT is, but there is a pretty good chance lay people have heard of ChatGPT, Gemini, and Claude by now.
To disagree with the person below/above me that ChatGPT is the word used generically, when someone uses Gemini or Claude or Copilot, they TELL you which one they used, because they are essentially saying "i didnt use ChatGPT by choice."
Gemini is the one most likely to be used without people knowing which one they used.
jmathai 17 hours ago [-]
Agree. I didn’t mean stronger in that sense. I meant how it naturally extends to Claude Design, Claude CoWork, Claude Code.
ChatGPT is too awkward to do that with.
Spunkie 2 days ago [-]
Def sounds like a bubble to me. In my own bubble, ChatGTP is so well known over the others that people will often slip and refer to other AI services collectively as ChatGTP.
e.g. "I put it in chatgtp and..." when they actual asked Gemini.
satvikpendem 2 days ago [-]
Ironic that you spelled it wrong though thrice.
Spunkie 2 days ago [-]
Yeah I'm pretty sure I switch between chatgbt, chatgtb, chatgtp, and chatgpt quite often without realizing it.
Still everyone knows what I'm talking about.
HDBaseT 2 days ago [-]
It sounds like you have no clue what you're talking about.
The people in real life who say ChatGBT or similar are either so out of the loop the technology doesn't even matter to them, or simply are just stupid.
There is no way a person who can't get "GPT" right is even worth listening to.
kilroy123 2 days ago [-]
Agreed. My Mom, who is a grandma, uses ChatGPT every day. Lots of nontech people use it.
koolba 2 days ago [-]
Is she paying for anything though?
kilroy123 2 days ago [-]
Yeah, now she does. She got hooked. Which is very interesting to me. Same with my partner.
ramoz 2 days ago [-]
The vast enterprise industry (non-technical) is now aware of Claude/Anthropic.
vasco 2 days ago [-]
Everyone knew Altavista too
nozzlegear 2 days ago [-]
My sister, who isn't in tech and would be called a "normie" by people more online than myself, told me she switched to Claude a few months ago because of Anthropic's fight with the pentagon. IMO the unbubbled public certainly knows about Anthropic/Claude, especially given their Super Bowl ad and their stance on standing up to the pentagon/Trump admin.
pavlov 2 days ago [-]
I remember seeing expensive multi-page ads for Claude in the New Yorker over a year ago.
Their marketing has been working the high end of the “regular people” market for a good while.
baal80spam 2 days ago [-]
Do ordinary people really know what Anthropic is?
__s 2 days ago [-]
They know that "claude's the good one"
linuxftw 2 days ago [-]
Of course not. Normal people are using gemini, it comes pre-installed on Android now.
boringg 2 days ago [-]
This is the answer for gen pop. Gemini is going to mop up the floor on most use cases as its ingrained in google search.
postalrat 2 days ago [-]
They certainly know claude. I keep telling them they are all about the same.
cmiles8 2 days ago [-]
They know the cool kids have ditched OpenAI and now use Claude
pembrook 2 days ago [-]
Ironically their tussle with the US federal government is what made them a household name [1]
There's no better way to create awareness of a brand than to get it featured in the most popular reality TV show globally at the moment: "Thing Trump Did: Season 2."
Deepseek perhaps would be the top threat on a pure price/performance metric for either of them. It doesn't look like OAI is going for the value play .
gruez 2 days ago [-]
Comparing $/MTokfor models makes as much sense as comparing $/ghz for CPUs. Models have different tokenizers and take varying number of "thinking" to get to a solution. A far better proxy is how much it takes to do a run, which takes all of that into account. Such metrics are much harder to gather, but once source claims $3357 for gpt-5.5 vs $4686 for opus, the opposite of your conclusion.
There is no conclusion , I only stated the only objective fact to compare with that will not change for you to me.
Everything else is subjective to your setup, use case, configuration tuning and so forth.
More importantly bean-counters and decision makers at even 150+ seat orgs are looking at pricing sheets and enterprise contracts not how it performs for some team in a specific harness today to make million dollar annual contracts. It is not common for procurement teams to do commission the level of detailed analysis or large scale pilots that will actually hold for the duration of contract.
That doesn't mean that GPT-5.5 is selling less than Claude at all, just that cost is not the primary driver if list price is not cheaper, there is reason these are published in the same format by every vendor, because the common metric is how finance likes to compare with.
Spartan-S63 2 days ago [-]
Most variants of GPT-5.5 are less chatty and token-intensive than Opus 4.8/4.7, so despite the output token price being higher, it generates fewer tokens, so the net cost is lower.
Per-token pricing is totally sensible from the provider-perspective on mapping COGS to revenue, but for a consumer, different models will produce more or less tokens, meaning the cost calculation is multi-dimensional.
manquer 2 days ago [-]
You can configure model to be terse/concise with output style ? There are plenty of popular projects like https://github.com/JuliusBrussee/caveman which do it for you even.
Input/Cache/Output ratios are use case and configuration dependent . Any benefits in one model can usually be roughly to another with configuration tuning, and discussions devolve into subjective experience.
Pricing sheet is the objective way to compare cost.
epolanski 2 days ago [-]
Sure, but people were claiming just 6-12 months ago that OpenAI had "brand moat" and "first mover advantage" that others would not have.
Considering that their models are de facto extremely close in performance you would think that these arguments would've held, but they clearly don't.
alecco 2 days ago [-]
Anthropic is at the mercy of 3rd party datacenter contracts. AFAIK OpenAI will soon run mostly on on their own GPUs.
I don't like Altman and I am still upset about his memory deal last year but he prepared for the current shortages months before anybody else. Meanwhile, Anthropic seems to lack any plans besides third party contracting. IMHO they got very lucky with xAI and Google having spare capacity and willing to rent it. But what about next year?
lumost 2 days ago [-]
Which also leaves OpenAI vulnerable to NVidia's aggressive pricing. To my knowledge Anthropic is relatively well positioned across multiple compute vendors/hardware providers.
abirch 2 days ago [-]
It also leaves OpenAI vulnerable to any GPU breakthroughs. You could imagine company X comes up with a XPU that is 100% faster than what's currently there.*
* NVidia GPU, Google TPU, Apple SoC, etc.
jiveturkey 2 days ago [-]
We are still in the short-half-life phase of GPUs. If a 2x faster GPU is on the horizon, why wouldn't OpenAI already be in line to buy? They aren't buying just 1, they are buying multiple datacenters' worth. So they wouldn't be a low priority, back of the line customer.
A short half-life means you are going to quickly dispose of what you have now, anyway. In fact most current datacenters can't even handle Vera Rubin, so I don't think there's short term risk here.
karmasimida 2 days ago [-]
You have missed the point
Nvidia has probably monopolized several upstream supplies to manufacture critical chip components for next 2 years, the HBMs and Optics component from LITE, as well as TSM capacity. Let alone those power components they funded themselves.
Let's say you have a genius design, but you will have it close to impossible to compete with Nvidia in getting it to volumes.
Jensen is a player, he isn't fooling around with all these Asian trips just to wine and dine
zozbot234 2 days ago [-]
nVidia can only 'monopolize' these components for itself inasmuch as other industry players are not seriously interested in them. This can change rather quickly.
karmasimida 2 days ago [-]
How ? Define quickly please.
bombcar 2 days ago [-]
Isn’t everyone at the mercy of NVidia deciding that everyone has to use NVidAI and refusing to sell anything to anyone?
manquer 2 days ago [-]
> their own GPUs
Everyone has critical risk on multiple parts of the supply chain. GPUs and Memory are just things OAI mitigated for.
Power - Bigger bottleneck than GPU or RAM perhaps, New Grid connected capacity is typically 10+ year timescale with lot of regulatory friction. Captive capacity is also quite constrained - now Gas turbines have 7+ year wait time.
There are plenty of hard constraints that OAI cannot easily solve either.
outside1234 2 days ago [-]
Don't worry, they will buy up OpenAI's contracts once they implode.
dopa42365 2 days ago [-]
The same 3rd party datacenters from the same few companies that everything else runs on? If there's demand, hyperscalers will supply.
thereitgoes456 2 days ago [-]
Stargate is not real.
It is not clear that running one's own datacenter is a competitive advantage. Why do you think OpenAI can handle that?
Gomotono 2 days ago [-]
Stargate as a project is real, they only stoped the Stargate UK thing.
Anthropics relativ longterm contract with xAI def shows that they can fill the capacity vs Musk not. OpenAI and Anthropic are both using a lot of capacity so its fair to say that this is an advantage.
If they stay very close competitive (which they are), your own datacenter does reduce token price.
llm_nerd 2 days ago [-]
>Anthropic is at the mercy of 3rd party datacenter contracts
I mean, this is a bit like complaining that McDonalds doesn't have their own herds of cows. OpenAI actually isn't in the business of buying GPUs or running data centres, and it's pretty weird to think that's an advantage (though it comes up constantly on here, as Anthropic keeps eating OpenAI's lunch).
There are many suppliers that are desperate to fight for Anthropics business, and it has shown an agility to embrace whatever advances in the industry come along. Anthropic is now running across a million or so Google TPUv8s, for instance. If tomorrow someone else comes out with a better GPU/TPU, they can embrace it in a heartbeat.
All while OpenAI sits on their rapidly depreciating GPUs.
Or...actually they won't, because OpenAI doesn't take business advice from HN. The vast majority of OpenAI's compute is from Microsoft, Oracle and so on. They're smart enough to not become a big hardware purchaser when that isn't their business. The core claim of your comment simply isn't true at all, nor is that the direction OpenAI is moving.
bombcar 2 days ago [-]
To be fair apparently a big part of McDonald’s success is having their own real estate.
CuriouslyC 2 days ago [-]
Anthropic is riding a hype wave as a result of brilliant marketing. OpenAI has the better products, higher reliability and better community relations. I don't expect the situation to continue.
kromokromo 2 days ago [-]
I disagree. They have been winning lately because of better harnesses and interfaces. New actual decent features are shipped almost weekly on Claude code and Claude desktop.
outside1234 2 days ago [-]
OpenAI has a broken business model
hereme888 2 days ago [-]
I agree
fontain 2 days ago [-]
I’m not so sure. We only need to look at Uber’s example of companies realizing they’re spending way too much and trying to rein it in. Claude has excellent revenue but it is highly dependent on very rich technology companies continuing to spend lavishly without seeing returns. The music will stop at some point and Anthropic will be hit the hardest. OpenAI may have less revenue but it is distributed across many, many more customers and use cases, it’s resilient. And even if Anthropic do, somehow, manage to keep their customers spending huge amounts on Claude, they’re very vulnerable to being undercut by OpenAI given codex is pretty much at parity. Anthropic seems more vulnerable to me.
Archonical 2 days ago [-]
I think it's somewhat guaranteed that the music will at least die down a little bit. We saw this with cloud companies being bitten by cloud cost optimization initiatives. I can't imagine we won't see the same with AI, especially as the workforce stops trying to tokenmaxx to save their role.
Gomotono 2 days ago [-]
If you look at the adoption curve of Claude, I don't think we have reached anything near peak.
Analemma_ 2 days ago [-]
Every week there's at least one post on the HN front page bitching about API errors from Claude because Anthropic doesn't have enough serving capacity. I really don't see any signs they're "spending too much", the actual evidence on the ground seems to be exactly the opposite: constant exasperation that they're not spending enough.
CuriouslyC 2 days ago [-]
Unlike OpenAI, a lot of Claude's infra problems are self-inflicted and not completely raw-capacity related.
newaccountman2 2 days ago [-]
What he means is the customers realizing they are spending too much on Anthropic.
claytongulick 2 days ago [-]
I just finished talking to a dev manager friend of mine at a household name company.
He told me they are massively pulling back on the AI stuff.
Right now the lashback is about cost, because that's the most easily measured pain point.
Soon, we'll start seeing a deeper understanding of the quality issues. At that point, it's likely this whole experiment gets firmly put in a bin of the toolbox where it belongs.
icedchai 2 days ago [-]
I know people at medium size companies where they are tracking AI costs very carefully. They are pulling back to levels under $100/week in AI spend per engineer, encouraging use of lower quality, lower cost models, etc.
vidarh 2 days ago [-]
You can run models near 24/7 per developers at that price with judicious choice of subscriptions, so that's not really saying much.
Most people don't yet have mature enough setups to fully exploit that level of use.
icedchai 2 days ago [-]
With open models, perhaps.
But $100/week isn't going to get you 24/7 use of Claude Sonnet.
vidarh 2 days ago [-]
$100/week will get you both a higher tier of Claude and significantly more tokens with GLM5.1 or Kimi, both of which are competitive in a decent harness (for Kimi that means their own CLI - it works well in that, but has a lot of quirks that requires special treatment). Just slightly more will get you all three.
mitjam 2 days ago [-]
Enterprise accounts pay per token.
vidarh 2 days ago [-]
Of all the companies I've done work for over the last couple of years, only one have used an enterprise account.
1 days ago [-]
decimalenough 2 days ago [-]
I don't doubt you, but $100 is approximately the cost to company of one hour of dev time. If companies end up being willing to spend only 2% of their dev budget on AI, this bubble is not going to last long.
icedchai 2 days ago [-]
I agree. $100/week is absurdly low if you want to allow for any real experimentation and productive use of these tools.
fontain 2 days ago [-]
I mean Anthropic’s customers are spending too much on Claude. Anthropic’s customers are encouraging tokenmaxxing amongst their employees; measuring employees by token usage. That’s great for Anthropic’s short term revenue numbers but terrible long term because at some point companies will realize tokenmaxxing is not good. OpenAI is much less exposed to tokenmaxxing, which is a good thing.
solenoid0937 2 days ago [-]
> at some point companies will realize tokenmaxxing is not good
Why? Have we figured out the limits of what agents can do?
> OpenAI is much less exposed to tokenmaxxing
I don't think this is true, from my own experience & chatting with my acquaintances.
fontain 2 days ago [-]
Tokenmaxxing is the practice of measuring employees by how many tokens they use, encouraging employees to burn tokens needlessly, it is unrelated to what agents can do.
If a task can be completed with 100k tokens but employees are considered better performers if they complete it with 500k tokens instead… that’s unsustainable and cannot possibly benefit Anthropic in the long term.
At some point, Amazon and Uber and so on and so forth are going to realize that actually, employees using 100k tokens or even 50k tokens is better than 500k and Anthropic’s revenue will fall off a cliff.
solenoid0937 2 days ago [-]
Oh. I thought tokenmaxxing was just removing token limits
fontain 2 days ago [-]
I think removing limits is fine. There’ll be overspend and at some point adjustments in expectations as we learn more about the value that can be delivered which will likely result in a reduction in spend, but even now, during this period of relative immaturity about measuring the value of output, so long as more tokens = more output, I don’t think the introduction of limits represents much of a risk to Anthropic and OpenAI. Tokenmaxxing is uniquely bad because it is not tied to any additional value (more tokens for the same output).
And I could be wrong about tokenmaxxing being a Claude specific problem but as far as I can tell, all of the major companies encouraging employees to maximize their token usage are Claude Code users. And the music has to stop on that at some point, whether because the companies run out of money or because they learn better ways of measuring productivity in the AI age. And if tokenmaxxing is what is driving Anthropic’s lead in revenue, it could be catastrophic to lose that, because Anthropic are spending billions of dollars per month on the infrastructure to support it.
If tokenmaxxing is evenly distributed between Anthropic and OpenAI then they’ll both hurt but equal hurt shouldn’t disadvantage either much.
2 days ago [-]
wslh 2 days ago [-]
This business and financial race is probably the craziest in human history, so zig-zags are expected. One company may take advantage on one curve while another is stuck in the pits.
karmasimida 2 days ago [-]
OpenAI isn't shaky or vulnerable, this market will need at least 2 players.
I see most of the surge here comes FOMO AI spending which will have to be dialed down later half of the year, otherwise those companies will have to layoff to fund their AI bill, which is harmful to their business.
Anthropic grabs its bag at the peak, but feast is over.
dbbk 2 days ago [-]
The other player is Google.
dyauspitr 2 days ago [-]
Having used Codex though I don’t think OpenAI needs to worry. It’s a solid product and they will share the market.
ignoramous 2 days ago [-]
How? OpenAI and Antrophic are basically the Big 2 racing away at light speed; the others who can't get near them are perhaps shaky & vulnerable. And sure, there's a garden full of those.
cmiles8 2 days ago [-]
Because the market almost certainly can’t support two foundation model labs given the increasingly little difference across models and the massive sums of cash required to keep it all going. There is no big 2, just a race to survive and be the big 1.
an0malous 2 days ago [-]
It probably can't support any because there's no moat and smaller, open source models are catching up. This is like investing $1T into mainframe computers in 1980.
treis 2 days ago [-]
There's at least two markets here. Consumer ad driven and worker augmentation markets. Likely a 3rd as a backend infrastructure provider to a bunch of value add companies.
I think Google has caught up enough to certainly be a player in the consumer ad driven market.
I also don't think only one foundation model adds up. Now that the trail is blazed a dozen companies can likely make a good enough model. The question is if there's a moat to make it winner take all
dchftcs 2 days ago [-]
China will make sure they have a frontier lab, there's plenty of chance for Google to catch up once the compute crunch gets more serious.
TacticalCoder 2 days ago [-]
Google needs to catch up on what? Devs mindshare? The latest Opus 4.8 carefully selected benchmarks made sure to pick Gemini 3.1 Pro and not Gemini 3.5 Flash: 3.5 Flash is beating Opus 4.8 on several of the benchmarks Anthropic posted but simply was ignored.
I don't think SOTA-wise Google has a lot of catch up to do.
dchftcs 2 days ago [-]
Gemini 3.5 Flash is not good at coding in practice. Gemini 3.1 Pro too, in particular is known to be bad at tool calls. Many companies would love to have alternatives to Claude Code (as it's a significant risk to depend on one vendor), so far most of the buzz is about moving to Codex but much fewer talk about moving to Gemini. All these benchmarks are not very informative, the Chinese labs do better on these benchmarks than in practice, for example.
solenoid0937 2 days ago [-]
Disagree, both are coexisting fine today.
claytongulick 2 days ago [-]
A series "H" for $65 billion and no path to profitability is existing fine?
Yeah it's very murky, but if they're arguably profitable under some definition of profitable, then it's ridiculous to claim there's "no path to profitability"
Especially if you assume that 6 months ago they weren't very close to this version of profitable
solenoid0937 2 days ago [-]
Sure, if you think there is even a small chance that OpenAI and Anthropic will get to AGI, it's practically a bargain.
They don't seem too far off to me.
2 days ago [-]
watwut 2 days ago [-]
If it cant support two competing compamies, something is very wrong. Oligopoly is bad, monopoly worst.
Well functioning market is supposed to have many, as in a lot, companies with similar products. To create competition.
dotcoma 2 days ago [-]
isn't Google going to win the race anyway ?
throwaway85825 2 days ago [-]
In terms of personal assistant AI the monopolists have a massive advantage because they control the platforms and can box out competitors from deeply integrating with the OS.
dotcoma 2 days ago [-]
I don't see it as merely boxing out competitors or integrating with the OS.
People have been going to Google for a quarter of a century, and I imagine they will continue to do so.
As far as I am concerned, I moved to DDG in 2013 and never looked back, but I find Google AI intriguing.
conradkay 2 days ago [-]
That feels really hard to argue now that Anthropic/OpenAI are so much bigger
How the hell do you crush a ~1T company on the one thing they have all their focus on?
SXX 2 days ago [-]
Data. Google has access to unphasmable amount of real human-created data with zero expectations of privacy (wink wink Apple): videos, photos, search, navigation, mobile app usage including competition platforms, emails, etc.
Both Anthropic and OpenAI only has access to whatever they can buy or steal.
And it's becoming increasingly hard to get fresh uncontaminated data for training. No amount of money can buy that.
conradkay 2 days ago [-]
I suspect it's less true now that synthetic data has worked so well, and multimodal doesn't seem to transfer as well as many would've hoped
> Both Anthropic and OpenAI only has access to whatever they can buy or steal.
A trillion can buy you quite a lot! Like offer some company a ton of money for data, and if they say no simply buy said company. Bonus points if it's someone like Atlassian who's stock price is getting hammered largely because of you.
dotcoma 2 days ago [-]
You start with a 90% share in search, I guess.
Me, not a SWE, I have never even tried Claude
(and I am not sure I prefer ChatGPT over Google)
2 days ago [-]
andriy_koval 2 days ago [-]
Google likely has its market share too, you can track how fast Cloud revenue increased.
decimalenough 2 days ago [-]
It's a gold rush and Google is both selling shovels and digging for gold itself.
dbbk 2 days ago [-]
Google is the present winner, they're beating everyone else
fsuts 2 days ago [-]
OpenAI were perhaps hampered by the Microsoft deal that was just renegotiated allowing them to go onto AWs etc?
andy_ppp 2 days ago [-]
I wonder if being consistently candid is a superior business strategy?
henry2023 2 days ago [-]
“Caballo que alcanza, gana”
ripvanwinkle 2 days ago [-]
[dead]
outside1234 2 days ago [-]
Anthropic is the Google to OpenAI's Yahoo.
p1esk 2 days ago [-]
Do you mean Facebook to Yahoo?
8f2ab37a-ed6c 2 days ago [-]
Say you join Anthropic now as an employee. What are the chances of your equity appreciating in value? I don't think we have any historical precedents to this.
bix6 2 days ago [-]
Well presumably nobody investing in this current round expects anything less than a 3x
hiddencost 2 days ago [-]
inflation
I suspect we'll have our first $10T company in the next 2-3 years. That's only doubling.
shreezus 2 days ago [-]
Anthropic is one of the few companies that will likely hit an infinite valuation if they succeed in their mission.
singingtoday 2 days ago [-]
Jokes aside, I do see potential of them hitting multiple trillions of valuation.
nearlyepic 2 days ago [-]
“Infinite valuation”? Give me a break.
stuxnet79 2 days ago [-]
TAM will go parabolic once bots are exchanging goods and services with other bots with us meat space humans sitting on the sidelines.
epolanski 2 days ago [-]
It's always a very borderline idea to hold equity in the company you work, argument being that if the company goes south or severely underperforms, you may find yourself laid off and with worthless equity.
So you're assuming lots of risk and putting it all in the same basket.
There's no shame in getting 100k $ worth of stock, selling it and putting it on some vanguard fund and diversifying, in fact it's statistically the best move you can do.
Of course, you can be like those many googlers that did this and then regret in hindsight.
iooi 2 days ago [-]
So close to being the first kilocorn. A unicorn = 1 billion, this is almost 1k.
someperson 2 days ago [-]
Hasn't SpaceX achieved that though?
And Saudi Aramco before they IPO'd
Jblx2 2 days ago [-]
Kibicorn has a nicer ring to it ($1,024 billion).
robinhood 1 days ago [-]
The amount of dollars that exist in this economy is mindblowing.
Ten years ago, we were talking millions, and this was already incredible when companies could raise a couple of them.
Now, headlines are only about hundred of billions. I do not know what to think about that, apart from the fact that I wish that we were putting that money to enhance human lives in general. Of course, people will say that these tools will help humans in the future, but 1) at what costs, and 2) I would prefer, I don't know, bridges or infrastructure, or free healthcare, or food for everyone.
mark_l_watson 2 days ago [-]
Until Anthropic, OpenAI, and Tesla have IPOs and are then bound by some laws to be truthful, I don’t want to bother about their possible valuations.
I do care about: how useful their products are vs. cost and how secure are their businesses. Actually I only care about the first thing since these services are hot swap-able with some effort.
GenerWork 2 days ago [-]
As someone who knows admittedly knows nothing about startup funding rounds, how many more rounds of funding can they do before an IPO? Is it effectively infinite?
tomwheeler 2 days ago [-]
Effectively infinite. Databricks is a good example. They're still private after 13 years and closed a Series L round last year. Stripe is similar.
Having been through an IPO before, it was good for employee liquidity, but bad for the culture and long-term success of the company.
charlie0 2 days ago [-]
Dead capital. There's no need for public funding until they are reasy to cash out at the top, if ever.
solenoid0937 2 days ago [-]
How do investors cash out? Do they sell to new round investors?
barbarr 2 days ago [-]
Going off the other reply, I wonder if a highly-active secondary market means that companies can raise series [A-Z]+ rounds effectively forever, where each "round" just refers to a giant purchase of shares under strict company supervision. Is this the new game for startups?
DenisM 2 days ago [-]
If investor pool becomes too loose the company becomes de facto public, subject to all SEC-enforced regulations.
The judgment is subjective though, so pushing the boundaries could be a calculated risk.
bix6 2 days ago [-]
Correct. There is also a secondary market.
epolanski 2 days ago [-]
> There's no need for public funding
You're assuming private liquidity to be infinite and private credit (that fuels VCs) to always have favorable rates.
charlie0 7 hours ago [-]
When you have companies like SpaceX and Anthropic raising billions, it is hard to believe that private credit isn't virtually infinite.
misiti3780 2 days ago [-]
so how do stripe employees get liquidity? can anyone sell their secondary shares?
tomwheeler 2 days ago [-]
I can't speak for the specific case of Stripe, but it's fairly common for private companies to have a "tender offer" in which employees have the opportunity to sell some portion of their equity. This is often done in conjunction with a new investment round.
Marsymars 2 days ago [-]
Outside of big tech, it's also fairly common for private companies to simply not offer equity to employees.
Company I worked for had a deal where I could sell back to the company at any time.
The price was determined by a formula based on revenue and such, so I always knew what they were worth.
I was not allowed to sell to anyone else though.
jiveturkey 2 days ago [-]
There's a newish term for this: RLO, Recurring Liquidity Opportunity. These are tender offers at some recurring interval. Even some companies that have a shorter lifespan (say 7 years) offer this.
clint 2 days ago [-]
Stripe might buy back the shares at a good price. They might be able to sell on secondary markets.
newaccountman2 2 days ago [-]
Private/secondary markets.
vanuatu 2 days ago [-]
regular tender offers
dkdcdev 2 days ago [-]
I believe the canonical example is Databricks on round L
nerdsniper 2 days ago [-]
I believe Databricks series L round raised $4B in late 2025, but earlier this year they raised another $5B so technically they've maybe completed series M round and are "on" series N round now? The press releases are a bit confusing to me.
tomwheeler 2 days ago [-]
It's semantics, but the latest raise might have been a follow-on to Series M, not a new round (to be clear, I know nothing about their finances, just speaking from experience at another company).
boringg 2 days ago [-]
What happens when you make it to Z do they then start doing Z1 rounds or does it reset to AA like excel (though that would be confusing)? Hex?
nitwit005 2 days ago [-]
The AA scheme has already been used for "down round", so they might have to get creative.
jmathai 2 days ago [-]
I imagine there are ways for existing investors to achieve liquidity while still raising venture funding. But an IPO is "the" liquidity event and I imagine there will be pressure from investors for that.
I also imagine that venture funding rounds have a lower ceiling than the public markets - but at these rounds I'm not so sure!
toasty228 2 days ago [-]
Once they reach series Z does it go back to A or do we get a new format like AA, AB ?
Npovview 2 days ago [-]
We live in Unicode times. We switch to Greek alphabets.
α β γ δ ε ζ η θ ι κ λ μ ν ξ ο π ρ σ τ υ φ χ ψ ω
tomwheeler 2 days ago [-]
After those are used up, it moves to Devanagari, Hangul, Katakana, Hiragana, and then Kanji.
rvz 2 days ago [-]
Depends on the investors if they see growth. The downside is dilution. Preferably they just want the Series I as the IPO in this case.
They cannot raise forever, SpaceX has done more rounds but the timing is most important.
ielillo 2 days ago [-]
usually you would go through seed funding, the series a,b, and possibly a1 and b1. If you entered c or d territory it meant that you still had a chance but vc would be following you very closely. After d, you could raise money, but it would be under very unfavorable conditions
vidarh 2 days ago [-]
The number of rounds is irrelevant. Having crunched the data, what is relevant to terms is simply as you'd expect the rate of growth. The only reason it rarely happens with fast growing companies is that the liquidity of an IPO is attractive. As a result, companies doing many rounds are disproportionately companies that are performing too poorly to try and IPO.
re-thc 2 days ago [-]
Yes, whatever you like
wina 2 days ago [-]
they can do as many as they want. but at some point investors need/want to exit their positions and push for an IPO. That point is different for every company.
notarobot123 2 days ago [-]
Well the market clearly thinks most of us will be without software engineering jobs by the end of the decade. There's clearly enough proof that the trend is heading in that direction to justify huge capital investments.
mettamage 2 days ago [-]
I wish I could invest into it, I'd at the very least have invested in their Series F. It was a no brainer by that point. If anyone could teach me how to get into stuff like this, that'd be awesome. I'm from the Netherlands, so not American. Though I'm married to an American.
outside1234 2 days ago [-]
Don't worry, it will be all dumped on the SP 500 as soon as it is at its hype peak.
pestatije 2 days ago [-]
just talk to your banker they sure know how to invest your money
mettamage 1 days ago [-]
Aaaaand it's gone!
esafak 2 days ago [-]
[dead]
bix6 2 days ago [-]
This is all getting a bit tiring. Show us the S1 already!
fsuts 2 days ago [-]
Everything will be clearer after they IPO
Until they IPO and the investors make their money, who knows what is behind the revenue stated
slashdave 2 days ago [-]
What happens when they run out of letters in the alphabet?
adamfarhadi 2 days ago [-]
I’m biased but perhaps they can consider expanding to the Swedish alphabet and introducing series Å/Ä/Ö funding? Then they should have enough headway to be valued in the order of quadrillions of USD.
jedberg 2 days ago [-]
Series AA?
Databricks raised an L round last year, so they'll have to solve it first.
mutator 2 days ago [-]
This did round involve a secondary? If yes, any data to suggest that these secondaries are leading to increased spending outside of housing and propping up the local economy?
jiveturkey 2 days ago [-]
There was a secondary sale in April.
whalesalad 2 days ago [-]
They're going to run out of letters pretty soon.
gruez 2 days ago [-]
Anyone in finance should know that excel doesn't run out of letters (for columns) either. It just rolls over to AA, AB, etc.
onychomys 2 days ago [-]
They eventually stop, Excel has a max column size of 16,384. Not sure what letter combo that would be (ZZZZZZZZZZ or something?), but it does eventually halt.
Does this mean no IPO this year? What are the likely chances a company at this stage needs two rounds of funding within a year?
system2 2 days ago [-]
They will have to IPO before the ai hype dies.
4ashga 2 days ago [-]
That announcement is a bit short on details. I suppose that, like in the previous rounds, there are some strings attached and they'll not get all of it at once.
Hynix is participating with a new circular deal. Hynix is also valued at $1 trillion now, which is positively insane.
This scam will implode harder that the housing bubble.
techpression 2 days ago [-]
The circulation of money in AI is deeply troubling (NVIDIA being one of the worst I believe), either the bubble doesn’t pop and corruption like this is considered legal (self-inflating money amongst friends) or it pops and the financial hurt will be felt for a decade.
shtopointo 2 days ago [-]
Smart move from Anthropic.
They're the belle of the ball right now, everybody is talking about them, everybody wants to invest in them, so they can call the shots.
Then they'll have money in the bank for a long time no matter what happens – IPO, market downturn, etc.
Takes off lots of pressure so they can continue focusing on the product.
king_zee 2 days ago [-]
you talk like claude of course you're glazing them
isodude 2 days ago [-]
I wonder if the investors used Claude to see whether they should invest or not hmm..
What is this, Series for ants? It barely covers Andrej's sign-on bonus.
dyauspitr 2 days ago [-]
To put this into perspective, India has built 4000+ miles of new railway lines over the last five years including a Shinkansen style bullet train. That’s more than all the railway lines in the country of Switzerland for about $12 billion. The money being thrown around here is mind blowing.
BobbyTables2 2 days ago [-]
Indeed! And the returns from such an investment in infrastructure will likely be profound.
I feel AI is a bit different, as in there is a spectrum ranging from “utterly stupid” (early ChatGPT), “very helpful” (kinda now), “SkyNet 2.0” (what good are humans!).
As algorithms and technology improve, AI will be both cheaper and more capable. Companies like Anthrophic wouldn’t be the vaulted celebrities as they are today. At that point, I’m not sure what value much of society can provide…
At one time, blacksmiths had a valued place in general society. Today, not so much…
yoyohello13 2 days ago [-]
NGL, I would gladly trade opus 4.8 in exchange for my city fixing the roads. Just imagine if even a quarter of the money being dumped into AI went to something that actually improved peoples lives. Man that would be a sight to behold.
limflick 1 days ago [-]
AGI in the next 20 years is more likely than all the roads in my country getting fixed anytime in the future.
dyauspitr 2 days ago [-]
Well, if everyone is out of a job, the government is going to have to do something so I’m not too worried about it. If I had to be an optimist about it, I would say that this might be the perfect catalyst to build out a lot of renewable energy (for data centers) and maybe actually implement UBI.
pestatije 2 days ago [-]
give a bot a vote
dyauspitr 2 days ago [-]
Who is a bot? Boot on head, sharpie in pooper.
field_reader 2 days ago [-]
[flagged]
terekhindc 1 days ago [-]
[flagged]
heloqui 2 days ago [-]
[dead]
johnmorris100 2 days ago [-]
[dead]
johnmorris100 2 days ago [-]
[dead]
aanet 2 days ago [-]
I'll have some of that joint they be smokin'
/s
enraged_camel 2 days ago [-]
[flagged]
InsideOutSanta 2 days ago [-]
*run-rate revenue
Without more information, this number is impossible to interpret.
Analemma_ 2 days ago [-]
This has become a meme which is way out over its skis. Yes, run-rate is not the complete story, but "impossible to interpret" is way overstating the case.
InsideOutSanta 10 hours ago [-]
No, it's not. We do not know how Anthropic calculates it, or even if they calculate each number they report identically, hence the number is impossible to interpret.
vb-8448 2 days ago [-]
Unless you have access to Anthropics book your claim is as baseless as Ed Zitron's ...
surgical_fire 2 days ago [-]
With all that revenue, while being the most expensive model, they are still unprofitable without trickery.
This really doesn't paint the good picture you seem to imply.
Lionga 2 days ago [-]
What do they then need another $65B for? To sell 200$ plans that cost them 1000$ to fullfill.
I can have $47B in revenue if I sell something that cost $80B to produce ez pz.
Maxatar 2 days ago [-]
Revenue is not profit.
NewJazz 2 days ago [-]
[flagged]
rvz 2 days ago [-]
[flagged]
underyx 2 days ago [-]
you also can't write openai without a pen
bensyverson 2 days ago [-]
Airtight logic
goodbirb 2 days ago [-]
Some new form of mystic numerology/gematria-based investment logic?
Imustaskforhelp 2 days ago [-]
I think your conclusion might be right that AGI does just feel a bunch of hype but the reasoning in middle feels flawed...
like how 13^2=169 and 31^2=961 or 10^2+11^2+12^2=13^2+14^2
moralestapia 2 days ago [-]
9k karma, whew
malthaus 2 days ago [-]
nice, that's another 4 years of spacex data center usage runway!
OK, so their self-reported run-rate revenue hit $47bn in early May.
For comparison:
Apr 6th 2026: https://www.anthropic.com/news/google-broadcom-partnership-c... - "Demand from Claude customers has accelerated in 2026. Our run-rate revenue has now surpassed $30 billion—up from approximately $9 billion at the end of 2025."
So that's $30bn at the start of April.
Feb 12th 2026: https://www.anthropic.com/news/anthropic-raises-30-billion-s... - "Today, our run-rate revenue is $14 billion, with this figure growing over 10x annually in each of those past three years."
That was $14bn on Feb 12th.
And $9bn in December (according to the above April 6th link.)
Existing corporate command and control has optimized for people control, because people cost money and performed work. Control their assignments, and you control costs and what's worked on.
Widespread unmetered AI turns this on its head, because suddenly each employee is directing their own work and the AI spend that comes with it.
F.ex. Bob in accounting may think it's a brilliant idea to rebuild Lotus 1-2-3.
That may help Bob, but 10x'ing Bob's spreadsheet output doesn't change the company's profitability, because it wasn't a limiting factor. It was to Bob, but not to the company's revenue generators.
HN didn't like this article when it hit, but I thought it made a good point that corporate C2 is going to be the first thing AI adoption breaks: https://www.forbes.com/sites/jasonwingard/2026/04/23/vibe-co...
Increasing AI spend without profitability improvements is a symptom that C2 is failing (or was insufficient to begin with).
Seen through a charitable "CEOs know what the fuck they're doing" lens, the preemptive layoffs are about forcing AI efficiency gains in areas CEOs expect them: instead of allowing those departments' remaining employees to build their own apps, they're forced to deploy AI to cover for their missing 3 team members.
Unfortunately, the layoffs were executed before there were solid results about which departments could benefit from AI use (and without a plan for continuity of institutional knowledge), so... we'll see.
I get what you're saying about how every employee becomes a liability when they are let loose with an AI, and totally agree. But I think the layoffs have a much more financial root because they are so widespread across so many companies and even industries (not just tech)
Layoffs + access to AI forces most of that AI use to make up for the layoffs.
Which is a pretty shitty way to burn your employees out... but it has a method.
(Also, I'd say AI infra companies vs everyone else are apples to oranges. This point would be more applicable to a non-AI infra company that's paying for AI use rather than AI infra capital)
Turn a potential existential competition into (a) maybe I win or (b) I commoditize the market and you can't make money at it either.
1) I have a new extra cost 2) How does that make me more profit, and improved cashflow
I know I'll be bombarded with metrics about productivity, feature completion, bug fixes etc. But someone is going to have to tell me how that equates to more sales. And who is going to do that? I'll be worried that the feature wishlist will just creep up now we can handle more throughput, and yet everyone will be telling me that I don't need to worry about the lack of new customers this quarter, one more new feature and we will catch up in Q4. You can see how that could make one grumpy. Then when the sales don't come, I'll tell the CTO that they need to balance the books, if they want to use expensive tools to make each developer more productive then they will need fewer developers...
...but I don't want that, I want more great features that people will pay for, and faster. But they have to pay for themselves.
When one is under the grip of irrational exuberance and FOMO due diligence goes out the window.
For example look at Sequoia’s investment into FTX and article about Sam that accompanied.
If for no other reason than they would rather spend their days procrastinating on everything else, so they can obsess over some of the most interesting numbers they will ever see.
ARR seems like a interesting concept which hides away the concrete details and is counting on a futuristic possible commitment especially for companies at this early stages. The crucial detail is probably the money being spent, that is what is fueling this sale of equity at this point.
Yes, it is a meaningful financial measure of actual progress, exactly where progress is most needed for a new company.
> which hides away the concrete details
A normal metric isn't a magic trick. It's just the number it is.
> The crucial detail is probably the money being spent
Accelerating (not just fast) revenue growth at an astounding rate, with absolute numbers that are enormous for a new company, bonkers for a 2500 employee startup [0], is the crucial "detail".
There are lots of companies with deep pockets and compute, making great AI efforts with teams of smart people, that would love to be doing, but are not doing, what Anthropic is doing and doing well. That might be the second crucial detail.
[0] https://en.wikipedia.org/wiki/Anthropic
ARR as a number for steady business or public company is much simpler, ARR for companies at this early stage fueled by the motivation for IPO is more than just a number. Its an attempt to convince the market for a certain outcome without revealing the books
> Accelerating (not just fast) revenue growth at an astounding rate
This acceleration is fueled by somewhat artificial demand (AI mandates by executives across the board, still figuring out the realistic use cases) and subsidized pricing. The expenses matter cause they are the ones which are going to dictate if the business is sustainable or not.
Of course ARR is simpler for less dynamic companies. Tautology.
Of course a PR is a communication to the market. Also a tautology.
You are framing functional numbers and behavior as fraudulent because... a newsy PR isn't a formal disclosure?
Cynicism is fine. However: accusations should reference some evidence, not just distrust.
If you think investment banks will catch them during the IPO: They have launched many overpriced IPOs in 1999/2000 and also want a high IPO price.
Historically, listing rules asked: "How much money did you make last year, and do you fit our standard corporate governance box?"
Today, NASDAQ's rules ask: "Do you have the massive market capitalisation, sufficient institutional public float, and transparent liquidity to ensure fair and orderly trading?"
Here are the obvious ones:
1. Free float - Every company that intended to list was required to have at least 20% free float.
2. Index weights were based on the free float.
3. Time before inclusion into the Indices (min 12 months) now 15 trading days
4. Lockup period - minimum 12 months up to 24 months - not 180 days.
I wish people would understand that if America had a functioning criminal justice system, no one would have heard Elon Musk nor Donald Trump.
"The passive funds holding trillions of dollars of 401(k)s and other investments are rushing to change their rules as the IPOs of SpaceX, OpenAI and Anthropic draw closer."
Those index providers are the same interest class with VCs. With such moves they inflate demand post-IPO (hoping it holds for 180+ days), but also allows them to lure buyers in private secondary market and offload that shit pre-IPO.
https://www.wsj.com/finance/stocks/stock-indexes-are-contort...
Not to mention even a total shitshow from an obvious crackpot like Theranos got $1.2 billion total funding, and a 9B valuation. Or FTX.
In my new job I haven't written a single line by hand. I now almost entirely work in claude code / codex and in github PRs. Occasionally I open vscode to read code, but very rarely. My company probably spends ~$100 a day for my token use. I'm not even going crazy with parallelization or subagents and such.
I 100% believe the demand growth based on my personal experience.
https://www.bbc.com/news/articles/c5y4jpg922qo
Update: this inspired a note on my blog: https://simonwillison.net/2026/May/29/anthropic/
This one client, then, is 12.8% of Anthropic's run-rate revenue? That does not exactly fill me with confidence that it's a meaningful number. Doesn't this suggest run-rate revenue will fall off a cliff if Anthropic customers start applying cost controls?
As an anecdote, Github is changing their copilot plan to usage based billing next month. They released a tool that allows their users to estimate what their bills will look like under the new plan based on their past usage. There are some screenshots online from users showing their plans will go from $40/month to $3-5k/month. I imagine this is happening everywhere. These tools absolutely can do more than they were capable of just six months ago. But if the true costs are as high as it appears, folks are going to be much more judicious with how they use them moving forward.
People confuse price with cost all the time. The price of Opus has dropped from $75/1M output tokens to $25. That's the price. The cost is much lower and according to Dario, about a month ago, they had about a 73% margin.
I don't understand how anyone would use GitHub copilot...it's basically running a custom harness and using close to API pricing for Opus. This is why Microsoft is cooked in this game.
But yeah, I don't understand why people switch from subscription to API prices for Claude. They're way higher, but again that's price and not necessarily the cost to Anthropic to serve.
And FYI, most enterprise accounts were forced to switch to a hybrid monthly seat license plus API based usage earlier this year. So that is why we are probably seeing so much alarm over Ai bills at the enterprise level. Companies went whole hog on agentic workflows not fully appreciating the costs structures of their new plans. Didn't help that pretty much every VC and board was probably breathing down their neck that if they didn't jump on the AI bandwagon they would get left behind.
Do you have a source for this? I missed this and it does not match my impression.
There were a lot of people (here included) that were absolutely abusing github - getting Opus to generate its own subagents for days on end with a single premium request. If THAT'S $3k/mo, I'm honestly not worried.
I was just asking sonnet, 5.4, opus, in single agent session to fix a problem for 20 minutes...if THAT'S $3k/mo, then AI is truly cooked.
I think for your use case, the most likely outcome is that you are going to need to be on a $100-200 month plan if you want access to the cutting edge models. But on the other end of the spectrum, you could probably get closer to $10-20 month using Chinese models. My usage was closer to yours, with the occasional tokenmaxxing sprint just to experiment a bit and test out the limits of the plan. I am not quite sure what I will be doing next month once it moves to API billing but I suspect I will move to openrouter and one of the open source CLI harnesses.
If they really did count that one customer as $6B it means they've gone from $30B in April to a mere $41B in May.
im really not sure why he keeps parroting on about this. its all irrelevant frankly. companies play games. non-gaap revenue recognition, adjusted operating income.... boo-ya.
wait for the somewhat official doc's to come out, then its worth talking about.
What we are talking about is entirely based on this one term in the announcement: run-rate revenue. This is a meaningless term just like how “clinically proven” for vitamins is a way for the companies to use weasel words to imply something without truly being able to back it up, but also not actually lying. There is no legal definition of “clinically proven” so, what exactly would you sue them for? The same thing is true for run-rate revenue. They can cherry pick numbers to use to generate the run-rate revenue value and they are not lying, but this isn’t exactly honest either. We have no transparency and run-rate revenue is not an accounting term.
This goes back to my entire point, this is a vibe. This announcement does not provide much in the way of substance, so a reader will take it to say whatever they want.
IDK if these tricks would work anymore but then again fraud is legal now so who knows.
In the US?
Hahaha, that was a good one, buddy.
like when you owe a bank a billion dollars, thats the banks problem.
Hard to imagine what a world with 100GW of compute looks like.
[1] https://epochai.substack.com/p/frontier-labs-dont-use-most-a...
^^ This quotes 1.4GW at the end of 2025. Add 0.3GW at Colossus 1, and some initial fraction of 1GW Trainium2 from [2]
[2] https://www.anthropic.com/news/anthropic-amazon-compute
I think token counts and GW are a gross over simplification here. Not all tokens are the same in the amount of GPU time they consume or the size of the GPUs they require or the amount of energy they consume. There's a huge optimization potential here once these companies get serious about consolidating the business they have and executing much more efficiently. Given enough time, these companies can heavily optimize their operations. Short term growth and not slamming the brakes on that is their primary concern.
I have been trying Claude Code with DeepSeek 4 apis, and the experience is barely different. In fact the margin of error is so small that harness and prompting account for the most impact in output quality.
But, here's the catch: I spend barely more than a handful of dollars per day of regular usage. In fact DS4 via api is cheaper than Claude 100$ subscription.
I really think that very soon many will start realizing that the alternatives are extremely close in performance but dramatically different in pricing.
funny enough - those 2 might not meet. then what happens ?
Instead of ARR they should report actual revenue for once.
A report that previously took ~4 hours per month for a very expensive resource now takes 30s to validate and can be run completely ad hoc by the events managers.
I see Claude as much more of a productivity tool than say, ChatGPT, so if they were to release an image model, I would expect it to be useful for presentations etc. But, then the images need to be more like infographics, which are difficult for an image model to get right. On the other hand, they are much easier for a strong coding model to implement, while also allowing users to make controlled edits.
But then Nano Banana happened, and ChatGPT Images 2.0, and now image models aren't a toy any more - you can get real, useful work done with them.
Which is a problem for Anthropic, because companies that are buying a suite of AI tools for their employees may well value the ability to create usable posters, leaflets, and most importantly presentation slides.
Anthropic don't have anything to offer there, where Google and OpenAI have two of the best models anywhere.
I agree that $400/month is a HUGE amount, but there might be a path to that.
Seems like a lot of money until I consider how much time it would have cost me to get this far in any of them a few years ago. So long as the thing you're doing is worth doing (business, pleasure, curiosity, utility) then it's a bargain.
That's not that much money in the grand scheme of things, especially for software engineers - and you also have to include folks who pay for Claude or other AI tools on their own or for their own mundane purposes. More startups, &c. I don't use these tools much for work but I pay for a subscription and find it very valuable for my own personal uses.
> I don't know how much killing girls in Minab pays, but it looks like there is a lot of fake revenue reported here.
Well, the Iranians have us beat on that. They just use assault rifles to mow down 30,000+ of their own people: little girls, medium girls, and big girls too and we use a fancy bomb on accident to blow up a school they are launching missiles from to kill even more little girls.
ez
(not sarcastic)
It’s probably still impressive growth, but probably not as impressive as it looks.
> An AI consultant tells Axios one of their clients recently spent half a billion dollars in a single month after failing to put usage limits on Claude licenses for employees
Times that by 12 to annualize it and you get to add an extra $6 billion to the number!
Everyone is in the same bag, get it IPO’d and then let the retail investors and tracker funds worry long term
And many more that are 50% of what they were: Snowflake, Coinbase
And many more that went back to private companies and then were sold off: Carbon Black, etc...
I'm actually too lazy to go list out all of them.
But employees, beware, of those gnarly lockup periods post IPO where all the better classed options than yours get to exit.
this gives a nice comfy exit to many late-stage investors, etc.
and, of course, it's hard to say that it's great that these companies are mere shadows of themselves post-IPO, but also it's impossible to non-misleadingly assess each IPO as if they were in a vacuum.
obviously Coinbase is/was a stupid venture, but at the same time it was a pretty good bet at the time. and the same stands for a lot of these.
Ipos are somewhat notoriously risky investments.
But with OpenAI and SpaceX IPOing roughly at the same time it will likely be more than fraction of a % in this/next year.
of course, public markets nowadays are definitely paying a pretty serious "agent-principal premium". (since public exits are usually very good for the C-suite and for all those vested stocks.)
but that's the cost of access to equity (compared to PE - which nowadays underperforms public markets https://www.hamiltonlane.com/2026-market-overview/performanc... )
so yeah, it seems it would make sense to buy the post-IPO dip, but then you would need to have some kind of formula for that, and ... that seems ripe for gaming by speculators ... so all in all, it's just more efficient to do what the rule of the index says. (and of course there's already speculation at the discontinuity.)
yes, directly buying a stock at IPO sounds really strange for me. (because either you know it's undervalued, but then it's insider trading. if not, then why compete with irrational fanatics?)
Lots of professional investors are passing on the SpaceX IPO for example, which is why they had to increase the share of the retail investors.
They are dumping them on your 401k -- especially SpaceX.
So IPO is not particularly a liquidity event for investors as much as a valuation/pricing event. Indeed, the tech IPO's that have done the worst were the ones where shareholders wanted liquidity.
Clearly none of the multi-trillion dollar companies could find a buyer now if they really needed to sell themselves, so they're not really "worth" that much. (Nor are their founders, who can't sell their shares without tanking the stock.)
So these stocks are more like derivatives: a way to bet on the future where betting volume is huge relative to the underlying asset.
index funds will be the cause of the next catastrophic collapse
2. Shit goes to 0. Your 401(k) invested into S&P 500 takes a dive (dump phase)
3. Retail holding bags (full of shit) phase.
Case study: Tesla, with a P/E ratio in the hundreds along with declining sales and TAM, is a part of the S&P 500 and, consequently, of many people's 401(k)s.
And then when you sell your units, hopefully in aggregate the index is worth more than it was when you entered...
We got "dumped" Google and Facebook, so... Those probably made up for all the other "dumps".
We also got "dumped" TSLA, which is meme-ing in the trillions at the moment.
You can short Anthropic at IPO if you want...
A trillion dollar valuation seemed so hard back in the day and now there are so many companies in that list. What's the next level?
Is this just signs that $ is no longer the inflating at the same rate over time and its the realistic inflation that is reflecting in the stock market?
Prices of all goods surely has to follow to make up for the revenue needed to sustain these valuations and also the salaries to sustain the prices.
Unfortunately, those who are not in the loop is not going to have a good time.
Yeah, looking back. At the time, I distinctly remember people were going batshit over the insane FB valuation. It wasn't at all obvious it was justified.
Hindsight is 20/20.
While we could claim that such a company can grow by consuming a larger share of the GDP ... this would not bode well for future political stability, and nationalization would be a major topic.
So your left with a fast take off scenario, a job apocalypse, or a massively reduced growth rate.
If you try to configure the index-fund to avoid this problem it is not longer passively managed as each new stock needs to be evaluated in a (at least) semi-subjective manner.
IPO was in the 50B $ valuation range, and at the time, there wasn't any hint it made any financial sense.
Of course, hindsight is hindsight, but for every Facebook there's been countless IPOs of tech companies shrinking their IPO valuations by 90% in the following years.
Situations change.
Because when Facebook IPO'd everyone was saying the stock market was a dumping ground...
Same with Google...
Same with Pets.com and WebVan...
But you, of course, can buy on their IPO. They need every bagholder they can get :)
I moved all my money outside US index and global index funds because of SpaceX and OpenAI. At least until these IPOs have passed I will not move any money back. The sheer size of these IPOs might trigger a market crash.
Companies that reached a level of maturity where going public make sense don't keep doing funding rounds to cover the rate at which they bleed money.
A lot of the money that is deployed by VCs comes from pension funds and asset managers that ultimately manage money for the average Joe.
Well, I will enjoy it while it lasts.
Run-rate is taking a recent measurement and multiplying it out so it will span a year. Basically, it assumes they keep all of their current contracts, and don't gain any new ones.
Forward-looking revenue is an estimate of what the integral will be from now to a year from now.
For a growing company, run rate is between past revenue and estimates of future revenue.
Forward-looking revenue estimates are often made up from whole cloth so are highly untrustworthy. But a run-rate is saying "we've already been making this much money, we just need to maintain where we are and that's how much we'll bring in". Backward-looking revenue for something like Anthropic is meaningless because almost all of their customer base is recently acquired - they're growing like crazy, not 20% per year.
So, it's saying something but without more details it's also vague and always partially forward looking. I prefer the TTM metric (Trailing Twelve Month revenue).
Along the lines of "in the 4.75 hours starting at 02:35 yesterday we collected invoices worth $X so our run-rate revenue is now $Y"?
Occasionally it can be a snapshot if you've just completed a big contract - but it's what you expect to get per month if you're not growing or shrinking for the typical SaaS that charges per month (and assuming yearly pre-paid contracts renew etc.)
When there are interesting stories that can't be told with GAAP metrics, accountants derive new metrics. Just because they haven't gone through the standardization process yet doesn't mean they're bullshit - investors in Anthropic can hire auditors to ensure the Anthropic metrics are still meaningful. There are a very small number of deep pocketed investors in Anthropic - they're not a public company like Enron trying to sell to the WSB crowd, or like 2007 CDOs being sold to dentists.
And run rate has been a widely recognized metric for SaaS as long as it has existed - it has meaning and can be audited.
it's
in other words, "if every month was as good as this one, here's how much we'd make in a year"that means, of course, if you make $1000 in january, your RRR is $12000.
...even if you end up making $0 every other month and thus only $1000 total that year.
thats why RRR is perhaps harmful. especially when it's not growing. it can be much bigger than the actual revenue. in anthropics case it's rapidly climbing, though, so it underestimates revenue if that growth keeps up
Investors are betting real money on a payout. It seems disingenuous to think that they’re all idiots.
Things change fast in this space. Anthropic had a big boost from having the premier coding model for a while, but GPT-5.5 has closed that gap at a time when a lot of Anthropic customers are looking for cheaper alternatives.
Anthropic is coming off of a recent change to their enterprise billing that substantially changed the pricing for many users. They were smart to do the fundraising before the effects of that change could fully propagate.
My wife knows about Claude because that's what I use and we pay for. She uses it also as a result. And inevitably she will talk about Claude to her friends.
For enterprise, Anthropic is crushing it. In the manufacturing sector I anecdotally hear a 2:1 ratio of Claude to ChatGPT for teams who are settling on a platform.
Unfortunately even though we have a degree or two of seperation from most federal contracts the punitive DoD blacklisting had enough of a chilling effect on our legal team to make them drag their feet on approving any contract involving Anthropic.
So I pitched OpenAI Business with Codex so we could drop our Github Copilot Business subscription before the billing change takes effect June 1st which was approved without pushback.
I felt some responsibility for finding an immediate solution to dump Copilot since I was the one who recommended adopting Copilot in the first place, ugh... Our prices would have quadrupled based on the single month Microsoft in their beneficence allowed previewing with their tool to simulate what the post-rug pull pricing would have looked like.
Codex becoming more or less a 1:1 replacement for CC made that a no brainer given our options and the exploitative value proposition of Copilot under the new pricing model (which Microsoft evidently hoped companies like us would just accept despite being a third tier option in the dev space these days).
Absolutely not, you live in a bubble. Everybody knows about ChatGPT.
Few non-programmers have heard of anthropic or claude, nor do they care. But they all know what ChatGPT is.
Maybe Im projecting my distaste for being psychologically manipulated, but I dont think users would continue using a genAI that embeds ads directly into the response when they can just switch to gemini where they only see banners.
https://cdn.openai.com/pdf/a253471f-8260-40c6-a2cc-aa93fe9f1...
Many people I know have started simply calling it "Chat".
"I had Chat help me write this" (I didn't, I promise)
They own "chat.com", I think OpenAI should pull the trigger and move to the shorter domain and finish the rebrand.
They ran a super bowl ad. It's all over the construction industry. Claude is still not quite the Kleenex that ChatGPT is, but there is a pretty good chance lay people have heard of ChatGPT, Gemini, and Claude by now.
To disagree with the person below/above me that ChatGPT is the word used generically, when someone uses Gemini or Claude or Copilot, they TELL you which one they used, because they are essentially saying "i didnt use ChatGPT by choice."
Gemini is the one most likely to be used without people knowing which one they used.
ChatGPT is too awkward to do that with.
e.g. "I put it in chatgtp and..." when they actual asked Gemini.
Still everyone knows what I'm talking about.
The people in real life who say ChatGBT or similar are either so out of the loop the technology doesn't even matter to them, or simply are just stupid.
There is no way a person who can't get "GPT" right is even worth listening to.
Their marketing has been working the high end of the “regular people” market for a good while.
There's no better way to create awareness of a brand than to get it featured in the most popular reality TV show globally at the moment: "Thing Trump Did: Season 2."
[1] Proof: https://trends.google.com/trends/explore?date=today%205-y&q=... (see the massive spike in January of this year)
GPT-5.5 is a bit more expensive than Opus ? Current list prices
Deepseek perhaps would be the top threat on a pure price/performance metric for either of them. It doesn't look like OAI is going for the value play .https://artificialanalysis.ai/?cost=intelligence-vs-cost
Everything else is subjective to your setup, use case, configuration tuning and so forth.
More importantly bean-counters and decision makers at even 150+ seat orgs are looking at pricing sheets and enterprise contracts not how it performs for some team in a specific harness today to make million dollar annual contracts. It is not common for procurement teams to do commission the level of detailed analysis or large scale pilots that will actually hold for the duration of contract.
That doesn't mean that GPT-5.5 is selling less than Claude at all, just that cost is not the primary driver if list price is not cheaper, there is reason these are published in the same format by every vendor, because the common metric is how finance likes to compare with.
Per-token pricing is totally sensible from the provider-perspective on mapping COGS to revenue, but for a consumer, different models will produce more or less tokens, meaning the cost calculation is multi-dimensional.
Input/Cache/Output ratios are use case and configuration dependent . Any benefits in one model can usually be roughly to another with configuration tuning, and discussions devolve into subjective experience.
Pricing sheet is the objective way to compare cost.
Considering that their models are de facto extremely close in performance you would think that these arguments would've held, but they clearly don't.
I don't like Altman and I am still upset about his memory deal last year but he prepared for the current shortages months before anybody else. Meanwhile, Anthropic seems to lack any plans besides third party contracting. IMHO they got very lucky with xAI and Google having spare capacity and willing to rent it. But what about next year?
* NVidia GPU, Google TPU, Apple SoC, etc.
A short half-life means you are going to quickly dispose of what you have now, anyway. In fact most current datacenters can't even handle Vera Rubin, so I don't think there's short term risk here.
Nvidia has probably monopolized several upstream supplies to manufacture critical chip components for next 2 years, the HBMs and Optics component from LITE, as well as TSM capacity. Let alone those power components they funded themselves.
Let's say you have a genius design, but you will have it close to impossible to compete with Nvidia in getting it to volumes.
Jensen is a player, he isn't fooling around with all these Asian trips just to wine and dine
Everyone has critical risk on multiple parts of the supply chain. GPUs and Memory are just things OAI mitigated for.
Power - Bigger bottleneck than GPU or RAM perhaps, New Grid connected capacity is typically 10+ year timescale with lot of regulatory friction. Captive capacity is also quite constrained - now Gas turbines have 7+ year wait time.
There are plenty of hard constraints that OAI cannot easily solve either.
It is not clear that running one's own datacenter is a competitive advantage. Why do you think OpenAI can handle that?
Anthropics relativ longterm contract with xAI def shows that they can fill the capacity vs Musk not. OpenAI and Anthropic are both using a lot of capacity so its fair to say that this is an advantage.
If they stay very close competitive (which they are), your own datacenter does reduce token price.
I mean, this is a bit like complaining that McDonalds doesn't have their own herds of cows. OpenAI actually isn't in the business of buying GPUs or running data centres, and it's pretty weird to think that's an advantage (though it comes up constantly on here, as Anthropic keeps eating OpenAI's lunch).
There are many suppliers that are desperate to fight for Anthropics business, and it has shown an agility to embrace whatever advances in the industry come along. Anthropic is now running across a million or so Google TPUv8s, for instance. If tomorrow someone else comes out with a better GPU/TPU, they can embrace it in a heartbeat.
All while OpenAI sits on their rapidly depreciating GPUs.
Or...actually they won't, because OpenAI doesn't take business advice from HN. The vast majority of OpenAI's compute is from Microsoft, Oracle and so on. They're smart enough to not become a big hardware purchaser when that isn't their business. The core claim of your comment simply isn't true at all, nor is that the direction OpenAI is moving.
He told me they are massively pulling back on the AI stuff.
Right now the lashback is about cost, because that's the most easily measured pain point.
Soon, we'll start seeing a deeper understanding of the quality issues. At that point, it's likely this whole experiment gets firmly put in a bin of the toolbox where it belongs.
Most people don't yet have mature enough setups to fully exploit that level of use.
Why? Have we figured out the limits of what agents can do?
> OpenAI is much less exposed to tokenmaxxing
I don't think this is true, from my own experience & chatting with my acquaintances.
If a task can be completed with 100k tokens but employees are considered better performers if they complete it with 500k tokens instead… that’s unsustainable and cannot possibly benefit Anthropic in the long term.
At some point, Amazon and Uber and so on and so forth are going to realize that actually, employees using 100k tokens or even 50k tokens is better than 500k and Anthropic’s revenue will fall off a cliff.
And I could be wrong about tokenmaxxing being a Claude specific problem but as far as I can tell, all of the major companies encouraging employees to maximize their token usage are Claude Code users. And the music has to stop on that at some point, whether because the companies run out of money or because they learn better ways of measuring productivity in the AI age. And if tokenmaxxing is what is driving Anthropic’s lead in revenue, it could be catastrophic to lose that, because Anthropic are spending billions of dollars per month on the infrastructure to support it.
If tokenmaxxing is evenly distributed between Anthropic and OpenAI then they’ll both hurt but equal hurt shouldn’t disadvantage either much.
I see most of the surge here comes FOMO AI spending which will have to be dialed down later half of the year, otherwise those companies will have to layoff to fund their AI bill, which is harmful to their business.
Anthropic grabs its bag at the peak, but feast is over.
I think Google has caught up enough to certainly be a player in the consumer ad driven market.
I also don't think only one foundation model adds up. Now that the trail is blazed a dozen companies can likely make a good enough model. The question is if there's a moat to make it winner take all
I don't think SOTA-wise Google has a lot of catch up to do.
Especially if you assume that 6 months ago they weren't very close to this version of profitable
They don't seem too far off to me.
Well functioning market is supposed to have many, as in a lot, companies with similar products. To create competition.
People have been going to Google for a quarter of a century, and I imagine they will continue to do so.
As far as I am concerned, I moved to DDG in 2013 and never looked back, but I find Google AI intriguing.
How the hell do you crush a ~1T company on the one thing they have all their focus on?
Both Anthropic and OpenAI only has access to whatever they can buy or steal.
And it's becoming increasingly hard to get fresh uncontaminated data for training. No amount of money can buy that.
Claude Code and Codex are a big advantage, vs Gemini CLI which might be killedbygoogle soon? https://news.ycombinator.com/item?id=48196867
> Both Anthropic and OpenAI only has access to whatever they can buy or steal. A trillion can buy you quite a lot! Like offer some company a ton of money for data, and if they say no simply buy said company. Bonus points if it's someone like Atlassian who's stock price is getting hammered largely because of you.
Me, not a SWE, I have never even tried Claude
(and I am not sure I prefer ChatGPT over Google)
I suspect we'll have our first $10T company in the next 2-3 years. That's only doubling.
So you're assuming lots of risk and putting it all in the same basket.
There's no shame in getting 100k $ worth of stock, selling it and putting it on some vanguard fund and diversifying, in fact it's statistically the best move you can do.
Of course, you can be like those many googlers that did this and then regret in hindsight.
And Saudi Aramco before they IPO'd
Now, headlines are only about hundred of billions. I do not know what to think about that, apart from the fact that I wish that we were putting that money to enhance human lives in general. Of course, people will say that these tools will help humans in the future, but 1) at what costs, and 2) I would prefer, I don't know, bridges or infrastructure, or free healthcare, or food for everyone.
I do care about: how useful their products are vs. cost and how secure are their businesses. Actually I only care about the first thing since these services are hot swap-able with some effort.
Having been through an IPO before, it was good for employee liquidity, but bad for the culture and long-term success of the company.
The judgment is subjective though, so pushing the boundaries could be a calculated risk.
You're assuming private liquidity to be infinite and private credit (that fuels VCs) to always have favorable rates.
https://www.investor.gov/introduction-investing/investing-ba...
https://www.law.cornell.edu/wex/tender_offer
https://carta.com/learn/equity/liquidity-events/tender-offer...
https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...
(secondary markets are sometimes an option, depending on stock transfer restrictions)
FTX bought 8% of Anthropic for $500m in 2021.
https://www.forbes.com/sites/josipamajic/2026/03/18/ftx-owne...
The price was determined by a formula based on revenue and such, so I always knew what they were worth.
I was not allowed to sell to anyone else though.
I also imagine that venture funding rounds have a lower ceiling than the public markets - but at these rounds I'm not so sure!
α β γ δ ε ζ η θ ι κ λ μ ν ξ ο π ρ σ τ υ φ χ ψ ω
They cannot raise forever, SpaceX has done more rounds but the timing is most important.
Until they IPO and the investors make their money, who knows what is behind the revenue stated
Databricks raised an L round last year, so they'll have to solve it first.
https://support.microsoft.com/en-us/office/excel-specificati...
[0] - https://learn.microsoft.com/en-us/answers/questions/4888467/...
Hynix is participating with a new circular deal. Hynix is also valued at $1 trillion now, which is positively insane.
This scam will implode harder that the housing bubble.
They're the belle of the ball right now, everybody is talking about them, everybody wants to invest in them, so they can call the shots.
Then they'll have money in the bank for a long time no matter what happens – IPO, market downturn, etc.
Takes off lots of pressure so they can continue focusing on the product.
https://huggingface.co/Anthropic
I feel AI is a bit different, as in there is a spectrum ranging from “utterly stupid” (early ChatGPT), “very helpful” (kinda now), “SkyNet 2.0” (what good are humans!).
As algorithms and technology improve, AI will be both cheaper and more capable. Companies like Anthrophic wouldn’t be the vaulted celebrities as they are today. At that point, I’m not sure what value much of society can provide…
At one time, blacksmiths had a valued place in general society. Today, not so much…
/s
Without more information, this number is impossible to interpret.
This really doesn't paint the good picture you seem to imply.
I can have $47B in revenue if I sell something that cost $80B to produce ez pz.
like how 13^2=169 and 31^2=961 or 10^2+11^2+12^2=13^2+14^2