Good news! Having reviewed and updated my household finances, I'm now personally making 16.8 billion a quarter after certain adjustments. Free money, please.
The company also said it was profitable after making certain adjustments,
...
And it said its gross margin, or revenue minus direct costs associated with generating that revenue, was 85 percent after making certain adjustments.
Good news! Having reviewed and updated my household finances, I'm now personally making 16.8 billion a quarter after certain adjustments. Free money, please.
I'll concede that the revenue is realer than I thought it would be. That framing of "after making certain adjustments" is quite possibly the most sus statement to accompany it though. My belief...
I'll concede that the revenue is realer than I thought it would be.
That framing of "after making certain adjustments" is quite possibly the most sus statement to accompany it though. My belief that this is going to collapse horrifically is unchanged.
Kinda like when our president promptly declared he's the most healthy president of all time instead of just "doing fine" when asked about his health.
To be fair, despite all available verifiable evidence suggesting the president should be in very poor health, he's still somehow kicking. Though as a media man I suppose a fair amount of his time...
To be fair, despite all available verifiable evidence suggesting the president should be in very poor health, he's still somehow kicking. Though as a media man I suppose a fair amount of his time in public could be smoke, mirrors, and A LOT of drugs, and he could be in kind of rough shape in private. It's amazing what having access to the absolute best medical resources in the world can do. Heck, as far as I can tell even the spirit of Mitch McConnell was dragged out of hell and forced back into his body by the best medical treatment money can buy.
**I want to be clear that I'm not trying to say this as a pure joke. There's kind of a morbid joke here, but also my very real reflection that even when it seems like something should be on the verge of collapse, it might be able to hang on for a lot longer than anyone could have expected. And for some of these AI companies, I suspect a few of them genuinely might "fake-it-till-they-make-it" long enough to actually become stable companies.
If pointing out that they're not getting these numbers using GAAP ages poorly, I will be very impressed. Maybe they're doing just as well as they would like people to think they're doing. But if...
If pointing out that they're not getting these numbers using GAAP ages poorly, I will be very impressed. Maybe they're doing just as well as they would like people to think they're doing. But if they were, would they need to be adjusting their numbers?
When they release more complete accounting, prior to their IPO, if it invalidates these numbers that they gave to investors... that would be bad for them. Is the implication that they're that stupid?
When they release more complete accounting, prior to their IPO, if it invalidates these numbers that they gave to investors... that would be bad for them.
My belief is that they're building hype. If their more complete accounting has more realistic performance numbers, but their previous hype got people excited, the realistic numbers won't be an...
My belief is that they're building hype. If their more complete accounting has more realistic performance numbers, but their previous hype got people excited, the realistic numbers won't be an invalidation, just... a different set of adjustments. And aren't you so excited for the opportunity to buy in advance so you can make money off the retail investors?
I don't think it's entirely bullshit, just... adjustments.
Meh, every large tech company I've ever looked makes adjustments in their reports. GAAP doesn't fit very well with what these companies do. Investors are aware that adjustments are made; they are...
Meh, every large tech company I've ever looked makes adjustments in their reports. GAAP doesn't fit very well with what these companies do. Investors are aware that adjustments are made; they are interested in different things and these presentations cater to the metrics they actually care about. I don't think it's likely that Anthropic would lie given that if they are angling for that IPO, all their numbers will come out shortly anyway.
Also, Simon Willison comments on a quote from the Financial Times:
Also, Simon Willison comments on a quote from the Financial Times:
Anthropic told investors on Wednesday distillation accounted for 5 per cent to 7 per cent of revenue and that they had cut it back “materially” in the third quarter of the year, said four people with direct knowledge of the matter.
Napkin maths says that if distillation was 5% of their Q2 revenue (before they cut it back in Q3) then unnamed "other labs" spent over half a billion dollars distilling Anthropic's models in that quarter!
The S-1 isn't out yet so this isn't the formal "road show" for the IPO. Apparently meeting with investors before filing is called "testing the waters." But it seems unlikely that the numbers...
The S-1 isn't out yet so this isn't the formal "road show" for the IPO. Apparently meeting with investors before filing is called "testing the waters." But it seems unlikely that the numbers they're giving to investors would be much different from the numbers that show up in the S-1.
I updated my chart to have a new tab for quarterly revenue. Also, there's a tab comparing the actual quarterly numbers to estimates based on previously leaked run rates. Looks pretty close! (The chart is AI-generated.)
So, despite some skepticism from cynics, the leaked run rates don't seem to have been misleading after all? Everything's consistent.
I think the cynics desperately want this to be a bubble, likely for a variety of reasons. I am not sure. Revenue is already this high, and climbing, and they haven't found consumer product market...
I think the cynics desperately want this to be a bubble, likely for a variety of reasons. I am not sure. Revenue is already this high, and climbing, and they haven't found consumer product market fit yet. Once they do, the sky could really be the limit.
I think the cynics are certain this is a bubble, as bubble economics have basically dictated my entire life. DotCom, SubPrime, AI. The market has already pretty much come around to the idea that...
I think the cynics are certain this is a bubble, as bubble economics have basically dictated my entire life. DotCom, SubPrime, AI. The market has already pretty much come around to the idea that Oracle may well be doomed because of it... a scenario that was utterly laughable 10 years ago.
And while I can't speak for others, I'm personally hoping for the bubble to pop before any significant portion of the datacenters come online and really start ruining our lives in all sorts of ways.
Well, sure, Oracle does look shaky. But I’m guessing most if not all of the big tech firms were founded during your lifetime. If you remember the the dot-com era, you should remember that Amazon...
Well, sure, Oracle does look shaky. But I’m guessing most if not all of the big tech firms were founded during your lifetime. If you remember the the dot-com era, you should remember that Amazon was founded then. Do you think you would recognize another one if you saw it?
With the revenue numbers they have and alleged profitability, Anthropic’s trajectory looks quite a bit like the rise of another big tech firm to me, for better or worse.
I think a key distinguishing factor is that 80%+ of Amazon's customers were not primarily the dotcom companies that failed, which is very much the case for Anthropic and OpenAI. Imagine if you...
I think a key distinguishing factor is that 80%+ of Amazon's customers were not primarily the dotcom companies that failed, which is very much the case for Anthropic and OpenAI.
Imagine if you will that instead of being an online bookstore with a solid business model, Amazon was only AWS in 1998. And that 80% of their total revenue was the vast majority of dotcom businesses that didn't have a sustainable business model. Do you think that version of Amazon would have survived?
Most of the names we remember today are the ones that survived which were either already profitable or very close. Autotrader comes to mind. But of the ones that were airing superbowl ads in 2000? Less than half survived. And even the companies that survived had a 90% drop in valuation back to reality...Amazon included. It went from $107 to $7. Didn't recover till 2007.
If all of the AI margins are in API, what happens when all those heavily subsidized companies reselling that API usage at a loss right now go poof? Are the current $200/mo subscribers going to pony up $12,000/mo instead to make up that difference?
So yes, I think I'm reasonably concerned that between Anthropic and OpenAI IPOs they're looking to hit valuations of $3T+, and when the bubble pops and the actual valuation of $300 billion between the both of them, wiping out $3,700 billion of our 401ks.
That’s the wrong way to look at it. Your 401k would inflate by $3,700 billion and then deflate by $3,700 billion. If you look at the SP500 over the dot com bubble, it more or less was on the...
and when the bubble pops and the actual valuation of $300 billion between the both of them, wiping out $3,700 billion of our 401ks.
That’s the wrong way to look at it. Your 401k would inflate by $3,700 billion and then deflate by $3,700 billion.
If you look at the SP500 over the dot com bubble, it more or less was on the trajectory it was on before the bubble by 2004. Things go up, things go down.
Which would be fine if it then didn't go down again in like crazy a few years later around 2008-2009. My 401k didn't return to its 2008 levels until 2015 (admittedly, I seemed to get hit harder...
Which would be fine if it then didn't go down again in like crazy a few years later around 2008-2009. My 401k didn't return to its 2008 levels until 2015 (admittedly, I seemed to get hit harder than most did).
Correct me if I'm wrong, but isn't my 401k (well not mine, I've moved almost all out of tech indexes) going to be buying in at that $3T+ initial valuation? And if that valuation goes pop, the...
Correct me if I'm wrong, but isn't my 401k (well not mine, I've moved almost all out of tech indexes) going to be buying in at that $3T+ initial valuation? And if that valuation goes pop, the amount they invested into it goes down by 90%?
We're getting in at the top of the rollercoaster. Not the bottom.
No, you actually can’t buy stocks from an IPO. Only institutional investors are allowed to buy IPO stocks. Some brokerages let you “buy” IPOs but what’s actually happening is that the brokerage is...
No, you actually can’t buy stocks from an IPO. Only institutional investors are allowed to buy IPO stocks. Some brokerages let you “buy” IPOs but what’s actually happening is that the brokerage is buying the stock as an institutional investor and then you buy an IOU from them.
Either way your average 401k will have no direct exposure to either company at IPO, and would only get exposure more than a year later if they sustain their market caps. But even then, the only way your 401k goes down by 90% would be if your 401k is entirely OpenAI equity. Which would be on you tbh. Index funds operate on weights, the proportion of an SP500 ETF that would own either company, assuming they even make it to the SP500, would be very small.
Also you’re definitely not getting in at the top. Remember that the reason why the stock market has been doing well is because of the (alleged) AI bubble. You have indirect exposure to the benefits of AI valuations through the Mag7.
I did amend while you were writing that it was just the portion invested, but point taken. I thought my 401k was managed by an institutional investor, and thus while indirect my pool of money is...
I did amend while you were writing that it was just the portion invested, but point taken. I thought my 401k was managed by an institutional investor, and thus while indirect my pool of money is very much part of that stake.
I diversified to indexes that weigh tech exponentially less than most others do. Because I think we're very close to peak NVIDIA, and the collateral damage, even though I do genuinely doubt 90% of full market, is likely to be at least 30% overall, and much higher in tech.
That's kind of the point, though. The market has gained TREMENDOUSLY from the AI bubble. The average American 401k has went up significantly because they had exposure to companies like Nvidia...
Because I think we're very close to peak NVIDIA, and the collateral damage, even though I do genuinely doubt 90% of full market, is likely to be at least 30% overall, and much higher in tech
That's kind of the point, though. The market has gained TREMENDOUSLY from the AI bubble. The average American 401k has went up significantly because they had exposure to companies like Nvidia 5x-ing their valuations.
So it will go down when the bubble pops. But you can't measure the loss from the top down. You have to measure it from the start of the bubble to the end of the bubble.
Bubbles happen, they happen all the time in the end. For something like retirement investments, just don't think about it that much. Keep investing into diversified assets, and trust the average.
The problem being, that the vast majority of my contributions come during bubble rises. So 80%+ of my 401k is my contributions at peak bubble growth, courtesy of when my career started. Then I get...
The problem being, that the vast majority of my contributions come during bubble rises.
So 80%+ of my 401k is my contributions at peak bubble growth, courtesy of when my career started. Then I get to spend another 10 years getting it back to just the money I put in, maybe keeping up with inflation.
So yea, I'm hoping to avoid that pitfall this time. At least now the market has already universally confirmed we're in a bubble so getting off now is likely safer than trying to ride it up a few more years.
You've only been contributing in the last 2-3 years? I mean, that's the importance of long time horizons. A hypothetical 401k that has been dollar-cost average invested into over 20 years has...
The problem being, that the vast majority of my contributions come during bubble rises.
You've only been contributing in the last 2-3 years? I mean, that's the importance of long time horizons. A hypothetical 401k that has been dollar-cost average invested into over 20 years has ridden its ups and its downs and is in a very good state averageout.
It's not a pitfall, it's just part of life. Trying to predict the bubble on your own is just a recipe for disaster. Just DCA, and let time do its thing.
Not just AI bubble. Started in 2008....not a great year for getting a career off the ground. However, courtesy of extremely low paychecks during the recession, the first 6 years amounted to a...
Not just AI bubble. Started in 2008....not a great year for getting a career off the ground. However, courtesy of extremely low paychecks during the recession, the first 6 years amounted to a pittance, as pretty much every penny earned needed to pay for regular life.
The next job finally started paying reasonably, so after 4 years of getting the rest of finances back to a stable state (eliminate all the debt from the broke years, get emergency savings, etc), it's now 2016 and I can finally start contributing more than the bare minimum to get match. So yes, almost all of my significant contributions have been during the rise of the NVIDIA bubble fueled by crypto and AI.
It doesn’t sound like it. ChatGPT launched in the end of 2022 and really the nvidia valuation skyrocketed occurred more towards the end of 2023. So it’s been a little over 2 years. If you’ve been...
it's now 2016 and I can finally start contributing more than the bare minimum to get match. So yes, almost all of my significant contributions have been during the rise of the NVIDIA bubble fueled by crypto and AI.
It doesn’t sound like it. ChatGPT launched in the end of 2022 and really the nvidia valuation skyrocketed occurred more towards the end of 2023. So it’s been a little over 2 years.
If you’ve been investing since 2016 then you’ve been investing in a “normal” market for 7-8 years and a bubble for 2 years.
I sounds like you think that Anthropic's customers are other AI companies, or something. I'm wondering what evidence there is for that? I did find this Ramp article which suggests that AI spend...
I sounds like you think that Anthropic's customers are other AI companies, or something. I'm wondering what evidence there is for that?
I did find this Ramp article which suggests that AI spend per employee and overall token spend may have peaked.
They also write:
80% of spend on OpenAI and Anthropic is driven by 1% of firms.
They do have a much wider customer base (Anthropic has ~40% of all firms in Ramp's data) but the whales are apparently spending a lot more.
The idea that we've hit the start of an S curve on spending already is a terrifying thought, which would suggest that peak AI spend seems to be on the order of $200B or less, spread among all the...
The idea that we've hit the start of an S curve on spending already is a terrifying thought, which would suggest that peak AI spend seems to be on the order of $200B or less, spread among all the players.
Which I presume will mostly boil down to the big players gathering up the remains of OpenAI and Anthropic in a few years.
Well, trend lines aren't a surefire way to predict the future anyway. There's a learning curve that could go either way: People figure out how to stop wasting money on AI People figure out new...
Well, trend lines aren't a surefire way to predict the future anyway. There's a learning curve that could go either way:
People figure out how to stop wasting money on AI
People figure out new ways to use AI
Probably both at once, and it's hard to say which will win out. But I think we're only at the beginning of figuring out how to use AI. It's going to take years. Some uses might not be discovered until after AI improves some more.
You don't have to be a cynic to recognize a financial bubble. Much like the dotcom bubble, the product is real, but the finances around it are massively inflated. In this case, one of the clear...
You don't have to be a cynic to recognize a financial bubble. Much like the dotcom bubble, the product is real, but the finances around it are massively inflated. In this case, one of the clear risks is that there is a lot of money going around in circles.
If you're not convinced, listen to someone with a vested interest like Bezos, he called it a bubble.
Currently it seems like the subscription plans for Claude are a better deal than the api pricing for open models, but this suggests they are heavily subsidised. It’s a common tech accounting trick...
Currently it seems like the subscription plans for Claude are a better deal than the api pricing for open models, but this suggests they are heavily subsidised.
It’s a common tech accounting trick to sell $10 notes for $5 and report massive growth, but when they have to actually make a profit they will be competing with the fact Chinese models do mostly the same thing for little more than the cost to run them.
Alternatively, subscription pricing might be sold at close to cost and the "API" rate might have massive profit margins? We'd need to know more about their actual costs to know which scenario is true.
Alternatively, subscription pricing might be sold at close to cost and the "API" rate might have massive profit margins? We'd need to know more about their actual costs to know which scenario is true.
I can tell you that a company (Yardi) is offering the Claude agent (under the name Virtuoso) and is charging $100,000 per account for access to this AI service. My wife works for a company that...
I can tell you that a company (Yardi) is offering the Claude agent (under the name Virtuoso) and is charging $100,000 per account for access to this AI service. My wife works for a company that uses Yardi and I just accompanied her on a big Yardi event (I didn't have access to the classes obviously but I am my wife's sounding board so I know what's going to the extent I understand it). And now she has had follow up meetings with her company about this and the only way they could see justifying buying into AI would be to get rid of employees and they are VERY against that. They also use a consultant that also doesn't know of a single company that uses Yardi and has purchased the AI suite. I can also say that big event was basically a huge AI marketing campaign. So if company's are spending Millions trying to get other companies to spend hundreds of thousands but no one is really biting, I dunno. Doesn't look great for Anthropic. Which I'm glad for. I'm an anti-AI guy. I have a huge problem with the environmental impacts of data-centers and of the theft needed for generative AI.
That sounds bad for Yardi, but I don't see what that has to do with Anthropic, which actually is widely used. (I've never heard of Yardi, but apparently they are "a complete AI platform for real...
That sounds bad for Yardi, but I don't see what that has to do with Anthropic, which actually is widely used.
(I've never heard of Yardi, but apparently they are "a complete AI platform for real estate.")
The "AI" part of that tagline is new. Yardi is one of the bigger names in real estate management software. No reason you'd know some unrelated industry's software, but they're not a small name.
The "AI" part of that tagline is new. Yardi is one of the bigger names in real estate management software. No reason you'd know some unrelated industry's software, but they're not a small name.
Every ERP/CRM system tacked on AI due to hype. These companies will all be fine because they were already immensely profitable, and the AI was a bolt-on that almost nobody wanted, and most of us...
Every ERP/CRM system tacked on AI due to hype. These companies will all be fine because they were already immensely profitable, and the AI was a bolt-on that almost nobody wanted, and most of us will be thrilled if they get rid of it.
Anthropic executives told a roomful of investors on Wednesday that the company’s revenue for the third quarter jumped about 45 percent from the prior quarter to $16.8 billion, as it pursues what may become the largest-ever initial public offering, two people with knowledge of the company’s presentation said.
[...]
In the meeting with potential investors at its headquarters, Anthropic sought to generate enthusiasm for the upcoming deal. Revenue was up from $11.6 billion in the second quarter. The company also said it was profitable after making certain adjustments, which is not a bottom-line number. And it said its gross margin, or revenue minus direct costs associated with generating that revenue, was 85 percent after making certain adjustments.
[...]
Anthropic, which filed to go public in June, has not yet made its prospectus — which lays out its financial performance — public. Wednesday was the first official opportunity for potential I.P.O. investors to get a glimpse into the numbers, which may give insight into the health of the broader A.I. industry and will help determine the valuation Anthropic will get in its listing.
Good news! Having reviewed and updated my household finances, I'm now personally making 16.8 billion a quarter after certain adjustments. Free money, please.
I'll concede that the revenue is realer than I thought it would be.
That framing of "after making certain adjustments" is quite possibly the most sus statement to accompany it though. My belief that this is going to collapse horrifically is unchanged.
Kinda like when our president promptly declared he's the most healthy president of all time instead of just "doing fine" when asked about his health.
To be fair, despite all available verifiable evidence suggesting the president should be in very poor health, he's still somehow kicking. Though as a media man I suppose a fair amount of his time in public could be smoke, mirrors, and A LOT of drugs, and he could be in kind of rough shape in private. It's amazing what having access to the absolute best medical resources in the world can do. Heck, as far as I can tell even the spirit of Mitch McConnell was dragged out of hell and forced back into his body by the best medical treatment money can buy.
**I want to be clear that I'm not trying to say this as a pure joke. There's kind of a morbid joke here, but also my very real reflection that even when it seems like something should be on the verge of collapse, it might be able to hang on for a lot longer than anyone could have expected. And for some of these AI companies, I suspect a few of them genuinely might "fake-it-till-they-make-it" long enough to actually become stable companies.
I suspect this is going to age as poorly as the idea that "run-rate revenue" is suspicious.
If pointing out that they're not getting these numbers using GAAP ages poorly, I will be very impressed. Maybe they're doing just as well as they would like people to think they're doing. But if they were, would they need to be adjusting their numbers?
When they release more complete accounting, prior to their IPO, if it invalidates these numbers that they gave to investors... that would be bad for them.
Is the implication that they're that stupid?
My belief is that they're building hype. If their more complete accounting has more realistic performance numbers, but their previous hype got people excited, the realistic numbers won't be an invalidation, just... a different set of adjustments. And aren't you so excited for the opportunity to buy in advance so you can make money off the retail investors?
I don't think it's entirely bullshit, just... adjustments.
Meh, every large tech company I've ever looked makes adjustments in their reports. GAAP doesn't fit very well with what these companies do. Investors are aware that adjustments are made; they are interested in different things and these presentations cater to the metrics they actually care about. I don't think it's likely that Anthropic would lie given that if they are angling for that IPO, all their numbers will come out shortly anyway.
Also, Simon Willison comments on a quote from the Financial Times:
The S-1 isn't out yet so this isn't the formal "road show" for the IPO. Apparently meeting with investors before filing is called "testing the waters." But it seems unlikely that the numbers they're giving to investors would be much different from the numbers that show up in the S-1.
I updated my chart to have a new tab for quarterly revenue. Also, there's a tab comparing the actual quarterly numbers to estimates based on previously leaked run rates. Looks pretty close! (The chart is AI-generated.)
So, despite some skepticism from cynics, the leaked run rates don't seem to have been misleading after all? Everything's consistent.
I think the cynics desperately want this to be a bubble, likely for a variety of reasons. I am not sure. Revenue is already this high, and climbing, and they haven't found consumer product market fit yet. Once they do, the sky could really be the limit.
I think the cynics are certain this is a bubble, as bubble economics have basically dictated my entire life. DotCom, SubPrime, AI. The market has already pretty much come around to the idea that Oracle may well be doomed because of it... a scenario that was utterly laughable 10 years ago.
And while I can't speak for others, I'm personally hoping for the bubble to pop before any significant portion of the datacenters come online and really start ruining our lives in all sorts of ways.
Well, sure, Oracle does look shaky. But I’m guessing most if not all of the big tech firms were founded during your lifetime. If you remember the the dot-com era, you should remember that Amazon was founded then. Do you think you would recognize another one if you saw it?
With the revenue numbers they have and alleged profitability, Anthropic’s trajectory looks quite a bit like the rise of another big tech firm to me, for better or worse.
I think a key distinguishing factor is that 80%+ of Amazon's customers were not primarily the dotcom companies that failed, which is very much the case for Anthropic and OpenAI.
Imagine if you will that instead of being an online bookstore with a solid business model, Amazon was only AWS in 1998. And that 80% of their total revenue was the vast majority of dotcom businesses that didn't have a sustainable business model. Do you think that version of Amazon would have survived?
Most of the names we remember today are the ones that survived which were either already profitable or very close. Autotrader comes to mind. But of the ones that were airing superbowl ads in 2000? Less than half survived. And even the companies that survived had a 90% drop in valuation back to reality...Amazon included. It went from $107 to $7. Didn't recover till 2007.
If all of the AI margins are in API, what happens when all those heavily subsidized companies reselling that API usage at a loss right now go poof? Are the current $200/mo subscribers going to pony up $12,000/mo instead to make up that difference?
So yes, I think I'm reasonably concerned that between Anthropic and OpenAI IPOs they're looking to hit valuations of $3T+, and when the bubble pops and the actual valuation of $300 billion between the both of them, wiping out $3,700 billion of our 401ks.
That’s the wrong way to look at it. Your 401k would inflate by $3,700 billion and then deflate by $3,700 billion.
If you look at the SP500 over the dot com bubble, it more or less was on the trajectory it was on before the bubble by 2004. Things go up, things go down.
Which would be fine if it then didn't go down again in like crazy a few years later around 2008-2009. My 401k didn't return to its 2008 levels until 2015 (admittedly, I seemed to get hit harder than most did).
Correct me if I'm wrong, but isn't my 401k (well not mine, I've moved almost all out of tech indexes) going to be buying in at that $3T+ initial valuation? And if that valuation goes pop, the amount they invested into it goes down by 90%?
We're getting in at the top of the rollercoaster. Not the bottom.
No, you actually can’t buy stocks from an IPO. Only institutional investors are allowed to buy IPO stocks. Some brokerages let you “buy” IPOs but what’s actually happening is that the brokerage is buying the stock as an institutional investor and then you buy an IOU from them.
Either way your average 401k will have no direct exposure to either company at IPO, and would only get exposure more than a year later if they sustain their market caps. But even then, the only way your 401k goes down by 90% would be if your 401k is entirely OpenAI equity. Which would be on you tbh. Index funds operate on weights, the proportion of an SP500 ETF that would own either company, assuming they even make it to the SP500, would be very small.
Also you’re definitely not getting in at the top. Remember that the reason why the stock market has been doing well is because of the (alleged) AI bubble. You have indirect exposure to the benefits of AI valuations through the Mag7.
I did amend while you were writing that it was just the portion invested, but point taken. I thought my 401k was managed by an institutional investor, and thus while indirect my pool of money is very much part of that stake.
I diversified to indexes that weigh tech exponentially less than most others do. Because I think we're very close to peak NVIDIA, and the collateral damage, even though I do genuinely doubt 90% of full market, is likely to be at least 30% overall, and much higher in tech.
That's kind of the point, though. The market has gained TREMENDOUSLY from the AI bubble. The average American 401k has went up significantly because they had exposure to companies like Nvidia 5x-ing their valuations.
So it will go down when the bubble pops. But you can't measure the loss from the top down. You have to measure it from the start of the bubble to the end of the bubble.
Bubbles happen, they happen all the time in the end. For something like retirement investments, just don't think about it that much. Keep investing into diversified assets, and trust the average.
The problem being, that the vast majority of my contributions come during bubble rises.
So 80%+ of my 401k is my contributions at peak bubble growth, courtesy of when my career started. Then I get to spend another 10 years getting it back to just the money I put in, maybe keeping up with inflation.
So yea, I'm hoping to avoid that pitfall this time. At least now the market has already universally confirmed we're in a bubble so getting off now is likely safer than trying to ride it up a few more years.
You've only been contributing in the last 2-3 years? I mean, that's the importance of long time horizons. A hypothetical 401k that has been dollar-cost average invested into over 20 years has ridden its ups and its downs and is in a very good state averageout.
It's not a pitfall, it's just part of life. Trying to predict the bubble on your own is just a recipe for disaster. Just DCA, and let time do its thing.
Not just AI bubble. Started in 2008....not a great year for getting a career off the ground. However, courtesy of extremely low paychecks during the recession, the first 6 years amounted to a pittance, as pretty much every penny earned needed to pay for regular life.
The next job finally started paying reasonably, so after 4 years of getting the rest of finances back to a stable state (eliminate all the debt from the broke years, get emergency savings, etc), it's now 2016 and I can finally start contributing more than the bare minimum to get match. So yes, almost all of my significant contributions have been during the rise of the NVIDIA bubble fueled by crypto and AI.
It doesn’t sound like it. ChatGPT launched in the end of 2022 and really the nvidia valuation skyrocketed occurred more towards the end of 2023. So it’s been a little over 2 years.
If you’ve been investing since 2016 then you’ve been investing in a “normal” market for 7-8 years and a bubble for 2 years.
I sounds like you think that Anthropic's customers are other AI companies, or something. I'm wondering what evidence there is for that?
I did find this Ramp article which suggests that AI spend per employee and overall token spend may have peaked.
They also write:
They do have a much wider customer base (Anthropic has ~40% of all firms in Ramp's data) but the whales are apparently spending a lot more.
The idea that we've hit the start of an S curve on spending already is a terrifying thought, which would suggest that peak AI spend seems to be on the order of $200B or less, spread among all the players.
Which I presume will mostly boil down to the big players gathering up the remains of OpenAI and Anthropic in a few years.
Well, trend lines aren't a surefire way to predict the future anyway. There's a learning curve that could go either way:
Probably both at once, and it's hard to say which will win out. But I think we're only at the beginning of figuring out how to use AI. It's going to take years. Some uses might not be discovered until after AI improves some more.
You don't have to be a cynic to recognize a financial bubble. Much like the dotcom bubble, the product is real, but the finances around it are massively inflated. In this case, one of the clear risks is that there is a lot of money going around in circles.
If you're not convinced, listen to someone with a vested interest like Bezos, he called it a bubble.
Currently it seems like the subscription plans for Claude are a better deal than the api pricing for open models, but this suggests they are heavily subsidised.
It’s a common tech accounting trick to sell $10 notes for $5 and report massive growth, but when they have to actually make a profit they will be competing with the fact Chinese models do mostly the same thing for little more than the cost to run them.
Alternatively, subscription pricing might be sold at close to cost and the "API" rate might have massive profit margins? We'd need to know more about their actual costs to know which scenario is true.
I can tell you that a company (Yardi) is offering the Claude agent (under the name Virtuoso) and is charging $100,000 per account for access to this AI service. My wife works for a company that uses Yardi and I just accompanied her on a big Yardi event (I didn't have access to the classes obviously but I am my wife's sounding board so I know what's going to the extent I understand it). And now she has had follow up meetings with her company about this and the only way they could see justifying buying into AI would be to get rid of employees and they are VERY against that. They also use a consultant that also doesn't know of a single company that uses Yardi and has purchased the AI suite. I can also say that big event was basically a huge AI marketing campaign. So if company's are spending Millions trying to get other companies to spend hundreds of thousands but no one is really biting, I dunno. Doesn't look great for Anthropic. Which I'm glad for. I'm an anti-AI guy. I have a huge problem with the environmental impacts of data-centers and of the theft needed for generative AI.
That sounds bad for Yardi, but I don't see what that has to do with Anthropic, which actually is widely used.
(I've never heard of Yardi, but apparently they are "a complete AI platform for real estate.")
The "AI" part of that tagline is new. Yardi is one of the bigger names in real estate management software. No reason you'd know some unrelated industry's software, but they're not a small name.
Every ERP/CRM system tacked on AI due to hype. These companies will all be fine because they were already immensely profitable, and the AI was a bolt-on that almost nobody wanted, and most of us will be thrilled if they get rid of it.
From the article:
[...]
[...]