Thank you for the article, but I think it might be helpful to review a couple of financial/accounting concepts for future analysis (not trying to be sarcastic at all!). All of those figures quoted include non-cash expenses like SBC, D&A, and one-time charges, which are usually embedded within those operating expense line items. The people stating the true losses are $8bn are likely referencing a cash flow metric that unburdens profitability of those non-cash charges. Calling it deception isn't quite accurate; within an investment firm, analysts strip these out to determine true steady-state cash generation or burn (done for a variety of reasons like true cash runway, SBC being non-cash - equity dilution problem and not cash problem, etc.). If you’re looking for a "no bullsh*t" figure for sustainability, industry standard dictates using UFCF or LFCF, not GAAP operating loss. Also, the characterizations of R&D and G&A are highly oversimplified, the R&D breakout for OpenAI likely carries much more depth than just training costs (I would posit it would be difficult to even say that is the majority of R&D cost). Thanks again for the article.
If the employees weren't getting stock they wouldn't work for free. Those are REAL expenses, disguised as "non-cash" and therefore less relevant to operating profitability.
Morgan's comments about potential one-time charges in the expense lines are fair though.
$5.7 billion on Sales & Marketing is where this piece stops being about AI and starts being about unit economics. Companies with genuine product-market fit don't outspend NASA on ads. They don't need to. The product sells itself because the value is obvious at the price being charged. $5.7 billion tells you one of two things: either the price needs to come down to match actual willingness to pay, or the demand isn't as deep as the TAM slides suggest. OpenAI is now reportedly considering cutting prices, which answers which one it is.
The Anthropic "profitability" is the other number worth pausing on. Profitable for a two-month window when SpaceX compute costs were discounted, after which the bill balloons to $15 billion a year. That's not a profit. That's a coupon. And the WSJ language around it, "about to," "on track," "first profitable quarter," is doing the same conditional heavy lifting that "we're almost there" has done in every pre-IPO narrative that didn't survive contact with actual financial reporting. The most important person at these labs is the CFO. The problem is that nobody is listening to the CFO yet.
Well spacex and tesla have been massive capex projects for years and they’re worth more than multiple ai companies combined. I agree the ad spend is insane
If OpenAI or Anthropic had a moat, the way that Facebook and Amazon did when they were IPO I'd see the path. What I see instead is Gemini, OpenSource, Chinese Models and SpaceX all running on the same track. Ultimately it seems this is going to become a commodity play from where I'm sitting. Fun to watch!
Stupidest technology I have ever seen, propped up by the stupidest business model I have ever seen. But those billions in marketing surely worked at making CEOs believe that they can slash their workforce and have all the work done by magical chat bots.
Oh, and I forgot that once you probe deep enough to get at the heart of the stupidity, it's really all about tech bros honestly believing they have created a Digital God and just need to feed it more information to become omnipotent. It's absolutely delusional.
The bit that jolted me is OpenAI's marketing expenditures. I thought only people and governments could have cluster B personality disorder. I just updated my knowledge of our world. Nowadays companies can get on the spectrum, too.
Prof G Markets are not talking heads. In the podcast Ed made a short references to personalities, too.
The "Sales and Marketing" figure is not just insanely high, it's clearly not just "ads" because if it were then we'd be confronted by those ads everywhere we go. So what is it? Other Ed (Zitron) reckons it includes the cost of all those subsidized subscriptions. Sounds cynical, but if I were trying to hide true inference costs, I know what I'd do..
Also, hinted at in the article (but not clarified) is the significance of that ARR claim from Anthropc. It's a 12x multiple of revenues from a cherry-picked month. You know, the one, when tokenmaxxing was still a thing and subsidies were keeping subscription prices at a fraction of their true costs. Good luck replicating that 11 more times, Dario
This is the result of businesses using AI without a clear intent, direction, or business identity. They just follow a trend, a trend that now costs them literally. As any problem, this one comes with a solution as well.
Language, behavior, identity…and proper AI calibration, this is where focus should be.
Good analysis, Ed. How many years of legitimate net profits at a reasonable rate will it take to "pay back" the billions that have already been set ablaze? When will the companies finally make one dollar more than has been put in?
Value the stocks that way, and they're are largely worthless. This is "number go up" speculating.
"Anthropic’s SpaceX contract (one of its largest expenditures) was heavily discounted (for reasons unclear)" The reasons are because Elon is petty and wants to help Dario with his pre-IPO narrative because they're both still mad at Sam for, in their words, stealing OpenAI. The Real Housewives of Generative AI.
Early SaaS burned cash the same way, cheap enterprise contracts, freemium for adoption, ugly unit economics. Then the pricing architecture matured: tiered enterprise plans, premium subscriptions, usage-based billing. Margins exploded.
AI labs are in that same early phase. If Anthropic or OpenAI can replace entire SaaS categories instead of just augmenting them, they don't just capture that revenue, they capture it at higher margins with better pricing leverage. The path exists. The question is whether they survive long enough to walk it.
I have people manage my money. That said. Ai is definitely going to create tremendous wealth and is solving many challenges in all fields.
Freedom
Freedom is the fuel that ignites Creativity & Compassion.
NOT Socialism NOT Communism & Definitely NOT Religious Theocracies! You may be hell bent on “Hate” because you “Don’t Like” Trump or simply brainwashed & a parrot because you refuse to study history… unfortunately you are enslaved in your own foolishness. #Peace comes after #Freedom No political/religious dogma kills a desire for FREEDOM. After WW2 133 out of 200 Nations left monarchies, dictatorships & theocracies to become Democratic in some form. The TREND worldwide is STILL towards forms of Democracy and that is FACTUAL! Just because Communists, Socialists, Dictatorships FLOOD social media/media with PROPAGANDA & fools repeat it (mainly for clicks & fake popularity), does NOT mean it is a trend.
We produce #music & #videos in various languages & genres.
Another thing no one seems to know or has the answer to is "what is the return on tokens?"
The shift from tokenmaxxing to tokenminimising within companies & the rise of open source models are direct questions to the profitability of LLMs at scale. Anthropic has a chance to become profitable but that hinges on crazy revenue growth whilst managing compute costs. Don't really see it with OpenAI in my view.
What I believe to be the future is specialised edge models as opposed to a GPT.
Thank you for the article, but I think it might be helpful to review a couple of financial/accounting concepts for future analysis (not trying to be sarcastic at all!). All of those figures quoted include non-cash expenses like SBC, D&A, and one-time charges, which are usually embedded within those operating expense line items. The people stating the true losses are $8bn are likely referencing a cash flow metric that unburdens profitability of those non-cash charges. Calling it deception isn't quite accurate; within an investment firm, analysts strip these out to determine true steady-state cash generation or burn (done for a variety of reasons like true cash runway, SBC being non-cash - equity dilution problem and not cash problem, etc.). If you’re looking for a "no bullsh*t" figure for sustainability, industry standard dictates using UFCF or LFCF, not GAAP operating loss. Also, the characterizations of R&D and G&A are highly oversimplified, the R&D breakout for OpenAI likely carries much more depth than just training costs (I would posit it would be difficult to even say that is the majority of R&D cost). Thanks again for the article.
If the employees weren't getting stock they wouldn't work for free. Those are REAL expenses, disguised as "non-cash" and therefore less relevant to operating profitability.
Morgan's comments about potential one-time charges in the expense lines are fair though.
$5.7 billion on Sales & Marketing is where this piece stops being about AI and starts being about unit economics. Companies with genuine product-market fit don't outspend NASA on ads. They don't need to. The product sells itself because the value is obvious at the price being charged. $5.7 billion tells you one of two things: either the price needs to come down to match actual willingness to pay, or the demand isn't as deep as the TAM slides suggest. OpenAI is now reportedly considering cutting prices, which answers which one it is.
The Anthropic "profitability" is the other number worth pausing on. Profitable for a two-month window when SpaceX compute costs were discounted, after which the bill balloons to $15 billion a year. That's not a profit. That's a coupon. And the WSJ language around it, "about to," "on track," "first profitable quarter," is doing the same conditional heavy lifting that "we're almost there" has done in every pre-IPO narrative that didn't survive contact with actual financial reporting. The most important person at these labs is the CFO. The problem is that nobody is listening to the CFO yet.
Well spacex and tesla have been massive capex projects for years and they’re worth more than multiple ai companies combined. I agree the ad spend is insane
If OpenAI or Anthropic had a moat, the way that Facebook and Amazon did when they were IPO I'd see the path. What I see instead is Gemini, OpenSource, Chinese Models and SpaceX all running on the same track. Ultimately it seems this is going to become a commodity play from where I'm sitting. Fun to watch!
I’ve been paying attention for a while, and I gotta say, on all fronts, AI just looks kinda… stupid.
Stupidest technology I have ever seen, propped up by the stupidest business model I have ever seen. But those billions in marketing surely worked at making CEOs believe that they can slash their workforce and have all the work done by magical chat bots.
Oh, and I forgot that once you probe deep enough to get at the heart of the stupidity, it's really all about tech bros honestly believing they have created a Digital God and just need to feed it more information to become omnipotent. It's absolutely delusional.
I like how they decided they had to start making stuff up to feed the models when they had consumed all the actual information available.
The bit that jolted me is OpenAI's marketing expenditures. I thought only people and governments could have cluster B personality disorder. I just updated my knowledge of our world. Nowadays companies can get on the spectrum, too.
Prof G Markets are not talking heads. In the podcast Ed made a short references to personalities, too.
The "Sales and Marketing" figure is not just insanely high, it's clearly not just "ads" because if it were then we'd be confronted by those ads everywhere we go. So what is it? Other Ed (Zitron) reckons it includes the cost of all those subsidized subscriptions. Sounds cynical, but if I were trying to hide true inference costs, I know what I'd do..
Also, hinted at in the article (but not clarified) is the significance of that ARR claim from Anthropc. It's a 12x multiple of revenues from a cherry-picked month. You know, the one, when tokenmaxxing was still a thing and subsidies were keeping subscription prices at a fraction of their true costs. Good luck replicating that 11 more times, Dario
This is the result of businesses using AI without a clear intent, direction, or business identity. They just follow a trend, a trend that now costs them literally. As any problem, this one comes with a solution as well.
Language, behavior, identity…and proper AI calibration, this is where focus should be.
Good analysis, Ed. How many years of legitimate net profits at a reasonable rate will it take to "pay back" the billions that have already been set ablaze? When will the companies finally make one dollar more than has been put in?
Value the stocks that way, and they're are largely worthless. This is "number go up" speculating.
"Anthropic’s SpaceX contract (one of its largest expenditures) was heavily discounted (for reasons unclear)" The reasons are because Elon is petty and wants to help Dario with his pre-IPO narrative because they're both still mad at Sam for, in their words, stealing OpenAI. The Real Housewives of Generative AI.
I don’t know if you can isolate the Google Gemini sub-business financials, but it would round out your analysis of the three big applied AI companies.
People were paying for it when they had unlimited accounts now that token base billing is in the jig is up.
Early SaaS burned cash the same way, cheap enterprise contracts, freemium for adoption, ugly unit economics. Then the pricing architecture matured: tiered enterprise plans, premium subscriptions, usage-based billing. Margins exploded.
AI labs are in that same early phase. If Anthropic or OpenAI can replace entire SaaS categories instead of just augmenting them, they don't just capture that revenue, they capture it at higher margins with better pricing leverage. The path exists. The question is whether they survive long enough to walk it.
I have people manage my money. That said. Ai is definitely going to create tremendous wealth and is solving many challenges in all fields.
Freedom
Freedom is the fuel that ignites Creativity & Compassion.
NOT Socialism NOT Communism & Definitely NOT Religious Theocracies! You may be hell bent on “Hate” because you “Don’t Like” Trump or simply brainwashed & a parrot because you refuse to study history… unfortunately you are enslaved in your own foolishness. #Peace comes after #Freedom No political/religious dogma kills a desire for FREEDOM. After WW2 133 out of 200 Nations left monarchies, dictatorships & theocracies to become Democratic in some form. The TREND worldwide is STILL towards forms of Democracy and that is FACTUAL! Just because Communists, Socialists, Dictatorships FLOOD social media/media with PROPAGANDA & fools repeat it (mainly for clicks & fake popularity), does NOT mean it is a trend.
We produce #music & #videos in various languages & genres.
https://open.spotify.com/playlist/37i9dQZF1DZ06evO3fX5Gb
There are some very funny comparisons in this article that are likely truer than they are satirical https://www.mcsweeneys.net/articles/ai-economics-for-dummies
What if AI told them to do it 😆 Joking, but not really https://nicspallitta.substack.com/p/the-machines-told-us-we-were-right?r=8krnih&utm_medium=ios
Another thing no one seems to know or has the answer to is "what is the return on tokens?"
The shift from tokenmaxxing to tokenminimising within companies & the rise of open source models are direct questions to the profitability of LLMs at scale. Anthropic has a chance to become profitable but that hinges on crazy revenue growth whilst managing compute costs. Don't really see it with OpenAI in my view.
What I believe to be the future is specialised edge models as opposed to a GPT.