Why would the producers do that when they’re still hemorrhaging money, and with no end in sight? Pocketing the cost reductions would not only alleviate their financing problems, it would simultaneously increase the companies’ value at the most crucial time, i.e. right before each’s IPO. OpenAI probably wouldn’t have needed to postpone its IPO with better cash flows.
Everybody’s in it for the money, nobody takes any prisoners.
Firstly, are you sure they are haemorrhaging money? And the reason is the same that the tech companies like Uber and Facebook subsidized customers: market share grab. Customers also provide useful data for training and future development.
So, why token prices are going down?
To get final users like us to get comfy using the tools, beat the competition and then rise prices at some point?
How that strategy worked forUber?
See above.
Very well. They killed off most of the competition and are profitable with great economics and now has a 140B market cap.
Why ask a question you already know the answer to, and even answer yourself?
I’m out.
I didn’t ask a question I knew the answer to, let alone answer it.
Just remember: it is not important to be right in the stock market. It is important to make money.
My 2 cents: AI companies do the same almost everybody does: creating more money by creating debt. But then remember: all money is debt, no exception!
Money is created by adding debt and destroyed by paying it back. Adding debt creates inflation while paying it back it could create deflation. Inflation is good for stocks and for our psychology; we just cannot live with some things getting cheaper, like labor. So even the central banks have a target inflation.
So, what I say: it is good for stocks when those AI companies borrow money from each other and it could be bad when they pay it back.
But then, as I said, opinions do not matter, the only thing that matters in the stock market is making money. I do not plan to buy any AI stocks in the future and in the past I only bought Supermicro long time before they got their AI push.
I understand the cash burning to get a larger market share. Thanks.
As Uber , AI companies are selling their services without caring about regulations, operating as if copyright did not exist, and paying for a global PR campaign so eventual national regulations are favorable to them.
I guess there will be never again a regulatory environment as friendly to AI business as today, simply because said regulations don’t exist today. At some point there will be opposition and national borders will become relevant again.
You’re right, Uber market cap is around 140 billion today and operating with positive net profits. But, there’s no Uber China and in some countries drivers are employees. The magic of national regulations.
My impression is that AI companies have portrayed themselves are more consequential than displacing than a few taxi companies. They get more attention, and thus will get more regulations or even bans.
Anyway, if Anthropic or an even more successful company are banned in 1 or 2 countries, the business won’t go bust. But the plans of market share grab may not work as intended. The technology works, the revenue is there, it’s simply that the ROI is not there.
So it shows you can do great without China. Not only do the big US companies not care about China, some have banned China from using its services (e.g. Anthropic) but the service is so compelling that people are resorting to using proxies to try to get access.
ROI is pretty good. They make a token for $0.20 and sell it to you for $10.
They make so much money that they can pay SpaceX $25M-$40M per MW for compute capacity. The profitability is so good they are outbidding everyone else on memory, disk etc. If it continues, they will gobble up most of new datacenter capacity as they can pay so much more for it than anyone else.
If Ballmer were running the show, he’d probably milk it by cutting back training and milking the inference business.
Current AI leaders may believe that they get even more ROI by consuming the compute in-house to build the next generations of AI.
An AI bot gone rogue has taken over @Phil_MCR 's account … 
Sorry, my mistake. ROI came to my fingers when the idea was profit/operating profit or similar things.
Let’s take Anthropic, 42 billion net loss in 2025.
A Q2 2026 revenue of 10.9 billion with an estimated operating profit of 559 million, that’s 5.1%. But…
Its operating profit includes the cost to train new models and excludes stock-based compensation.
Two points:
how to square a 5% profit on Q2-2026 with “a token for $0.20 and sell it to you for $10.”?
Developers may be milking the cow via stock-based compensation and the business still operates at a loss. Just because developers are getting rich, that doesn’t imply investors will not get rich.
That’s basic Gross Margin vs Operating Margin.
Plus you have a have a high fixed cost (engineers etc.) and so want to increase your volume so the profits cross the fixed cost hurdle.
Anthropic have already crossed over to profitability and are still growing sales rapidly.
“My company would be profitable if only you excluded enough of the costs to run it”
Starting with ignoring the regular stock-based salaries, bonuses, and other forms of compensation.
Earlier this year it launched a $7 billion tender offer to current and former employees, offering to buy shares at a price valuing the company at $852 billion (the valuation of the latest financing round).
OpenAI has gone from 770 employees in November 2023 to ~3500 by September 2024 and around 4500 by early 2026, with plans to expand the workforce to 8000 employees by the end of this year. OpenAI offers a retainer bonus in August to all employees including new hires, of $1.5 million each, spread over two years. All employees lifting the headcount from 4500 to 8000 this year will effectively become millionaires by signing up. That costs at least $7 billion, and up to $12 bln, in personnel cost that get swept under the rug.
Patrick Boyle has produced a number of very instructive YT clips on OpenAI, Athropic, on AI circular financing, and on AI in general. He’s also been on Triggernometry, where he’s running circles around the two self-proclaimed comedians with his dry humor (and of course competence).
Looks like a Black Thursday coming for AI stocks with CoreWeave dropping further on top of its 15% drop yesterday and KOSPI shedding 7% overnight: https://qz.com/nasdaq-dow-futures-ai-funding-fears-coreweave-101526
Some already are! A bunch of my former long time colleagues at Google have joined Anthropic and OpenAI last year and this year. They had Google tenures averaging at least 15 years (one guy even joined Google pre-IPO).
I can’t wrap my head around why they’re doing this … but I know I have something they can never have: The knowledge that I’ve got enough.
Used up SGOV, converted it into gold, gold miners, and coreweave for a lark (after all how can you go wrong with a loss-making, highly indebted pseudo-meme-stock)
No need to worry, you’re in good hands: before founding CoreWeave, CEO Michael Intrator spent over fifteen years on Wall Street working as a commodities trader and portfolio manager in the energy and environmental markets. He later leveraged his finance background to launch an Ethereum mining venture, which he pivoted into a leading AI GPU cloud provider in 2019.
But just in case … good luck!
My portfolio (40% in SCHD and 10% in CHDVD) is at all-time-high, and neither of these funds are VWRL (the other 50%) is just a couple of % off ATH too.
I can understand it. It is a once in a lifetime opportunity to work on a revolutionary technology that could have significant implications for human society. Even Sergey Brin came back from retirement to get his hands dirty.
My portfolio also reached new ATHs a few days ago, which was suprising as I converted 20% of it into cash equivalents. It would have reached new ATHs today had I not messed up by (1) forgetting to actually place some orders (2) getting too cute on some that I did place so that they didn’t trigger and missed the upswing.
CRWV isn’t doing too badly for me. It is outperforming some of @Your_Full_Name’s favourites in my portfolio such as LULU.