Now you can keep track of how many billions the AI companies are losing on AI.
-
Now you can keep track of how many billions the AI companies are losing on AI. (Red is spending, green is revenue.) https://isaiprofitable.com/
@MikeElgan I like how literally everyone's broke except NVidia
This is so stark lmao
-
Now you can keep track of how many billions the AI companies are losing on AI. (Red is spending, green is revenue.) https://isaiprofitable.com/
@MikeElgan hii
-
Now you can keep track of how many billions the AI companies are losing on AI. (Red is spending, green is revenue.) https://isaiprofitable.com/
@MikeElgan that definitely looks sustainable
-
Now you can keep track of how many billions the AI companies are losing on AI. (Red is spending, green is revenue.) https://isaiprofitable.com/
@MikeElgan hii
-
Now you can keep track of how many billions the AI companies are losing on AI. (Red is spending, green is revenue.) https://isaiprofitable.com/
@MikeElgan This even seems to be a favourable framing of the numbers... the circular financing is mentioned, but, from the numbers, OpenAI looks like it's almost making some money....
Until you remember that microsoft has been subsidizing the hell out of them, so a lot of their costs are probably hidden in the microsoft spend bar, rather than their own... -
Now you can keep track of how many billions the AI companies are losing on AI. (Red is spending, green is revenue.) https://isaiprofitable.com/
@MikeElgan its like that time i spend time learning how to drive a car. My instructor made me drive in circles ending my tour where i started. Valuable time i could have spent walking and cycling and actually getting somewhere. Not to mention the costs per hour.
-
Now you can keep track of how many billions the AI companies are losing on AI. (Red is spending, green is revenue.) https://isaiprofitable.com/
@MikeElgan this is insane, how are they eating all these losses?!
-
Now you can keep track of how many billions the AI companies are losing on AI. (Red is spending, green is revenue.) https://isaiprofitable.com/
@MikeElgan Looking at the whole chart, #Nvidia is at the bottom, and is the only one making any money, and it's making buckets of money.
"When there's a gold rush, sell shovels."
-
Now you can keep track of how many billions the AI companies are losing on AI. (Red is spending, green is revenue.) https://isaiprofitable.com/
@MikeElgan Oh to be the guy selling the picks and shovels to these idiots.
-
@sebastian @cohentheblue @julesbl @PaulaToThePeople @MikeElgan At the very least we should encourage individuals to have their own domains and publish their own content. There's no need for fancy websites, either. Simple text-based websites with meaningful content are the last holdout of a useful internet.
@KennedyRichard @sebastian @julesbl @PaulaToThePeople @MikeElgan I have a web site. I don't pay for it, it's not on my personal domain but it's my site, easy to port elsewhere if needed. Jekyll generated static html based on markdown source. Link in my mastodon profile.
There's plenty of personal web sites out there with excellent content. I have read lots of it. Most of what is worth a read is similar generated static html with minimal design.
There are striking, simple, elegant designs.
-
Now you can keep track of how many billions the AI companies are losing on AI. (Red is spending, green is revenue.) https://isaiprofitable.com/
@MikeElgan
Disgusting amount of money.
And it's killing the entire PC industry. -
@MikeElgan I like how literally everyone's broke except NVidia
This is so stark lmao
@zkat @MikeElgan this list needs the memory companies
-
Now you can keep track of how many billions the AI companies are losing on AI. (Red is spending, green is revenue.) https://isaiprofitable.com/
@MikeElgan This looks a bit distorted in places to make the argument. Unless you need material to fume about the AI industry, use better sources. The economic insanity is bad enough.
-
Now you can keep track of how many billions the AI companies are losing on AI. (Red is spending, green is revenue.) https://isaiprofitable.com/
@MikeElgan So everyone is trying to figure out the scheme Nvidia has. Got it.
-
@MikeElgan And exactly how many trillion valuation did they add to the their Big Tech sugar daddies?
Let's be honest, last time I checked that was some time ago, alone the difference in MSFT valuation between "growth" and "mature" company P/E was over 2 trillion USD.
That makes the burnt capex look like good PR costs, wouldn't you agree?
@MikeElgan Or to put it differently, the graph is nice. Beside the aspect that the "spent capex" in most rows is "estimated," the primary issue is that it's a way to narrow perspective.
Let's skip the aspect that there are really useful uses for AI, but let's ignore them.
The whole AI hype (epicenter USA) is primarily a market manipulation. Market as in securities, but secondarily they are manipulating anything that is convenient.
E.g. the OpenAI LoI con job was not only a brilliant securities -
@MikeElgan Or to put it differently, the graph is nice. Beside the aspect that the "spent capex" in most rows is "estimated," the primary issue is that it's a way to narrow perspective.
Let's skip the aspect that there are really useful uses for AI, but let's ignore them.
The whole AI hype (epicenter USA) is primarily a market manipulation. Market as in securities, but secondarily they are manipulating anything that is convenient.
E.g. the OpenAI LoI con job was not only a brilliant securitiesmanipulation for the involved companies (OpenAI, Samsung, Hynix if memory serves), it also led to an artificial shortage and allowed Samsung & Hynix to raise their memory prices over 100% in a concerted action.
(And no, OpenAI if they wanted to reserve an allocation "RAM for AI", the natural way would have been to sign purchase order with NVidia, not a non-binding LoI with the RAM makes for unfinished wafers in unspecified amounts at unspecified times …) -
manipulation for the involved companies (OpenAI, Samsung, Hynix if memory serves), it also led to an artificial shortage and allowed Samsung & Hynix to raise their memory prices over 100% in a concerted action.
(And no, OpenAI if they wanted to reserve an allocation "RAM for AI", the natural way would have been to sign purchase order with NVidia, not a non-binding LoI with the RAM makes for unfinished wafers in unspecified amounts at unspecified times …)It raises the market valuation for all the sugar daddies, and that is literally like printing money for these companies. Companies with such high, unnatural P/E ratios tend to use their stock to pay for anything they buy, be it companies they acquire, highly sought-after employees, …
Mature companies cannot compete; they have to pay cash.
-
Now you can keep track of how many billions the AI companies are losing on AI. (Red is spending, green is revenue.) https://isaiprofitable.com/
-
@MikeElgan So everyone is trying to figure out the scheme Nvidia has. Got it.
@mrgrumpymonkey @MikeElgan
Nvidia got lucky (most business succeeds by luck not judgement IMO!), as did their investor Softbank who cashed out when the price plateaued in Nov '25.This is interesting on where they invested next (TL;DR - they are attempting to own the whole supply chain, based on much cheaper silicon tech):
https://dataconomy.com/2025/11/11/the-why-behind-softbank-nvidia-stake-sale/Once the first grifters leave, the bubble's days are numbered I rekon (I hope!).
-
@KennedyRichard @sebastian @julesbl @PaulaToThePeople @MikeElgan I have a web site. I don't pay for it, it's not on my personal domain but it's my site, easy to port elsewhere if needed. Jekyll generated static html based on markdown source. Link in my mastodon profile.
There's plenty of personal web sites out there with excellent content. I have read lots of it. Most of what is worth a read is similar generated static html with minimal design.
There are striking, simple, elegant designs.
@cohentheblue @sebastian @julesbl @PaulaToThePeople @MikeElgan That is great. I'll bookmark your site when I'm on my PC (I'm on phone right now).
Funny that a lot of people are using static website generators these days. I myself use one as well. Didn't actually find one to my liking but managed to make a simple one w/ Python and the Python Markdown lib.
It seems everyone reaches that moment in life where they just want noise-free work, no distractions, no bs. Just write, generate, publish.
