JP Morgan weist auf KI-Ausgaben hin und sagt, dass ein Jahresumsatz von 650 Milliarden US-Dollar erforderlich sei, um nur 10 % Rendite auf den KI-Ausbau zu erzielen – das entspricht einer Zahlung von 35 US-Dollar für jeden iPhone-Benutzer oder 180 US-Dollar für jeden Netflix-Abonnenten „auf Dauer“.
https://www.tomshardware.com/tech-industry/artificial-intelligence/usd650-billion-in-annual-revenue-required-to-deliver-10-percent-return-on-ai-buildout-investment-j-p-morgan-claims-equivalent-to-usd35-payment-from-every-iphone-user-or-usd180-from-every-netflix-subscriber-in-perpetuity
25 Kommentare
Imagine spending that much money on helping society instead, and to help upskill current and emerging workforce.
But instead we get greedy companies burning through billions for a small chance of laying off workers and having all their work be automated.
$35 or $180…..a month
Been saying it for a while. AI takes 10 units of energy to produce 1 unit of output. It’s horribly inefficient, which has led to a tech company circular dependency pumping up that bubble.
1M poured into AI = 100k output roi = cut that 100k /year worker for that roi
AWS chief Garman doesn‘t seem to think there is a bubble. Interesting though between the lines is a recent interview with him in German newspaper Handelsblatt.
https://www.handelsblatt.com/technik/it-internet/wir-wuerden-ihm-gern-helfen-amazons-cloud-chef-spottet-ueber-microsoft/100171072.html
here are a few translated paragraphs:
Matt Garman, head of the world’s largest cloud provider Amazon Web Services (AWS), counters fears of a growing speculative bubble around artificial intelligence (AI) in an interview with Handelsblatt. “We are seeing huge demand. And that will continue for the foreseeable future,” said Garman. He does not consider this development “a bubble.”
As the market leader, AWS is the main driver of an unprecedented expansion of IT infrastructure, the financial dimension of which has recently continued to grow. Economists now see the more than $400 billion hardware battle as the main reason for the US economy’s growth.
…
Amazon is spending $125 billion this year – primarily on AI chips and data centers. This is roughly equivalent to AWS’s annual revenue and far exceeds what other tech companies are spending.
Critics fear that the enormous investments may not pay off. An analysis by the Handelsblatt Research Institute recently identified a negative trend in the ratio of investments to cash flow, particularly at Amazon: While investments are expected to increase by 45 percent in 2025, cash flow will decline by 32 percent.
Garman is promising even higher spending in 2026. Given the strong demand, this is only logical, he said. He wants to further expand the dominant position of his company, whose infrastructure enables internet services such as Netflix, Snapchat, Airbnb, and the networking of BMW vehicles.
His company would now gain revenues on the scale of a Fortune 500 company every year, the manager said. The Fortune index comprises the 500 largest publicly traded companies in the US. He also announced new hires of young university graduates – despite the recently announced job cuts. “It makes no sense,” Garman said, “to cut off the talent pipeline.” … Microsoft CEO Nadella, who recently admitted that he was unable to connect “a bunch of AI chips” due to a lack of electricity, is met by Garman with ridicule (“For us, that would be a huge planning mistake”): “We would be happy to help him if he needs support with supply chain planning.” Leaving these chips lying around is “very expensive,” Garman said. “That’s lost revenue that you can’t get back.”
Is that why all the Coreweave, Nebius, Rigetti stock price been plummeting recently?
Maybe they should use ai to design more cost effective ai models because those prices are not reasonable
Sure, but that is only at the current cost of AI infrastructure, because it is still young. In the future it will cost more.
Ai is a tool. Not a solution. The sooner the higher ups realize this, the better we’re all going to do.
Ai is an excavator in a land where everyone has shovels. You’ve still got to learn how to properly use the damn thing and what situations to actually use it in. You’re not going to need an excavator to dig a 12in hole for your kids dead hamster.
And another brick is pulled from the wall.
It’s only a matter of time until it crumbles.
But once we get every company to layoff their junior staff, we can just keep doubling the subscription price and they’ll just have to pay.
Pop
Happen soon please
Epic FOMO here
Well yeah, the “income” now is nearly all based on deals with other AI/tech companies. It’s fake money being passed around. When the consumer base doesn’t materialize to justify the spending, it’s all going to fall
Hold on to your butts.
Anyone who’s not an idiot knew the math NEVER made sense.
It’s all propped up on hope, stock-buy backs, round-tripping, and bullshit.
For what? A glorified encyclopedia? Lol. Im going back to motorola flip this is horseshit.
I am starting to think that AI might destroy society, or at least change a dramatically. A lot of people will be left behind. But the cat is out of the bag now and these oligarchs got nothing else better to do or spend on so they’re going to strive for AGI. At any cost. I did make a lot of money off of tech stocks the last few years though.
Let’s burst that bubble already, it’s been crazy and is allowed to get even crazier yet.
The tech companies are worth trillions: let them blow up some few hundred billions over AI experiments.. even if it fails that may result in some value after a decade.. like it Happened during dotcom.
Let’s not fret over their AI spend, in fact use the systems freely while the party is on. Soon there’d be Chinese sponsored LLMs and AI programs.
In other words, J.P. Morgan is telling the investors they’re lighting money on fire and getting scammed by basically the same people who were selling NFTs a few years ago.
This bubble will end in tears unless we have some kind of once every few centuries breakthrough in hardware that makes it magnitudes better. Statistical regurgitation engines isn’t particularly useful, generically. They might be for _very_ narrow usecases, but that’s not how they’re marketed or used in the vast majority of cases.
Once someone decently big runs out of cash to burn, the vicious circle of domino effects will start. Hold on to your butts, this ride’ll get bumpy.
I’m assuming J.P Morgan cashed out in their long positions and are looking to cash out on their shorts.
Uh yeah. This is maybe the most impactful technological race the world has ever seen
Ai is already permiating a large portion of your life without knowing it. Sure, the most obvious ones are chat gpt and the Ai Wendy’s ordering menu but a lot of the companies you rely on also use them, datacenters, streaming services, games, etc all leverage them. You can’t really valuate these things from a consumer spend perspective and why the market is as bullish as they are. Also, these companies spending billions aren’t dumb and the average redditor commenting about it is not smarter than the ivy league educated people working at those companies, valuing these companies, etc.