this post was submitted on 18 May 2026
223 points (96.3% liked)

Fuck AI

7070 readers
1734 users here now

"We did it, Patrick! We made a technological breakthrough!"

A place for all those who loathe AI to discuss things, post articles, and ridicule the AI hype. Proud supporter of working people. And proud booer of SXSW 2024.

AI, in this case, refers to LLMs, GPT technology, and anything listed as "AI" meant to increase market valuations.

founded 2 years ago
MODERATORS
VerbFlow@lemmy.world
MrMcGasion@lemmy.world
TootSweet@lemmy.world
BigMikeInAustin@lemmy.world
cynar@lemmy.world
drmeanfeel@lemmy.world
pavnilschanda@lemmy.world
CriticalMedicine@lemmy.world
WonderfulWanderer@lemmy.world
Communist@lemmy.ml
eatCasserole@lemmy.world
SpaceNoodle@lemmy.world
NutWrench@lemmy.world
Soup@lemmy.cafe
iAvicenna@lemmy.world
Tinks@lemmy.world
wizblizz@lemmy.world
corus_kt@lemmy.world
Prandom_returns@lemm.ee
JimSamtanko@lemm.ee
TrickDacy@lemmy.world
TheFriar@lemm.ee
ArmokGoB@lemmy.dbzer0.com
HawlSera@lemm.ee
andrew_bidlaw@sh.itjust.works
MeDuViNoX@sh.itjust.works
33550336@lemmy.world
Nougat@fedia.io
Lost_My_Mind@lemmy.world
Quill7513@slrpnk.net
glowing_hans@sopuli.xyz
e8d79@discuss.tchncs.de
ThefuzzyFurryComrade@pawb.social
223
The Time Bomb Went Off: AI's All-You-Can-Eat Era Just Ended in Real Time (www.thestateofbrand.com)
submitted 1 day ago by brianpeiris@lemmy.ca to c/fuck_ai@lemmy.world
48 comments fedilink hide all child comments
 

Excerpt:

The IPO Math Forces the Issue

Both OpenAI and Anthropic are on IPO timelines for the second half of 2026. OpenAI completed the largest private funding round in history in April, $122 billion at an $852 billion post-money valuation. Anthropic has reportedly surpassed $30 billion in annualized revenue. Massive numbers, both of them. Also both attached to companies that are still burning cash at extraordinary rates.

Public markets will not tolerate the gap between subscription revenue and compute cost that has defined the past three years. The moment either company files, analysts will demand unit economics that show a path to margin. Usage-based billing is the fastest way to demonstrate that path.

None of this contradicts the repricing thesis. The pricing war is the last land grab before the gate closes. Both companies are spending aggressively now to lock in users whose switching costs will make them sticky when prices rise. OpenAI offers two months free. Anthropic offers 50% more capacity. Both expire in July. What comes after July is the real pricing.

top 48 comments
sorted by: hot top controversial new old
[–] rekabis@lemmy.ca 28 points 16 hours ago (1 children)

So first they get devs hooked on AI.

Then they watch and wait as dev skills degrade. Like, the actual skills to get the work done without AI.

Then, once devs are unable to be productive without AI anymore, they turn down the screws, pulling profit from those who cannot do without anymore.

It’s a tactic of capitalism that is as old as time. But I think it’s been done too quickly this time, as there are still plenty of devs which have sat on the sidelines, waiting for things to shake out, and who haven’t had their skills erode away from AI usage because they just haven’t been relying on it or even using it.

[–] Yeller_king@reddthat.com 6 points 12 hours ago (1 children)

Problem is there is tons of competition and anyone with a video card can run models on their PC for the cost of electricity.

[–] axexrx@lemmy.world 6 points 9 hours ago

Thats why theyre trying to buy up all the hardware-

[–] quick_snail@feddit.nl 9 points 18 hours ago (2 children)

My first thought is that companies that built dependence on AI workflows will just start hiring humans again.

But I think it's more likely they'll just switch to free, self-hosted AI models.

[–] ICastFist@programming.dev 7 points 17 hours ago (1 children)

But I think it’s more likely they’ll just switch to free, self-hosted AI models.

Implying anyone left there will even have the know-how to set one up

[–] quick_snail@feddit.nl 1 points 9 hours ago* (last edited 9 hours ago)

Wow, you just convinced me to learn how to do something with AI.

It'll be a very useful skill to help companies migrate from third party to self hosted AI in the next decade

[–] Cityshrimp@lemmy.world 5 points 17 hours ago (1 children)

Still not “free” even if self-hosted and I would think perf and accuracy would be noticeably worse.

My company started using AI in coding process a lot more now but its only because the AI responses are useful 80% of the time. If its lower than that, not worth the time to use for most tasks.

[–] quick_snail@feddit.nl 1 points 8 hours ago

I don't really think it matters. Companies that have adopted AI are already fucked. It hurts more then helps.

But they'll be willing to thow VC money at hiring people to setup AI systems, despite how much worse it leaves them

[–] whalebiologist@lemmy.world 9 points 19 hours ago* (last edited 19 hours ago)

This is article slop, please don't make me read things that were written by bots, have no authors, feature ai generated images, citing themselves as a source... People would happily Ai-generate anti-ai blog posts and sell them to us all day long if we let them.

This stuff is from press releases, primary sources are not hard to find.

[–] zeroConnection@programming.dev 8 points 19 hours ago* (last edited 19 hours ago) (1 children)

There's a mass move to Chinese models now, which are like 10 times cheaper already, and will probably be about 50 times cheaper when the new pricing kicks in for Claude and Codex.

Can't wait for the American AI companies to go bust.

[–] Buddahriffic@lemmy.world 2 points 15 hours ago

Not to mention China could subsidise them just to pressure the US companies even more.

I wonder if they are timing it for the price increase to let them stick to the promotions that cost them more before pulling the rug. Could even be a win to open source them if the US government tries to block it.

[–] melsaskca@lemmy.ca 4 points 20 hours ago

This reminds me of the "pet rock" craze in the early 70's. Everyone as getting them. They were everywhere. Myself, and a few others, never got one and never did. The whole "pet rock" thing came and went and passed us by. I plan to do the same for "AI".

[–] stoly@lemmy.world 98 points 1 day ago (1 children)

Honestly: those who based a business model on this are probably bad business people.

[–] Carnelian@lemmy.world 42 points 1 day ago (2 children)

I feel for the employees whose lives will be tumulted by their moronic management who went “all in”

[–] Zizzy@lemmy.blahaj.zone 11 points 1 day ago

If you think this will stop at just harming the employees of those companies, youre wrong

[–] stoly@lemmy.world 4 points 1 day ago

Well everyone graduates with their CS degree and thinks that not only will they be making $150k - $200k in their first job, they will save up so much money and be so in demand that it won't matter that they get laid off.

[–] jonathan@piefed.social 30 points 1 day ago (1 children)

The headline seems way off from the content. It talks about planned IPOs as the basis, and somehow a "time bomb went off" "in real time". What kind of chronology is that. I've read better AI generated stories than this dogshit click bait.

[–] UnspecificGravity@piefed.social 65 points 1 day ago (2 children)

The idea of people investing billions based on "annualized revenue" tells me that rich people have too much fucking money.

[–] yakko@feddit.uk 38 points 1 day ago

They've literally never had this much more, proportionally. If this isn't too much, there's no such thing.

[–] dumnezero@piefed.social 21 points 1 day ago

Money is 100% wasted on the rich.

[–] Naich@piefed.world 44 points 1 day ago (2 children)

Enshittification in 5... 4... 3...

[–] BradleyUffner@lemmy.world 20 points 1 day ago (1 children)

How do you enshittify shit?

[–] Tronn4@lemmy.world 13 points 1 day ago (1 children)

AI.... now with CoPilot! 🙃🙂

[–] BradleyUffner@lemmy.world 8 points 1 day ago

It's the turducken of shit!

[–] Thorry@feddit.org 22 points 1 day ago

Microsoft has already raised prices with their Github Copilot service. In some cases the new price is 27x higher. And the actually useful models are only available in the super expensive plan (and then also incur high fees for actually using it).

[–] brucethemoose@lemmy.world 39 points 1 day ago* (last edited 1 day ago) (1 children)

How does this work when “good enough” AI like Deepseek V4, GLM and such are so dirt cheap they’re basically free for businesses? And available from tons of providers, or even self hostable?

And that’s only going to get exponentially more dramatic. Bitnet models + nonlinear attention + ASICs alone is at least an order of magnitude cost drop, though there’s tons of lower hanging fruit to pick.

AI is a race to the bottom, not the top. Intelligence has basically topped out, efficiency has not. It’s never going to replace most workers, but just assist them or automate dumb tasks, increasingly cheaply.

And Tech Bros can’t make trillions off of that.

[–] NaibofTabr@infosec.pub 14 points 1 day ago (1 children)

How does this work when “good enough” AI like Deepseek V4, GLM and such are so dirt cheap they’re basically free for businesses? And available from tons of providers, or even self hostable?

Typically what separates enterprise-grade products and services from alternatives is a contract with an SLA... but that generally means there's some contractual requirements for the reliability and productivity of the product or service. I'm not sure that any of the overhyped chatbots are reliable enough to support such contractual obligations, or that there's a useful way to measure their productivity.

[–] brucethemoose@lemmy.world 11 points 1 day ago* (last edited 1 day ago) (1 children)

contract with an SLA

Plenty of hosters provide that. Cerebras, for example, fabs their own ASICs (seperate from Nvidia), builds them into servers, hosts a number of open-weights models themselves in friendly jurisdictions, and offers SLAs for enterprise clients; it doesn't get more "guaranteed" than that in AI Land, but there are tons of hosts to choose from.

https://www.cerebras.ai/build-with-us

The major sticking point is that the best open weights models are Chinese. This doesn't actually matter from a security standpoint anymore than buying a Chinese tire does; they're dumb weights anyone can finetune, host and run on whatever software/hardware stack one wants... But try explaining that technical distinction of "using Chinese AI" to executives responsible for entire corporations.

There are even attempts to "launder" Chinese models to make them palatable for western enterprise use. For example:

https://huggingface.co/microsoft/MAI-DS-R1

https://huggingface.co/unsloth/r1-1776

[–] NaibofTabr@infosec.pub 1 points 22 hours ago (1 children)

Plenty of hosters provide that. Cerebras, for example, fabs their own ASICs (seperate from Nvidia), builds them into servers, hosts a number of open-weights models themselves in friendly jurisdictions, and offers SLAs for enterprise clients; it doesn’t get more “guaranteed” than that in AI Land, but there are tons of hosts to choose from.

This makes sense for first-party hardware businesses like Cerebras that are renting or selling their platform to developer businesses (second party) for the purpose of creating AI-based software tools which they will then sell as services to other businesses (third party), and I can see that guarantees could be written in a contract for the first-to-second-party relationship.

What I don't see is that any such guarantees can be effectively written or enforced in a second-to-third-party contract, where an AI SaaS company is selling their software service to companies that don't do their own development, and expect that the service they have contracted will produce reliable results.

[–] brucethemoose@lemmy.world 1 points 20 hours ago* (last edited 20 hours ago)

Actually, what Cerebra’s does is no different than any generic host. They provide API access to LLM weights, though most providers will do it with some standard open source serving software like VLLM or SGLang.

And they all use the same open weights LLMs. They arent the software developer.

Cerebras doesn’t train their own model. And I think this is fine for service guarantees as long as the weights do not change, hence will provide the exact same deterministic results at zero temperature (and generally perform the same when used as a service).

My experience is that a lot of “enterprise” LLM stuff is used in bulk, for results that can be “good enough” with a reasonable error rate. Like (for example) extracting info from literally millions of documents. Or as RAG/querying their own internal documentation.

[–] ininewcrow@lemmy.ca 28 points 1 day ago (1 children)

The rich have already cashed out and secured their fortunes

It's the suckers, losers, poors and those who think they know better who are all going to lose everything

[–] NaibofTabr@infosec.pub 20 points 1 day ago (1 children)

They're going to try to stick taxpayers with the debt in a "too big to fail" way, and the current administration will help them.

[–] ininewcrow@lemmy.ca 13 points 1 day ago

Another classic example of ...

CAPITALIZE PROFITS!

SOCIALIZE LOSSES!

When the system ... any system fails, the owner class like to congratulate themselves on how they made it work so good ... but when it fails, they're the ones always groveling to the masses asking them to pay for everything.

[–] plyth@feddit.org 11 points 1 day ago (1 children)

The real pricing will come when companies can't go back to regular workers, as predicted here: https://feddit.org/post/30017257

[–] davidgro@lemmy.world 12 points 1 day ago (1 children)

Apparently that post is gone?

[–] plyth@feddit.org 7 points 1 day ago* (last edited 1 day ago)

The point was that prices will rise when companies are dependent on AI to the point that the owners don't have money to pay for their AI girlfriends.

The title was "they're ransoming my Al girlfriend back to me at exorbitant prices". But it's not only AI girlfriends. Once AI runs the company, the regular AI can also demand ransom prices.

[–] Damage@feddit.it 4 points 1 day ago (2 children)

Didn't Amazon sail on while losing money for years as a public company? Why would super-hyped businesses like these have issues?

[–] ICastFist@programming.dev 3 points 16 hours ago

Adding more context to what phcorcoran said, AWS was something that actually had an "end goal". A new company offering cloud servers where you can host your stuff to be reached by the internet at large, which was already a proven necessity even back in 2003.

AI is still trying to sell itself as something useful. Not only that, the fixed monthly cost makes zero sense, because tokens have an actual monetary value - there is a cost in processing, cooling, network, etc, which can be attached to it. You'd need an army of low-usage users to pay for the power users^[you know those f2p games where the players who use their credit cards are above all others? The free players' value is in being the punching bag of the paying players. Now imagine the reverse: you need to recruit 20 paying players, who are ok to take a beating, to keep one unprofitable whale in the game. It makes no business sense, but it's exactly what's going on at the moment with their monthly rates. It's no wonder every other week, AI users are reaching their monthly limit faster and faster] to have it make any sense. The alternative is actually charging per token, like pay 10 dollars and get 1k tokens or whatever.

[–] phcorcoran@lemmy.world 10 points 1 day ago

According to Ed Zitron, AWS cost about $52 billion between 2003 and 2017 (adjusted for inflation), and became profitable after that. OpenAI did 13 funding rounds at a total of $180 billion and the signs are that they are burning through that capital at an accelerated pace and nowhere near profitable. The scale isn't quite the same

https://www.wheresyoured.at/premium-ai-isnt-too-big-to-fail/