Ever run an AI model locally? If you want the most capability you need a fast GPU with 32-48gb RAM. And that’s all for you, ONE user.
Copilot has millions of users, with tens or hundreds of thousands of them hitting the AI all at once. Each one needs $thousands worth of GPU and RAM dedicated to them for the length of their query processing.
Where do you think the money to buy all that hardware comes from? You see OpenAI buying a double digit percentage of the world’s RAM production, you think they got it on clearance sale?
No, there are investors. Investors who are pouring hundreds of billions into this AI stuff. And they don’t do this because it’s fun, they do it because they expect a BIG return.
So what’s going on is just like your neighborhood drug pusher, only the drug pusher is more honest. He says ‘first hit’s free, man’. AI company says ‘AI models are an easy and cost effective way to modernize your workflow!’; they don’t tell you that once you’ve integrated them and fired all the humans who know how to do the work, the price is gonna go way up.
Because the fact is, there IS a real cost of AI compute. GPU time, or at the large scale, datacenter space, power, cooling, etc.
In another few months to few years, the C-suites will stop huffing the koolaid and will start doing cost-benefit analysis on where AI is and isn’t cost-effective vs. humans. With any luck (for the AI people) by that time the AIs will be good enough that it’s a clear benefit. If not this bubble’s gonna pop.
Has anyone ever actually had a drug dealer giving out free hits?
I feel like that’s the biggest lie DARE ever told.
DARE told so many lies. I was told that if I smoked weed, it would ruin my life and I would become a helpless drug addict. Then I smoked weed and that didn’t happen. Then they got me thinking “what else did they lie to me about?”.
When I was a teen, I was offered drugs for free more than once, so it’s not all a lie. Maybe it is just uncommon in many places.
I’m Brazilian, BTW.
But was it from a dealer?
Yes.
Yeah no. That’s not good business, for small dealers. Only large ones looking to improve their margins and completely without morals.
Pretty much the only people who do that are actual medical doctors (who’ve gotten convinced into believing some lies by pharma reps who are the lowest of the low) and sample offerers at large grocery stores.
I mean I’ve had free hits from my dealers, and given out free hits. Of weed. While someone is buying weed. So it’s not like I’m pushing it on anyone or anyone is pushing anything on me.
Drug propaganda is mostly lies yeah.
Legalise, educate, tax and regulate.
That’s why I would rather run locally. I control my data. Better for the environment. And if I ask a programming question, sure ChatGPT will come back in seconds, but I’m fine with waiting 30-60 seconds for my own AI. How impatient are we.
So, what you’re saying…is the AI Bubble is going to pop once the pencil pushers do the math? But they’re asking their local LLM for that… so it isn’t happening?
Not pop. Correct.
A lot of the managers aggressively pushing AI have little or no understanding of it themselves. They just hear of a technology that can make a human more productive by doing most of the work for them. So absolutely that’s worth a ton of money. It’s why many companies are encouraging if not demanding employees to start using AI- because in their mind, one employee fully utilizing AI can do the work of two standard employees. Of course they believe this because they’ve never actually had to use the damn thing themselves and thus don’t realize it doesn’t do all the work for you. Or worse they think it does and your wonderful code base turns into spaghetti.
Side note- A few companies even had leaderboards for who was using the most AI tokens. This led to ‘tokenmaxxing’, trying to consume as many tokens as possible to prove you are adopting AI. Things like 'Write unit tests for our company code base, then refactor the code base. Spin up an instance of Claude and another of ChatGPT to each generate unit tests of the old code and run them against the new code, then run the tests against each other to check each other’s work, submit full debug output to another instance of gpt 5.5 that will check for hallucinations… Keep that query going for a few paragraphs and you’ll have an army of AI workers all checking each other’s work while producing zero productive output but costing a fortune to run.
Ever run an AI model locally? If you want the most capability you need a fast GPU with 32-48gb RAM. And that’s all for you, ONE user.
Even then, that’s quite small. Top of the line frontier models would be looking at hundreds of gigabytes of video memory, and just as much RAM.
A terabyte of VRAM/RAM needed for something like CoPilot is probably a fairly sensible estimate.
Depends on what you want to do, the model, and optimization or quantization.
A lot of LLM stuff that seemed pretty amazing a few years ago - chatbots and the like that respond to questions in plain language - can run in comparatively light hardware. Coding agents can take more, but could also be optimized against a particular language and spit out useful snippets.
Image stuff can be pretty complex especially at higher resolutions and detail, and creating seamless video segments gets expensive on hardware, fast.
Quite true. The thing is, there aren’t billions and billions of dollars in chatbots. The billions are for the creative stuff and the code.
And that is where the reckoning / correction will come from, the bill has to come due eventually. When top end generative AI starts to have a real cost associated with it, then it’s no longer a blanket ‘everyone start using this immediately’ mandate, it prompts some consideration of cost versus output quality.
And itbos slower and a whose model then they are using.
It’s more complex than that. The weights of big models are distributed, and then tokens are processed in parallel for multiple users. The setup varies, but it could be 8 GPUs serving many dozens of users at once, or bigger sets with even more parallelism.
I think the bigger problem is that Copilot is… shit.
It’s probably some ancient, inefficient architecture, not something super sparse and hardware efficient like (say) Deepseek V4, or Kimi 2.6, or Gemini Pro.
And literally every interesting dev team Microsoft has ever acquired (Phi, WizardLM, many more), and any interesting innovation they figured out, has just disappeared into a black hole.
They don’t have custom hardware, either, like Huawei NPUs or Cerebras WSEs, or Google TPUs. They’ve written some very interesting papers on that, and proceeded to do squat with them.
Also, it is AWFUL for its size. Tiny models that are basically free run circles around CoPilot.
What I’m getting at is that CoPilot is probably the most inefficient LLM out there. Like, it’s impressive how bad it is.
Umm copilot is just linkage to other models. My work VS instance defaults to claude but there are several others available. “Copilot” itself is not its own model
Really? I don’t use it for work, but I swore I was hitting some internal MS model for chat/code, as it was one of the worst experiences I’ve had with LLMs over 24B.
It wasn’t profitable. This bait and switch was a long time coming.
Yup, the “first one is free” deal is always a trap.
Literal drug dealer strategy
In just glad it happened relatively soon. Should have happened sooner :-)
I’m sure they wanted to drag out the adoption longer than it was viable, too. Didn’t quite get everyone dependent.
Sigh. None of this is surprising in the least. The cost of AI infrastructure and compute (coupled with the complexity of the chain) makes it prohibitively expensive.
It only has appeared cheap because of investor money flowing like Niagara on the off-chance that it could be made cheap enough to be profitable after getting everyone addicted to using it. I really don’t think it’s there, and it’s definitely not cheap enough to continue flying for free much longer.
There will be some major bag holders soon. Who’s is gonna be? The funnelers or the funnelees? How does MS bounce back having integrated copilot into everything?
Given the recent NASDAQ changes, looks like US pension funds will end up with the bag. What a scam.
Microslop can suck it.
So a massive wealth transfer - right after Covid, ANOTHER unprecedented wealth transfer.
Side question: when are you people in the imperial core going to do something about this with all that God given Liberty™ and Freedom® you continually bomb and destabilize the rest of us for?
Shakes Magic 8 Ball: Signs point to never.
We got ourselves a keyboard warrior here.
A stunningly generic defensive sledge from user BassTurd, I guess.
As in the 2008 saying: “All that money has not really disappeared. It is just in the pockets of different people”.
B- but the DOW 🥺
I don’t understand how everyone repeats this shit… as if we have never ever before seen a renewable asset… no, it’s not expensive to run them. It is expensive to give away usage. It is expensive to train. It is expensive to buy hardware. It is very very cheap to run. But hypercapitalism ruins everyone
So how do they get back the money spent on hardware and training? They also have to update their hardware occasionally. It is expensive to run.
Everyone could’ve seen it coming from mile away
anyone who didnt is a fucking idiot
I’ve got some bad new about our managers and leaders…
tried to explain director, was met with a blank stare.
failing up is real.
what? free trial over? 📯 ai spam free-for–all has sucked so far
The squeeze could not happen soon enough 😌
For real. This whole “AI/LLM” nonsense could’ve been a nothing burger if they correctly charged it at the get go… Ah, well. Venture capital is suckers, I guess?
From what I’ve seen, the AI companies don’t even know what to charge. For example, claude.ai finally made an API for user costs and we get this fun fact:
Values for a given date may be revised for up to 30 days as late events arrive and reconciliation runs. For invoicing-grade totals, query dates at least 30 days in the past.
The problem is though they made it basically free. And that was a mistake cause people don’t like paying for stuff once it was free.
I’ve been using it for well over a year now.
Hit my limit in 1 day and canceled.
Yup, same. x57 for GPT 5.5 now?! The GitHub agent architecture was hot as hell and allowed me to work on projects without ever bootstrapping it locally, but there is no way I’m gonna get sucked into spending hundreds a month. It felt good seeing that $250 refund come through for my yearly plan. I’ll take those funds to Claude or Codex directly. Throttled usage is way better than being fucked over by a glorified reseller.
It felt good seeing that $250 refund come through for my yearly plan. I’ll take those funds to Claude or Codex directly
I look forward to seeing the follow up comment where you do the same with one of those when they inevitably do the same
Lol, all AI is shifting to this model. Those other companies will bait and switch too just a matter of time.
Yeah I don’t doubt that anthropic is burning money so fast right now. It’s actually wild to see, GitHub’s pricing might actually be the first glimpse of real pricing for AI we’re seeing.
The company with all money all phones all email all calendars and meetings and satellites and shit will probably not.
It’s too bad their models are also all garbage, or I’d be on that Google AI train. They’re desperately trying to stay relevant in a space they should be owning in the first place. You plant shit seeds, you get shit weeds. Even if the models were good, their track record shows it’s not worth the time investment adopting their stack because chances are it ends up in the infamous graveyard within 3-5 years.
They’ll throttle you more - it’s just different way of gouging price
Perhaps, but at least I’m not being charged for each attempt at doing work besides my time in the attempt. Copilot would bug out and that cost me usage credits. Borderline stealing from its users paying for limited credits per month.
I have pro+ and I filed an FTC complaint when the billing change was announced and encouraged others to do the same, with the hope that it may lead to a class action settlement in the future.
Good luck, buddy!
*plays the world’s tiniest violin*
Not me. As soon as my boss said, “Use the tool thoughtfully,” I pretty much don’t use it at all anymore.
The most thoughtful way to use it.
I think this is "stupid money’, again.
I hope this makes companies stop being “AI native”
My reaction, exactly.
Would love to be a fly on the boardroom wall of any one of those companies which did massive layoffs “because AI”. Probably lowkey panicking on how to get their devs back.
Deepseek V4 Flash exists as a small open-weight model and is crazy cheap for what it is capable of. ClosedAI/Anthropic and other non-free models are sketchy AF with their pricing and basically everything else about them.
Nice. Maybe now my company stops trying to shove it down our throats.
btw soon GitHub actions will no longer be free to self host
For private repos, unless something changed since I last heard about that
