OpenAI could reportedly run out of cash by mid-2027 — analyst paints grim picture after examining the company's finances
https://www.tomshardware.com/tech-industry/big-tech/openai-could-reportedly-run-out-of-cash-by-mid-2027-nyt-analyst-paints-grim-picture-after-examining-companys-finances
782 points · 201 comments · view on lemmy.world
201 Comments
roofuskit@lemmy.world · 219 pts · 48d
How can we make that sooner?
droopy4096@lemmy.ca · 48 pts · 48d
use their cheapest plan, burn their tokens, burn the hole in their budget. could backfire though as then "clever analysts" will claim that demand is up and eager bankers will shell out more cash
RedstoneValley@sh.itjust.works · 44 pts · 48d
Or educate the people around you that "AI" is not the magical fairy dust machine that can do anything imaginable. Might be better than giving them any money at all.
DevDave@piefed.social · 12 pts · 48d
The people actively using AI are not an issue of lacking education on how bad it is. Just a quick example is of an upper management asshole who absolutely "loves" AI. Reality is they shit out slop but their direct reports undo the damage to protect their jobs. I know of more than a few developers in a tough spot where they are credit card slaves so work invisible overtime to compensate. They did this to themselves and are trapped in AI hell.
RedstoneValley@sh.itjust.works · 6 pts · 47d
Interesting perspective. Software development works different where I come from, the dev part is still very much in control of the developers themselves. However the management lives in delusionville and the expectations are pretty much insane compared to the reality of the actual turnout. My comment above was more about educating the general public... AI results can look pretty compelling unless you decide to have a closer look, and a lot of non-dev people are falling for it. Either because they are dumb and gullible, lack analytic thinking or because they have no real contact with AI and are just exposed to the hype through the media.
DevDave@piefed.social · 2 pts · 47d
Adjacent comment: Using the electrician analogy, AI is like one of those really cool gadgets you see on Temu that don't actually do half of what they claim and will probably burn the building down. Another analogy is AI are like Roomba's: they kinda work, take forever to finish, clog up easily, and are guaranteed to miss a lot of stuff.
RedstoneValley@sh.itjust.works · 1 pts · 47d
An air conditioning device that needs no pipe to the outside you say? It defeats the laws of thermodynamics for only 99€? Shut up and take my money!
/s
DevDave@piefed.social · 2 pts · 47d
Oh I like that but you might scare the MBA's using techno babble like "thermodynamics" :)
kewjo@lemmy.world · 2 pts · 47d
it's one of those things that people need to ask about things they are an expert at so they can understand how valuable the results really are.
DevDave@piefed.social · 2 pts · 47d
I was a software consultant with "architect"* experience for the last 10 years of my career or whatever you want to call it. 50~80% of my job was basically saying the same thing their in house developers were saying. Maybe the other half was pushing tech stack upgrades (CVS -> SVN -> Git/Perforce ), (CGI to basically anything else), etc etc. My least favorite role was being a hatchet man, eg one of the Bobs from office space.
Professionally I used a shit ton of analogies and metaphors. Absolute most effective was comparing new feature development to an electrician adding a new light or receptacle. You may find you need to upgrade the service panel (database or other services), it can take time to pull new lines and doing so can interfere with existing equipment. Finally you can only pull so many new cables through an existing building before you need to do necessary cleanups and rearrangements (eg refactoring). Failing to do so may lead to brown outs (crashes) or the entire building catching on fire (eg Microsoft Dumpster Fire 11). Last bit is you can help explain the complexity of a feature as top floor, mid level, or basement (eg "soonish").
* Architect title/role was something I would try my best to bury. The companies that needed an Architect only figure that out when they discover MS Access DB isn't going to work /s but only a little bit. Also what the fuck is an "Architect", its not like any of our titles have any national standards.
This video is kind of triggering my deep seated hatred of MBA's but if you ever want to explain to the normies what your job is like, "The Expert" https://youtu.be/BKorP55Aqvg is perfect.
RedstoneValley@sh.itjust.works · 1 pts · 47d
Yeah, I can relate. I've been doing consultant work and I always tried to avoid being pushed into the "architect" role. My background for over 25 years was software development and we had a lot of success with agile methods (doing it right is hard but viable and it produces quality software) My experience is that a good dev team does not need an "architect" to tells them what's best. But all of this is now gone anyway, or at least taking an extended break. At the moment no one is investing in development teams, and the prospect of being able to fire all the developers because AI can do their job now is making CEOs giddy everywhere. Not going to happen (and this should be obvious for anyone who can judge the quality of code or the effectiveness of processes in terms of reliability, quality, cost, performance etc.) but that doesn't stop them from trying.
DevDave@piefed.social · 3 pts · 47d
I framed the "AI" craze as parallel to self checkout kiosks. They were sold as the future and a way of getting rid of those pesky human workers. Reality is they're an open wound on the company. Unfortunately it will cost more money to get rid of them so they're still there.
That's the difference with AI currently. Cancel your subscription, toggle it to off in your IDE, and its gone. I believe this is why they're trying to push "AI" everywhere, hoping it will stick somewhere.
Otherwise I liked a well manged agile development process. Heck of a lot less stressful than water fall.
edit: fax spalling erwor
MrKoyun@lemmy.world · 1 pts · 47d
educating people doesnt work very often. people do not give a shit about what's going on inside or behind of their glowing rectangle machine.
neukenindekeuken@sh.itjust.works · 13 pts · 48d
Why burn tokens when datacenters are much more flammable
anomnom@sh.itjust.works · 2 pts · 48d
Start with the gas turbines
morto@piefed.social · 5 pts · 47d
That would definitely backfire. The best approach against any corporation is not to use them and let their name be forgotten, therefore, the shares losing value and becoming less and less attractive to investors
boonhet@sopuli.xyz · 4 pts · 48d
The most expensive plan gets you more tokens per unit of currency actually (10x cost for 20x usage), but it's pretty expensive and there's not actual guarantee that they're still losing money on your subscription in 2026 and as you pointed out, being able to show demand and MRR will get them more loans.
neutronbumblebee@mander.xyz · 13 pts · 48d
Look to your superannuation plan. Most providers have a conservative plan that's little or no shares. If lots of people transfer the message will be loud and clear. At that point the markets will dump AI and things will eventually normalize. I suspect this is already where the elite have their money, given the present warnings.
fizzle@quokk.au · 1 pts · 48d
Nah.
The more people divest the more appealing these investments are.
neutronbumblebee@mander.xyz · 5 pts · 48d
So more AI losses per investor. Shares are just like card collecting it only goes up if the demand is increasing. Right now I suspect it's reached maximum insanity and is ripe for a correction.
john_t@piefed.ee · 3 pts · 48d
I'm sure the big players are betting Trump will bailout AI companies. They don't have to worry. So they keep pumping the share prices.
fizzle@quokk.au · 3 pts · 48d
That's not true.
It can feel that way when you think as shares as numbers on a computer but they represent a share in the ownership of a country and presumably that ownership is actually worth something.
If no one wants to invest because of the vibe, then you can buy the shares for a bargain.
Maybe it's a bit like buying a house where someone was murdered. If you don't care about the vibe then the reduced demand means you can get a bargain.
You can argue the shares are grossly overvalued, and that may be true, but my point is that shares have an intrinsic value and if demand reduces it increases the appeal to other investors.
ZILtoid1991@lemmy.world · 10 pts · 48d
Get a sniper rifle and shoot their generators.
Don_alForno@feddit.org · 8 pts · 48d
Nah. They lose money every time you use their shitty services. Sabotage would actually improve their cashflow.
Semi_Hemi_Demigod@lemmy.world · 3 pts · 48d
Only by using a significant amount of power and water
HazardousBanjo@lemmy.world · 167 pts · 48d
Oh boy, I can't wait for my tax dollars to go to bailing them out instead of food and healthcare
Endymion_Mallorn@kbin.melroy.org · 33 pts · 48d
Ah right, they're "too big to fail".
theneverfox@pawb.social · 24 pts · 48d
Don't worry, we can't bail them out this time. There's just not enough money, each bailout is exponentially bigger than the last one and this time the bubble is bigger than the rest of the global economy
We'll probably destroy the global economy buying them just a few more months anyways though
T156@lemmy.world · 14 pts · 48d
It's also arguably the only thing propping up the American economy right now. They don't have anything to bail with, if the AI bubbles goes on fire. Their economy may well come tumbling down in short order.
wewbull@feddit.uk · 9 pts · 48d
I don't think you even have to argue it. If you move the top 7 from the us economy, the US has been in recession for about a year I think.
Upgrayedd1776@sh.itjust.works · -7 pts · 48d
pretty sure as soon as the strategic reserve is at minimum levels the gas prices will pop the economy just like in 08, looking at fuel prices, there is some consensus that the derivatives bubble and the economy was accelerated or contributed by the big spike then too, for what its worth here is a snippet from a longer conversation with claude the other day about the timing and effects....
"The timing is on your side. Oil peaked around $147 in July 2008, but Lehman didn't collapse until September 15, and the bulk of the crash to the $30s happened over September–December, right on top of the acute credit panic. So the dominant driver of that specific collapse was the financial crisis — a credit freeze, a global trade contraction, forced deleveraging, and a stampede out of commodity speculation — not primarily consumers and firms conserving because gasoline got expensive. Presenting $147→$30s as a clean "high price destroyed demand" story conflates a financial/credit event with a price-response event, exactly as you say. Fair hit. The honest nuance is that they're entangled rather than fully separable. James Hamilton — probably the leading economist on oil-and-macro — argued in a 2009 Brookings paper that the oil spike itself was a meaningful contributor to the recession, not just a bystander: US miles driven and gasoline demand had already rolled over in late 2007 and early 2008, before Lehman, and the auto-heavy parts of the economy were buckling under fuel costs first. His stronger claim was that the oil shock alone might have produced a recession even without the financial crisis. That's a contested, minority-leaning view — the mainstream reading is still that the financial crisis dominated — but it means high oil wasn't innocent; it was one of the stressors loading an already over-levered system. So the two didn't just coincide, they reinforced each other. And here's the deeper point your question exposes, which is worth more than the correction itself: it's genuinely hard to find a clean example of price-alone demand destruction, because oil spikes and recessions almost always travel together. Spikes help cause recessions, and recessions crush oil demand, so they arrive as a package — you rarely get to observe the price-conservation channel in isolation. That entanglement is real, and I was papering over it. The cleaner illustration of the structural piece — the permanent kind of demand destruction — is the aftermath of the 1970s shocks. Global and especially US oil demand fell durably through the early 1980s and didn't recover for years: fuel economy standards, switching oil out of power generation and home heating, industrial efficiency. That's the sliver that sticks even after the economy recovers, and it's isolatable precisely because it persisted past the recovery — though even there the Volcker recession is a confounder, so it's not perfectly pure either. So let me refine the claim rather than abandon it. 2008 is a fine illustration that a demand collapse crashes the oil price, but a poor illustration that the high price alone caused the collapse. My underlying point — that $150–200 oil is self-limiting — still holds, but the mechanism is better stated as: high oil prices help trigger or deepen a recession, and/or a coincident recession does the demand-crushing, and it's that demand collapse (from whatever mix of causes) that breaks the price. Pure price-induced conservation is the slower, structural component I described last turn, not the fast circuit-breaker. The circuit-breaker is the recession — which, as you're implying, may have its own separate ignition source and just happens to also torch oil demand on the way down."
xvapx@lemmy.world · 1 pts · 47d
Conversations with LLMs are absolutely worthless as a source of knowledge.
Upgrayedd1776@sh.itjust.works · -1 pts · 46d
they are great for doing a lot of web research into complex subjects, create an inventory of the most provelant discussions and historical precedents and form all that into an easy to read report. Since the llm is not the source of knowledge but the search engine, your comment is absolutely worthless as a coherent reply to my post
xvapx@lemmy.world · 1 pts · 46d
No, LLMs are not a search engine, that's not how they work. They process and regurgitate text, and even in such "simple" uses there is risk of hallucination.
Your conversation with a LLM is absolutely worthless as a coherent reply to anything.
Upgrayedd1776@sh.itjust.works · -1 pts · 45d
LLM's are search engines by their nature, but on top of that they have search engines they reference, they have powerful grepping and search tools to iterate through those results come up with the prevailing theories and outliers and present that as an organized report to you, unless you can find something that is factually incorrect about my statement you are arguing against what you are seeing with your eyes which i dont quite get
Jaysyn@lemmy.world · 3 pts · 47d
Also, and the point that everyone here is missing, bailouts have an end point.
Giving them tax payer's money still won't make them profitable after the VC money goes away.
moustachio@lemmy.world · 11 pts · 48d
Better start preparing to grind the economy to a halt when they try it. This time any occupy Wall Street movements should be heavily armed and wearing plate carriers
Shellofbiomatter@lemmus.org · 13 pts · 48d
Maybe Americans are finally going to use their second amendment for the intended purpose?
moustachio@lemmy.world · 8 pts · 48d
We can only hope. But the left wing of the working class needs to wake up and realize they have that right too; and get armed if they are not already.
tixooo@lemmy.zip · 7 pts · 48d
Endymion_Mallorn@kbin.melroy.org · 92 pts · 48d
How can we make them run out in mid-2026?
Whostosay@sh.itjust.works · 25 pts · 48d
I was under the impression that they had an extremely large amount of negative money.
itsjustachairmary@lemmy.world · 14 pts · 48d
If Skyrim has taught me anything it's probably integer overflow
Quetzalcutlass@lemmy.world · 6 pts · 48d
Since their current cash wouldn't fit in a 32-bit integer that'd mean it's at least 64-bit. An integer underflow would require a debt of over 9,223,372,036,854,775,808 units-of-currency before it rolled over to positives.
Man, this bubble is even bigger than I thought! /s
krashmo@lemmy.world · 11 pts · 48d
Let the internet do what it does best, create all kinds of weird and unnecessary porn. That should do it.
Yttra@lemmy.world · 5 pts · 48d
I don't think that went well for Grok...
motruck@lemmy.zip · 5 pts · 48d
Definitely not in my feed tonight. Grok helping PDF filez.
Marshezezz@lemmy.blahaj.zone · 9 pts · 48d
gilokee@lemmy.world · 5 pts · 48d
baseball bats with nails in them!?
Yttra@lemmy.world · 5 pts · 48d
Mid-2026? How about right now? :)
kosmikheart@aussie.zone · 9 pts · 48d
That’s the same thing… oh dog time is flying
Yttra@lemmy.world · 3 pts · 48d
I was hoping for yesterday but true, we really are halfway done aren't we...
tixooo@lemmy.zip · 4 pts · 48d
tristynalxander@mander.xyz · 88 pts · 48d
mr_satan@lemmy.zip · 22 pts · 48d
While I agree with not buying being better that giving them money. I can't agree with leadership being smart.
This thing was never going to be profitable. So they either:
a) they're actively malicious — they were never planning on profiting which means this is basically a pump and dump scheme capable of triggering some form of recession;
b) they're fucking morons for believing their own fantasies, which judging by their public appearances would track.
Former seems just as plausible, but my bet is on the latter. These AI bro CEOs seem to be woefully average at best.
tristynalxander@mander.xyz · 28 pts · 47d
mr_satan@lemmy.zip · 8 pts · 47d
I want this response to be higher. This is an excellent take.
jubilationtcornpone@sh.itjust.works · 6 pts · 47d
I've worked too many places where "b" is absolutely the answer. Even small companies that you've probably never heard of where management is drinking the koolaid. They usually don't like being asked why someone would actually pay for whatever bullshit they're selling. The answer, almost without exception, is because one of their CEO buddies told them it was a good idea.
Knock_Knock_Lemmy_In@lemmy.world · 1 pts · 46d
The bet was that bigger compute gave more parameters which gave exponentially better intelligence.
Unfortunately the increase in intelligence is logarithmic under current methods, so the business model is unsustainable.
Windex007@lemmy.world · 21 pts · 48d
Counterpoint:
They already are losing money. They themselves do not project that they will be profitable until 2030. The idea that "smart people with spreadsheets won't let them lose money" is obviously wrong because they have done nothing except lose money, ever.
They get thier operating capital from funding, not sales.
Now, I agree that the gambit is wrong. So you're right. You're just right for the wrong reason.
Funding is just the cash value of market optimism for the future of your product. High usage props up optimism. Social media IPOs were valued very much based on active users, the idea being that more users meant more opportunity for profit.
The more people actively using these tools, even if they're just maliciously burning tokens, just add to the "active users" metric. Which makes funding easier. And funding is the ACTUAL way these companies "make thier money".
tristynalxander@mander.xyz · 9 pts · 48d
Windex007@lemmy.world · 10 pts · 47d
Yeah, I conceded that your call to action was correct.
Just wanted to add colour so people understand the mechanics at play. When you understand them, it lets people evaluate other things, without needing you to tell them what to do.
For example, if your comprehension of these companies is that the companies are acting with the goal of profitability l, they would see something like a fast-track onto an index post IPO as a bid for legitimacy, some kind of ego play.
If you comprehend it as a beast that can only subsist on funding with no viable product, it entirely changes how you comprehend the post IPO index listing desire.
UnderpantsWeevil@lemmy.world · 15 pts · 48d
Uh... famously untrue. Mental illness has plagued monarchs, pharaohs, and caesers for millennia. That's long before you get to all the quirked up white boys and trad rad girl bosses currently running things.
Okay, but you can review their balance sheets and their primary lines of credit. Case in point, Sam Altman is heavily reliant on three big financial partners - Softbank, Oracle, and NVIDIA. Two of these are - themselves - hemorrhaging money thanks to large capital outlays that have failed to produce substantive returns.
There are finance journalists who get out ahead of this and report their own analysis. And you can find that in a thousand different private journals, substacks, and podcasts. But you can also go do the grunt work yourself if you're ambitious.
OpenAI's financials have been disclosed already (although the official SEC filing is still upcoming). So... just read them if you doubt what you're reading in the news. But it's not some kind of secret that the business is operating at a loss, with a fixation on debt-fueled growth. The argument is over future projected revenue, which isn't something business leadership can be any more certain of than a passive media consumer.
I definitely get the knee-jerk impulse to announce "business professionals know more about business than internet idiots". Because, sure. True. But the idea that business professionals don't routinely make bad decisions has some pretty historic well-established counterpoints.
"The business people know more" line is akin to saying "This used car dealer must know more about the vehicles on his lot than I do, so I can trust him".
tristynalxander@mander.xyz · 4 pts · 48d
UnderpantsWeevil@lemmy.world · 5 pts · 48d
I'll concede that trying to soak Altman for $14k/mo when he's already hemorrhaging billions is a bit like pissing in the river.
I might ask what you plan to do with the $14k in tokens you're burning. If you've got a material use case for them, and Altman wants to sell you $20 for a nickel, go wild.
But half the joke of AI is that you're burning tokens to do nothing. That's why businesses recoil as soon as they're asked to justify at-cost AI spend.
mattyroses@lemmy.today · 2 pts · 47d
"every little bit helps" said the old woman as she pissed in the sea . . .
FukOui@lemmy.zip · 10 pts · 48d
I think companies these days are more about valuation (how much people think they are worth) vs profit (what they actually earn). AI companies are loss leaders, yet vcs still fund them anyways. Paying them more money will just add to hype and valuation (eg "our XX service grew YY% which projects to ZZ potential profit")
Doomsider@lemmy.world · 10 pts · 48d
The only way to stop them is give them more money.
non_burglar@lemmy.world · 9 pts · 48d
Hubris is what the rich and powerful suffer from, not stupidity. And hubris makes the same blind decisions as stupidity.
nforminvasion@lemmy.world · 3 pts · 48d
Right on point
rumba@lemmy.zip · 8 pts · 48d
Bailout in 3...2...1....
Jaysyn@lemmy.world · 4 pts · 47d
Probably not.
rumba@lemmy.zip · 2 pts · 47d
I sincerely hope not. Yet... quite a few things that can't happen have been happening.
morto@piefed.social · 7 pts · 47d
I see a lot of people in this thread stating and insisting that people should give them more money, which fells so surreal. I guess people's mind have been shaped into consumerism to such a degree that they can't think about a solution to things that doesn't involve buying something anymore. It's so hard to believe
XLE@piefed.social · 5 pts · 48d
I agree with you that nobody here should purchase a subscription with any of these companies, but I will watch happily from the sidelines while AI fanatics engage in this exact behavior, jumping from loss leader to loss leader.
DupaCycki@lemmy.world · 3 pts · 47d
Unless money is not the goal. Not the sole goal at least. As you said, they're not stupid. It's entirely within reason to assume they may be aiming for more than just money.
Who knows what they're planning. Nothing good for us - that's for sure.
luthis@lemmy.nz · 53 pts · 48d
Yes but you have forgotten about the technique used by the most skilled and intelligent company leaders to consistently outperform all predictions for years now: Corruption
Cruxifux@feddit.nl · 17 pts · 48d
Step one: make thing
Step two: make everyone hate that thing
Step three: call it “too big to fail” and force you to pay for thing you hate with your taxes
hcbxzz@lemmy.world · 6 pts · 48d
Ah yes, the most profitable business
arc99@lemmy.world · 31 pts · 47d
I will happy when this bubble bursts. OpenAI, Grok and several other companies offer nothing substantive and if they're burning cash and disrupting economies then just die already.
plyth@feddit.org · 3 pts · 47d
Like industrial farming, there is nothing new, but now everybody can create memes and create software. Society will change.
NostraDavid@programming.dev · -9 pts · 47d
You were saying? 🤨
bold_atlas@lemmy.world · 11 pts · 46d
At least you're honest about about what most are using it for. But have you ever tried to talk to a real one?
NostraDavid@programming.dev · 1 pts · 21m
Big-tittied goth GFs? Nope.
Mostly for a lack of available Big-tittied goth GFs.
Also, I was making a joke. I'm feeling you're taking this WAYYY to seriously.
SaharaMaleikuhm@feddit.org · 28 pts · 48d
Come on, burn faster!
NocturnalMorning@lemmy.world · 27 pts · 48d
How can I help speed this up?
scottmeme@sh.itjust.works · 1 pts · 48d
Use very non-cachable tokens without paying them a penny.
Use something like repomix to do maximum damage.
wewbull@feddit.uk · -1 pts · 48d
Tokenmaxxing. They lo
ose more money, especially on the subsidized plans.Doomsider@lemmy.world · 16 pts · 48d
Nice try OpenAI
Jaysyn@lemmy.world · 4 pts · 47d
More like nice try Nvidia.
HugeNerd@lemmy.ca · 4 pts · 48d
Can you tight money?
wewbull@feddit.uk · 3 pts · 48d
It certainly appears to be escaping them.
HugeNerd@lemmy.ca · 3 pts · 48d
Nice
ShinkanTrain@lemmy.ml · 2 pts · 48d
Rubber band
doingthestuff@lemy.lol · 23 pts · 48d
Any chance of a bunch of used gpus getting sold off from this?
jj4211@lemmy.world · 11 pts · 47d
Unfortunately they would broadly be systems that need about 15kw, and as part of a board that won't work as discrete parts.
So you get it, but you'll need a couple of 60 amp circuits to dedicate to it... Also you have no viable video out
FoxtrotDeltaTango@sh.itjust.works · 7 pts · 48d
I hope so
Jax@sh.itjust.works · 7 pts · 47d
No, they're going to use the data centers for surveillance
jabjoe@feddit.uk · 5 pts · 47d
Might be like the AMD BC-250 discarded from crypto. Hopefully! https://bc250.info/
H1AA6329S@lemmy.world · 4 pts · 48d
Getting those GPUs would be like getting a beat up horse. It's still a horse, just not much of a use anymore
FoxtrotDeltaTango@sh.itjust.works · 5 pts · 48d
I still has some use, it’s just that it’s not as good but still useful for people who want to save money on gaming pc builds
Jason2357@lemmy.ca · 6 pts · 47d
Those GPUs will be about as useful as a bitcoin miner for gaming. As in exactly not useful.
They may be really useful for running your own local LLM, as long as you don't mind the hefty power bill and jet engine noise.
mattyroses@lemmy.today · 3 pts · 47d
LLM coops might have a big future
JohnEdwa@sopuli.xyz · 6 pts · 47d
OpenAI isn't running ChatGPT on RTX 5090's, they use data centre hardware like the nVidia NVL72, which are entirely useless for anyone else. You wouldn't even be able to boot it as it uses more power while idling than a residential home has access to, and up to 150kW at full load.
GamingChairModel@lemmy.world · 7 pts · 47d
That's the last generation. They're moving from Blackwell to Rubin chips now, and the 72-GPU Rubin servers use up to 230 kW.
The typical residential connection in the U.S. has a 24 to 48 kW electrical connection. A block of houses might not have enough power infrastructure to power just one of these server racks.
doingthestuff@lemy.lol · 3 pts · 47d
A friend of mine was a big crypto miner but he was kinda crazy. He died young and I helped his remote family process his estate. They let me take his computer equipment. I got a dozen tired GPUs but my dad has been running his 4k monitor on one of those old 1080tis for about six years now and it's still doing its thing. No one has any info on his crypto accounts, all of that is just lost for now.
Blackmist@feddit.uk · 4 pts · 47d
Aren't they using low accuracy FP units that aren't suited for gaming anyway?
Jaysyn@lemmy.world · 4 pts · 47d
I was hoping the same, but have since been told by several people that know more about it than I do that they are useless for gaming.
JcbAzPx@lemmy.world · 4 pts · 47d
I wouldn't say entirely useless. They just aren't plug and play replacements.
Jaysyn@lemmy.world · 2 pts · 47d
I hope you're right.
muusemuuse@sh.itjust.works · 22 pts · 48d
Doesn’t matter. The regime got what they wanted out of this. The data centers are up, the tech is there, they will just snap it up cheap for use in surveillance and fabricating evidence of dissenters.
wewbull@feddit.uk · 11 pts · 48d
The data centres are far from up. Many of them look to have a few steel beams in place and that's it. Others are sitting empty due to lack of hardware or lack of power.
Chulk@lemmy.ml · 10 pts · 48d
The datacenters largely aren't built yet, as someone else pointed out. That doesn't even matter though, to your point.
The Epstein class has successfully destroyed large portions of the Internet through bot traffic. They've scraped virtually all human contributions and stolen it, while ruining traditional vehicles for information gathering. They've eviscerated consumer hardware, and thus shifted control of compute away from the masses. They've successfully manufactured consent for mass surveillance including harvesting our biometric data.
Once the AI bubble bursts, it will be one of the largest transfers of wealth in history as they force us to bail out this fradulent industry. If it doesn't throw us into another great depression, the damage it will do will ensure that things never go back to how they were pre-AI. Once that reality becomes apparent to the masses, they'll be primed and ready for Peter Theil's vision of "Freedom Cities"
Jaysyn@lemmy.world · 2 pts · 47d
No one is bailing them out.
FoxtrotDeltaTango@sh.itjust.works · 1 pts · 46d
I hope so
tristynalxander@mander.xyz · 1 pts · 47d
NostraDavid@programming.dev · 20 pts · 47d
Lemmy, this is the fourth year in a row that "OpenAI bankrupts soon" articles are being posted:
OpenAI isn't going anywhere.
Brkdncr@lemmy.world · 11 pts · 46d
Because they really should be failing. These companies are leveraging against themselves and getting propped up by the government.
They will get bailed out just like 2008 to avoid a cataclysmic global financial meltdown. I won’t be surprised if other countries chip in to bail them out.
InnerScientist@lemmy.world · 6 pts · 47d
This is the year of the OpenAI bankruptcy.
NostraDavid@programming.dev · 12 pts · 47d
And it's the year of the Linux Desktop! What a coincidence!
pingveno@lemmy.world · 3 pts · 46d
To be fair, the Linux Desktop use to have virtual no market share. Now it is close to 5%. Linux used to be absolutely painful to run as a desktop environment. Now, there are some rough edges, but it works pretty well.
Treczoks@lemmy.world · 20 pts · 48d
It will be a race which AI company holds out longest. All of them are making losses that no company can survive, and if they rise their prices enough to cover the costs, they will basically lose all their customers. Who will then struggle to hire the people back who still know how to do things without AI.
Auli@lemmy.ca · 9 pts · 48d
Then Google wins? Since they have the ad business sonar least they are making some money.
sobchak@programming.dev · 5 pts · 48d
I think they could significantly lower their costs if they turn their focus away from the race to "AGI." But, their valuations don't really make sense unless investors believe they will achieve AGI.
Treczoks@lemmy.world · 4 pts · 47d
I was there when the "Neural Networks" idea started. They thought they could program NNs to achieve AGI. Now they think they can do it with LLMs. But LLMs are just parrots with a large dictionary. They won't reach that point, either. An LLM is way too much rooted in words to be able to think.
melroy@kbin.melroy.org · -6 pts · 48d
Recent models are quite good. Like gpt 5.6 sol. Even better than mythos 5.
expr@programming.dev · 12 pts · 48d
How is this a response to a comment about AGI?
LLMs are not currently, nor ever will be, anything remotely resembling AGI. AGI is still entirely within the realm of science fiction, like teleportation or time travel.
Jestzer@lemmy.world · 1 pts · 48d
Hey, we have AGI! We’ve had it for years! It’s called making babies 😎😎😎😎😎
melroy@kbin.melroy.org · 1 pts · 46d
That's true. I'm just saying the models do get better and are quite impressive now.
But whatever the term means. I know where AGI stands for, but the term "agi" is still very vague. You can't compare a machine llm model with the human brain.
Olap@lemmy.world · 10 pts · 48d
Yet still hallucinates mass bullshit and can't count, can't architect, and can't write a decent test
sobchak@programming.dev · 4 pts · 48d
The "best" one I've tried is the latest Opus. I don't trust any of them to use for real work, so I mostly just play around with a local Qwen 3.6 27B or Deepseek v4 Flash. I have heard OpenAI's latest models produce less bloat than Anthropic's.
I should say I do find LLMs useful as a kind of search agent (both web and large unfamiliar code bases). And GhidraMCP is pretty cool (maybe just because I don't have much experience with reverse engineering).
HornyElf@lemmy.world · 18 pts · 48d
Some big company would invest in openai to keep the bubble alive.
chiliedogg@lemmy.world · 41 pts · 48d
That's how they've been operating for years. They'll go public soon enough though, so your 401k will take the hit instead of the parasites on top.
That's what the SpaceX IPO was about. They were listed on the NASDAQ way earlier than companies are normally allowed to be so the assholes up top could raid your retirement fund despite there being no path to profitability.
moustachio@lemmy.world · 11 pts · 48d
I can’t fathom how people actually contribute to a 401k when it’s run by Wall Street ghouls that stole everyone’s money 20 years ago and never suffered any consequences for it…
baggachipz@sh.itjust.works · 11 pts · 48d
Because that’s the game. If you don’t play it, they continue to make more and more, while your money decreases in value due to inflation. Best you can do is try to be the bird in the rhino’s back, picking up fleas and scraps.
samus12345@sh.itjust.works · 18 pts · 47d
"grim"
krisevol@lemmus.org · 5 pts · 47d
Every bubble has created something we use every day. AI isn't going anywhere.
JcbAzPx@lemmy.world · 18 pts · 47d
You use tulips every day?
Jtotheb@lemmy.world · 14 pts · 47d
You bundle and sell subprime mortgages every day?
modus@lemmy.world · 5 pts · 47d
You use the world wide web every day?
bold_atlas@lemmy.world · 2 pts · 46d
dot com bubble was about speculation over domain names. The internet was certainly not created by a bubble.
krisevol@lemmus.org · -1 pts · 47d
Mortgages are used everyday.
Railroad are used everyday
The interest is used everyday
Amazon is used everyday
Those are the big bubbles in recent history.
bold_atlas@lemmy.world · 5 pts · 46d
Railroads were not made or built by a bubble.
Amazon did not invent remote shopping by mail. That's been around since 1800s. They weren't even the first website to do that.
and loans and interests are abstract concepts that have existed for millennia so I don't know what you're talking about there.
Bubbles and speculation don't create wealth, new tech or products. Like what aspect of bubbles are you trying wash here? They're bad for everybody except for the few rent seekers at the top.
krisevol@lemmus.org · -1 pts · 46d
Yes the railroad was a bubble. A ton of people lost money in that investment.
Amazon was a bubble, it lost 92% value when it popped.
Bubbles don't create wealth, i agree
The housing bubble lost people a ton of money that that investment, and people also lost their pensions when the Bible popped.
A bubble is when people start heavily investing in a product to the point it's a losing bet because that sector doesn't need the money, or can't produce the returns bubble investors are looking for. So when the AI bubble pops a lot of people will lose money, but the product will be here to stay because of the investment.
My point is that when a bubble pops the product is here forever, where most people think a Bible popping is the product is a failure and will disappear.
TheBlackLounge@lemmy.zip · 3 pts · 47d
Don't forget the AI bubble is also a ram and SSD bubble.
elucubra@sopuli.xyz · 6 pts · 47d
That is not a bubble, it's a shortage caused by a bubble.
IamBlank@lemmy.zip · 14 pts · 47d
Fingers crossed! Let it burn.
AlteE@programming.dev · 13 pts · 48d
Oh no. That is another whole year to wait.
melroy@kbin.melroy.org · 2 pts · 48d
Well they most likely get bailed out by the government
Corkyskog@sh.itjust.works · 13 pts · 48d
Q3 2027 is when the house of cards falls. I have been saying this since 2025 and I will continue to make this prediction based off of loan and investment terms.
BarneyPiccolo@lemmy.today · 7 pts · 48d
I've been making the comparison to the first synthesizer in 1897. It was so huge, that it took up the basement of an entire city block in NYC. It was enormous, and relatively useless but it worked...technically. But 60 years of development later, and it can be put in a suitcase and carried around.
I see data centers and AI the same way. Sure, we can technically do it, but it's big, unwieldy, and wasteful. It clearly isn't ready for prime time.
Go back to the drawing board, address the real problems, including regulations, and get back to us in a decade or two, when they've figured out how to do this properly. Because right now it's a monstrosity that's more of a curiosity than an useful product.
DarrinBrunner@lemmy.world · 5 pts · 48d
"Figure out who to do this properly" means a completely different method. What they've built is way, way too expensive for too little return. What they want to do may not even be possible, too little is known about the human mind to assume it is. It's all assumptions and hubris at this point, no proof. Not everything we imagine is possible.
Auli@lemmy.ca · 0 pts · 48d
Still going to take massive amounts of data to train an AI.
brucethemoose@lemmy.world · 5 pts · 48d
I agree with this.
That lines up on the technical side, too, with how LLM “intelligence” is plateauing, cost of cheaper models is decreasing, where open weights are going and such. Q3 2027 seems about when an OpenAI coding subscription makes no sense.
That being said, I’d be wary of the “Facebook effect.”
Once a service gains a huge foothold, it can deteriorate for a long time without going away. Especially with regulatory capture. And for many smartphone users, OpenAI is the only AI they know.
Jaysyn@lemmy.world · 5 pts · 47d
The difference here is OpenAI has no revenue, Facebook has plenty of advertising revenue.
hark@lemmy.world · 1 pts · 47d
That's where advertising through AI comes in. People relying on AI suffer from cognitive deterioration which is the perfect opportunity to sneak in advertisements disguised as suggestions through AI responses.
anon_8675309@lemmy.world · 12 pts · 48d
That long? Sigh…
He’ll probably tell Trump he’ll help win the election and get all sorts of tax payer handouts.
orenj@leminal.space · 11 pts · 46d
Faster please. I want to buy computer parts again
lemmelemmy@feddit.org · 9 pts · 46d
Yeah they’ll just sell more data and get more money from oil barons or something. They’re not going anywhere.
trolololol@lemmy.world · 9 pts · 48d
How can you run out of something you never had?
TammyTobacco@sh.itjust.works · 14 pts · 48d
Oh they had plenty of other people's money. They have to burn someones money very very fast
HugeNerd@lemmy.ca · 7 pts · 48d
Can't be soon enough. The amount of AI sludge videos out there where I can't even tell anymore.
UnfortunateShort@lemmy.world · 1 pts · 48d
Didn't they kill their video generator?
HugeNerd@lemmy.ca · 1 pts · 48d
There must be many others. The amount of videos depicting things I've NEVER seen before pawned off as real is disconcerting. A dog bouncing a 3 foot exercise ball on his nose in a backyard? Yeah, that really happened.
Earthman_Jim@lemmy.zip · 6 pts · 48d
But I was told 2027 was when the AI would finally get good. /s
XLE@piefed.social · 3 pts · 48d
Best AI news I've heard recently!
Earthman_Jim@lemmy.zip · 2 pts · 47d
Agreed! I'm a professional animator. I fucking hate this shit.
Jaysyn@lemmy.world · 2 pts · 47d
Yeah, but Altman & Dario have been lying from day one.
Earthman_Jim@lemmy.zip · 2 pts · 47d
Oh, most definitely.
mattyroses@lemmy.today · 2 pts · 47d
The next six months will have Iraq turn around!
Thomas Friedman
hakunawazo@lemmy.world · 6 pts · 47d
It could be an Auryn. It symbolizes the fucked up circular market flow between software and hardware manufacturers and how their greed is a neverending story.
Rooster326@programming.dev · 5 pts · 46d
Is this just a sexual Ouroboros?
hakunawazo@lemmy.world · 4 pts · 46d
It's two of them, after they discovered 69.
justaman123@lemmy.world · 2 pts · 46d
https://www.larp.website/
These guys should adopt it as their logo. It's gotta be satire right!? Right?
sik0fewl@piefed.ca · 5 pts · 48d
Oh no...
Endymion_Mallorn@kbin.melroy.org · 7 pts · 48d
Yeah, it's taking too long.
Anonymous_Leaker@lemmy.world · 5 pts · 48d
Jaysyn@lemmy.world · 3 pts · 47d
Yeah, it's cargo cult garbage.
foodandart@lemmy.zip · 5 pts · 48d
I have one word for that: GOOD!
ApeNo1@lemmy.world · 5 pts · 47d
I like that the articles main image looks like a giant gold turd.
sirico@feddit.uk · 4 pts · 47d
An Ouroboros made of shit
NostraDavid@programming.dev · 1 pts · 47d
I presume it's supposed to be a (golden) snake eating itself. Could also be a mom protecting her eggs.
AbsolutelyNotAVelociraptor@piefed.social · 4 pts · 48d
What do you mean "grim"?
BeN9o@lemmy.world · 2 pts · 48d
it's bad, it means miserable, depressing.
AbsolutelyNotAVelociraptor@piefed.social · 3 pts · 48d
Exactly, so why are they describing this as grim?
BeN9o@lemmy.world · 2 pts · 48d
Ohhhh hah yeah this is only 'grim' for them! I assumed you didn't understand the word 🤦♂️ my bad!
ilinamorato@lemmy.world · 3 pts · 48d
"Grim" says ye.
"He he hee ha haw ha haaa haaaa heh haa ha ha ha hee haa haaaa haaaaw...!" says I.
nibble4bits@lemmy.dbzer0.com · 3 pts · 47d
Let's accelerate the the process. What's the AI query version of DDOS?
Yes, they could just set more limits.. but itll still shave the time sooner.
kokesh@lemmy.world · 3 pts · 48d
Sooner the better
HrabiaVulpes@europe.pub · 3 pts · 48d
Ain't it ironic that technology failed mostly because humans are still cheaper option for the same work?
roserose56@lemmy.zip · 3 pts · 48d
No shit! Let that sink in, I'm waiting. If the world don't want to learn from it's mistakes, let it burn.
Iamaquantummechanic@lemmy.world · 2 pts · 47d
I sold all my funds today. Gonna hold on to my cash until this bubble bursts.
thethunderwolf@lemmy.dbzer0.com · 2 pts · 47d
fym grim
Cordyceps@sopuli.xyz · 2 pts · 48d
Cool! I might just be on time to get out of my payment plans before the market crashes and interest goes up!
green_goglin@thelemmy.club · 2 pts · 48d
Puts
RMonk@lemmy.world · 1 pts · 47d
ryannathans@aussie.zone · 1 pts · 48d
US govt would step in if true
moustachio@lemmy.world · 5 pts · 48d
Why would they step in?
They have barely talked about stakes in AI companies to this point. It’s nothing like banking and housing market that were actually useful, and impacted people’s lives, unlike AI. LLMs are just an aggregate of stolen work that they sell back to gullible C level execs.
Goodeye8@piefed.social · 6 pts · 48d
They shouldn't. There are a lot of arguments why they shouldn't. But there's only one argument why they might, it's the most corrupt government in the entire US history and they're very gung ho about siphoning wealth upwards. No amount of logic or reasoning will work against corruption. If the government wants to bail them out they will find a way to bail them out.
ryannathans@aussie.zone · -1 pts · 48d
The Pentagon is militarily dependent on OpenAI now, used for targeting, aggregating intel, etc. That's a big part of the US govt hostility towards Anthropic, Anthropic won't permit used of their advanced models for war
moustachio@lemmy.world · 6 pts · 48d
The pentagon… the agency that can’t pass an audit, despite stealing most of our tax dollars… needs LLM companies to steal more of our money to bail their big investors out, for fucking up stealing our labor/art and trying to sell it back to execs to put us out of work?
If they do try and get a bailout we def need to be armed with ARs and Plate Carriers for our protests.
fxdave@lemmy.ml · 0 pts · 48d
Hey as a European I happily use your money and willing to pay only a cheap claude subscription 😊
But if you keep this industry and find a way to operate it on low cost, then the US wins big in the long-term.
baggachipz@sh.itjust.works · 3 pts · 48d
Yes they do and yes they will. They only have a couple stipulations but their models are already being used by the military.
ryannathans@aussie.zone · 1 pts · 48d
The key difference is not for weapons/targeting/surveillance
The rest is admin
GodofLies@lemmy.ca · -6 pts · 47d
Amazon wasn't profitable for many years until it was. Now make your own conclusions about AI.
Jaysyn@lemmy.world · 8 pts · 47d
Apples & oranges.
https://www.wheresyoured.at/cargo-culture/
jj4211@lemmy.world · 5 pts · 47d
There were a lot of companies that were unprofitable during dot com and failed.
The question would be if openai is like Amazon or webvan or Netscape or AOL or....
I would argue that openai is more overextended than amaxon was, and worse, they aren't really seen as the leading company anyway, nor do they have a strong hook. Anthropic is broadly more well regarded, and Google has the captive audience of phones and browsers.
Despite this reporting suggests openai maybe the most insane purchasing commitments.
I don't think it ends well for openAI, however durable AI market ends up otherwise.
jabjoe@feddit.uk · 1 pts · 47d
Until it was dominant and could abuse that power to squeeze everyone.
I'd argue this is dumping.
Not just Amazon either. A good few of the American monopolies. AI spending is the pattern repeating and it is a dark pattern for everyone.
hark@lemmy.world · 1 pts · 47d
I can run an open weights amazon from my house?