Apple Will 'Watch Everything Burn' When the AI Bubble Bursts(macrumors.com) |
Apple Will 'Watch Everything Burn' When the AI Bubble Bursts(macrumors.com) |
In 2020 I could write audio apps that worked incredibly well on a Pixel: despite LLMs! I still can get Siri to put tasks in the right todo app 70% of the time.
I updated to iOS 27 beta hoping something had changed, but nope: the biggest change is a massive black orb search bar.
The people who are going to end up making the most money on this are creditors and future businesses. When the AI bubble does pop there will be a massive glut of data centers and hardware available. Both Apple and Microsoft are realigning their entire businesses to brace for this. When the AI bubble pops businesses that sell hardware, like Apple and Microsoft, will face an immediate price shock because they have had to raise prices to account for more expensive hardware. That shock will be short lived and prices will fall accordingly with disruption to supply chain but otherwise minimal disruption to margins.
I look forward to the bubble popping because when retail hardware becomes cheap again all kinds of new business opportunities will open in the self-hosted service market.
Think Netflix. If you believe the money spent on Netflix over the course of a year is less than the entertainment you receive back you will likely continue to pay for it. If you believe the money spent on all entertainment subscriptions exceeds the value returned you will likely start cancelling some, or all, of the subscriptions. AI is not immune from economics or human behavior. When customer company budgets get tight the money spent on AI subscriptions might be reallocated or reduced to prevent reductions in head count.
Hell no.
They'll pick up companies in trouble at bargain-basement prices.
>Consumers and enterprises alike have been trained to pay a monthly fee for a service, and while these services might have limits or strictures, basically nobody buying software expects to have a metered service, let alone one that's both metered and with hard to measure costs.
Has Ed Zitron not heard about the cloud? Unpredictable AWS bills?
But most consumer's aren't paying per token for access to models, so unless that changes and the labs start charging API pricing to everyone, it's kind of a moot point.
A legitimate criticism will be Apple not fixing their memory problem in the next 2 to 4 years…
Users go on vacation, they slack off, they spend the day talking to each other. There are very few people who are really effective at burning tokens. how do you know the ratio? do you have insides? No :)
The biggest target is enterprise, and the economics for an LLM vendor look like this: price per token = R&D + inference + infra investments. When you buy a subscription, you are quite often buying a year ahead. That lets the vendor predict future infra investments against hard commitments, and sell expensive per token pricing to everyone else. And when a hard commitment sits unused because the user is busy, they sell it twice. It is loyalty in exchange for predictability, in exchange for the promise to always deliver SOTA to users.
Vendors control the harness. Tomorrow they simply roll out a router where reading the code and doing the final edits goes to a cheaper model, and their math suddenly becomes very sexy.
Isn't that hard to predict that their economic model is very easy to tune? and this is just first baby steps.
I personally pay per token ( do not have subs for work ). I did have once a $25k/mo worth of tokens, since i knew it was free so i was doing crazy experiments. Now , 2 month later, my bill was barely $1.5k since i moved into different stage with project. I do have team members who burn $500-600. pre router, pre optimization.
I switched recently to grok 4.5 and cursor router and my bill will go even further down. It rotates 4-5 different vendor models cheap and expensive too, depends on the task. Routers will flip entire LLM economy upside down.
If you have properties of the market where your costs will go down , the size of the market will increase and you are top contender. How is that a bubble or a bad market ?
Sure you have risks of underperforming and lose the competition, but how is that different from any business in the world ?
What's unclear to me is if this is a systemic issue that's going to cause credit to freeze up but imo the opacity of the shadow banking "system" does not help here. If you see one cockroach, etc.
In fact spacex story tells you next : investments in infrastructure is the best investment. If OpenAI or anthropic have committed infra in the worst case scenario they can re-sell it with margin .
The only way it will not payout suddenly we wake up in the world where ai fails to deliver . Which does not seems to be the case .
Apple’s just been sitting there with solid cashflow waiting for all this to implode and then have on-device models with their own chips.
Apple will own the end device while the rest of the world is fighting over a pure commodity. It will burn hard and Apple will laugh all the way to the bank.
I'd qualify this with for now. It's not entirely clear what happens in the event the AI industry implodes, but I do think there will be a glut of memory and hardware up for grabs. It very well could crash the consumer hardware market as a result. The AI obsession and investment has burrowed so deeply across the US economy that I can't think of a single industry that won't be impacted, including agriculture.
Maybe. On the one hand, if there is a strong ceiling to what users are willing/able to pay for a smartphone (might not be true, with all those people making $$ on the stock market), they will have to cut prices.
On the other hand, if one part gets more expensive, premium phones need smaller relative price increases than budget phones. Certainly, if they are willing to accept lower margins, premium product might get more attractive.
Also, iPhones tend to have less RAM than competitors, so this might hit other premium manufacturers harder than Apple.
Having to raise prices isn't a great look but when they come out the other side they can drop pirces and get a boost in volume or they can keep them high and enjoy a bigger profit margin.
The ones who will actually suffer is Apple 2.0 (presumably a different competing company), start ups, research, companies that are not worth trillions.
My outsider take is that Apple leadership never really believed in the capability of GenAI, and hence kept pursuing their pre-LLM strategy. An interesting signal about how that's working out now is that their AI org leadership was overhauled recently.
As for laughing all the way to the bank, I am sure memory manufacturers and Nvidia was definitely join them regardless of what happens.
Also comparing Apple (literally one of the largest companies in the world) to a company burning VC money makes no sense.
Compare that to Meta, who went all-in (so much so that they changed the company name) despite apparent overwhelming user apathy toward the product.
The estimates I've seen say that Apple spent ~$10b on their VR program, whereas Meta is in for about $80b.
And what about Google? They probably thought for sure when they met with Apple that they would wipe out that $20 billion yearly payment to Apple every year for the default search position but the best they could do was to come out with a measly $1 billion a year refund both scenarios sound like Apple took it in consideration and determined that it was not worth the price, that does sound like a plan (The Art of War type) when the rest of tech is going crazy over AI.
As for the rest, they’re the only major tech company to not have a serious case of AI FOMO and dumping all their cash into that FOMO. Thats not an accident.
I think it is pretty clear at this point that Apple has gone all-in on being the ideal edge silicon for AI. This leverages their core competencies, requires only modest investment, and will likely pay out no matter how the AI market eventually shakes out. They are in one of the only parts of the obvious future AI market where there isn't really a fight for greenfield turf with other big companies.
Staying in their lane is arguably the optimal business decision for Apple and they lose nothing by it.
The best price for a product is what I call the "suck air through your teeth" price. You want your customers to suck air through their teeth... and then pay the full amount anyway.
Uber set their per-developer token allowance to $1500 per developer per tool. That suggests to me that they think they can get at least that much ROI out of AI tooling.
Selling $1500/employee/month plans to companies is a great business to be in.
Whether companies pay will come down to the bottom line once the hype around the country club dies down a bit and for that it's still TBD on the actual product cost impact.
If Uber were asked to pay full price, they'd either get the same value for significantly more outlay, or keep their outlay capped but get significantly fewer tokens.
This would, I'm sure you'd agree, change the ROI outcome.
Zitron is arguing that if we all pay the true cost once the subsidies go, we'll either stop paying for the value we get today, or we'll find AI to be less useful than we do today, and those businesses won't be sustainable, and then <pop> goes the bubble.
I enjoy keeping an eye on https://isaiprofitable.com to see just where we are overall. Interesting reading if you like to think about such things.
it's coming out of the salary budget from what i've seen. i've seen a company both say "AI max. it's the future! don't be late" and then go on to increase limits to compensation across the board.
They can only subsidize you so far, because they may not be even be covering their variable cost at this point.
There's no software multiplier (build once — pay the cost once — sell many times).
Traditional SaaS is in a middle ground, there are operational costs associated with providing services, but per request they're usually negligible.
AI? I don't know, but it's not looking great from where I sit, unless there's a significant breakthrough in inference efficiency.
But that also makes the case for government subsidization + bailout, if you accept it as a capability critical to national security, but one that may not be profitable to run within the US on its own merits.
One logical gap in the SemiAnalysis 40x cost of tokens versus subscription: I don't know anybody actually maxing out their account limits.
Sure, if you're somehow always running stuff, you can max it out, but subscriptions like this allow people to max sometimes (or always), while others never come close to the max.
What is the average usage of subscribers? Only Anthropic and OpenAI know, as far as I can tell.
If anyone is in my age group, they remember when ATMs were free. It was the "heroin dealer" business model. It worked out well (for the banks). The Chinese did it with manufacturing.
Getting people hooked on subsidized junk, is an age-old (and highly effective) business model.
Think of all the shops that will soon be composed of people that simply can't even get out of bed, if their LLM is not available. If the LLM dealer starts raising the price, there's no choice. I suspect many of the valuations are taking this into account.
> "Junk is the ideal product... the ultimate merchandise. No sales talk necessary. The client will crawl through a sewer and beg to buy."
-William Burroughs
Curious that he references a METR study from July 2025, before the leap in model and harness performance towards the end of 2025/early 2026.
> The primary reason is that we have observed a significant increase in developers choosing not to participate in the study because they do not wish to work without AI, which likely biases downwards our estimate of AI-assisted speedup.
So compared to just last year, they had a hard time finding participants because too many didn't want to work without AI.
He selectively quotes the max theoretical enterprise pricing equivalent of fully using the private subscription. Did SemiAnalysis not also claim a very high margin?
He throws in doubt about the providers having decent margins, which he claims is made up by "AI boosters" rather than leaked financials and open-weight pricing.
Then next he talks about "the real cost" of inference as if it was in any way realistic that labs price the enterprise plans near cost, like he seems to imply.
Then next he claims AI is actually slowing developers down and there isn't much difference between the models.
It just seems delusional.
yes, because in order to take a short position you have to predict exactly when the bubble is going to pop, which is different from predicting that at some point it will
Edit: to be clear, I am talking about Ed’s contention that AI coding isn’t net productive.
I would love to replace my monitor with Apple Vision Pro for programming and productivity. I would gladly pay $1000 for that.
But at $4000 it really needs to put me in a Microsoft Flight Simulator cockpit.
The future of AI would be on-device models which are as powerful as current frontier models and also I can imagine companies have their own deployments of inference of open weighted models for most of the use cases and use the frontier models for extremely niche or higher intelligence tasks.
As an example I use Claude code heavily for every day development and Opus 4.8 was already good enough for my use cases and never used Fable. Also note that I use AI as a tool to help with my work and I do not offload everything I have to do to AI in a single prompt
This is an interview with Ed Zitron.
I am running new betas for macOS/iOS/iPadOS and Siri is actually useful for a much wider set of use cases. I asked Siri last week what models it was using and one of those listed was Gemini which is confusing because I enabled free use of OpenAI in the settings. Regardless, Siri is much more useful than it used to be.
I also think it's irresponsible to not broach the obvious implication of "PC components becoming prohibitively expensive" + "untold amounts of compute sitting in compute warehouses with nothing to do because the AI companies that used to own them folded". You probably won't even notice when everything in Best Buy becomes a thin client.
China is doing all the R&D for free and the whole thing turned out to be unprofitable anyway.
LOL WHAT.
Actually the right framing is Apple hedged the risk by maintaining good relations with China. They did not need to take any of the risk.
HSBC didn't engage in the unwise practices that led to the great financial crisis; their share price still dropped by 75% in 2008.
If the AI bubble does burst chaotically, then I'd expect all tech stocks to decline to at least some extent and for even the strongest survivors to remain in the doldrums for years (in the cases listed above, the share price of both IBM and HSBC remained flat for almost a decade).
Fable is great, but Opus can handle most coding tasks for a fraction of the cost, and Sonnet is good enough for average questions or word processing tasks.
Just more wishful thinking from our favorite AI skeptic
Didn't see national security mentioned a single time in all the current comments or in the article. Speculations about profit are great and there are many companies that will largely be irrelevant across all the major metrics which will not be in a good position, but there are always these people lost on the profit picture of anything.
Does anyone realistically believe that all of these smart people are rushing into AI to maximize profit? No, they're rushing into it, because it's important and awesome. The potential value, not simply measured in dollars, is enormous.
Despite China aggressively training on western outputs, their cloud services can't be trusted and their open weight models are too large for most people to run. Running any of these open weight models through proxies or 3rd party services isn't really expected to be more secure or more private than any of the big companies. If you aren't running it locally then it's irrelevant and most people cannot do that for any sufficiently big model at a usable performance.
Ctrl+F: "web search"
Almost nobody talking about the giant shift in web search. Web search also remains a hugely profitable business and AI is becoming the new web search. It is the default now for many people and that will likely continue to increase.
Is anyone going to argue that web search is unprofitable? No. Nobody. AI will get better and more efficient as needed. You'll have your efficient to run models and your expensive models, branched as needed. This is the biggest shake up we've seen in search in decades, but even with that, Google remains dominant.
Bing made AI search their default and it actually hurt them in my opinion, because it was very bad and mostly wrong. Google added AI search, but it's been a lot more accurate and grounded. Keep in mind that Google search remains the default for a lot of devices. You don't even have to install ChatGPT or Claude on your phone, because Google is already there and people are getting used to Google AI results.
When I talk to random people, do you think they tell me they're using ChatGPT or Claude? Basically never. They talk about Google Gemini or Google AI. Gamers have an ok chance of saying they use ChatGPT. Programmers have a good chance of saying they're using Claude or ChatGPT. Anyone else it's almost universally Google Gemini or just Google AI. You do a basic Google search and you're already within the potential flow of an AI conversation.
Making money is important so that you can increase the amount of reinvestment into the product, not strictly to maximize profit. From the situation a lot of these AI companies are in, acceleration should still be more important than profit and whether they can focus on acceleration may depend on their investment picture.
OpenAI and Anthropic are going to need a long term reinvestment strategy that accounts for Google having the potential to outpace them and become good enough that nobody bothers launching or loading a separate app anymore. Keeping smart people using your product is an important aspect of it all since you want your product to become smarter too.
Maybe Apple will simply pay for these services the way it does for Google search, but if they want to maintain their privacy philosophy they will either be dependent on models from these companies or have to roll their own.
Either way, AI is sufficiently complex and capable that there will remain many ways for companies to specialize and provide profitable value to people for decades to come. It adheres to physics. It takes energy. It takes time. User experience is important, but the more that private user and company data gets used the more that trust will continue to be relevant too.
I think some of these analysts are too caught up in a few numbers here and there to understand what is happening. Are OpenAI, Anthropic and Google simply going to explode and cry "oh no, we were dumb"? That feels like the picture this guy is painting and it feels a little naive. That also ignores the fact that they are basically developing the smartest AI that can theoretically help them navigate their business into the future.
Really? I thought the AI bubble happened because something genuinely novel had been invented. People immediately saw practical uses for it, and then it took on a life of its own - funding, superfunding, $trillions becoming part of day to day lingo.. etc.
AI is not in the same class as Vision Pro, which Ed Zitron is comparing it to. I personally found the overall tone to be a bit hyperbolic.
I sincerely hope, when all this madness ends, there will be enough data centre surplus gear we'll all be able to soup up our homelabs.
[1] https://images.macrumors.com/t/PtKleXF2po_g3qSuA6VMOE90WEM=/...
Do you even know what FOMO is? The money is being spent in physical infrastructure, not the the metaverse.
AGI means cruelty-free robot slaves.
To give you some idea, here’s an Asianometry video describing Taiwan’s prior DRAM attempts: https://youtu.be/ehT3U935Pww
What leveraged? yeahnah thats not going to be profitable.
Not true.
> He also described a shift toward running AI locally rather than in the cloud – a move motivated by privacy, security, and the rising cost of inference as agents consume more tokens. However, Brooks envisions a hybrid future in which agents decide what runs on-device and what gets sent to the cloud.
And coupled with their efforts in auditably-private cloud computing, that's a strong pitch.
Do those of you who have read the white paper have confidence in this claim (as I’d hope)?
Blog: https://security.apple.com/blog/private-cloud-compute/
> This leverages their core competencies, requires only modest investment, and will likely pay out no matter how the AI market eventually shakes out.
This looks more like a call or a small raise or whatever (I don’t play poker).
This is a common misconception. This is NOT true: memory designed for CPU performs horribly for GPU tasks. Mac/Strix Halo/NVIDIA Spark are performing horribly compared to a desktop video card with GDDR.
I spent quite a few weekends optimizing ROCM kernels for strix halo, i WISH I had GDDR instead of high-frequency CPU RAM; the bandwidth would be SO MUCH better.
I'm quantizing weights not to make computation faster, not because I'm out of memory, but because memory cannot move fast enough to be processed and it's cheaper to load quantized version, convert into bf16 compute, and discard; This happens every single time a token is generated for the whole model over and over.
Ed may be right about some of his claims, but he is 100% a crank, and he makes so many incoherent claims I'd say that if he's right, it's in the nature of a broken clock.
If you are making the case that he is full of shit, please provide some actual evidence of his main thesis that AI companies are not going about this in any sustainable way
He's right about some things -- off the top of my head:
-- OAI's perpetual fundraising, due to their losses
--Coreweave
--Circular financing
--Revenue/capex imbalance
...all forensic economics analyses.
He's often wrong when on the topic of capabilities and adoption -- he mistakes a snapshot for a trajectory, and treats current limitations as permanent:
--Hallucination/unreliability an unsolvable problem
--OpenAI failing
--AI capabilities have plateaued / models will stop getting better
--Saying Microsoft's cancelled leases meant the bubble was popping
--Agents a marketing fad
--Gross miscalculation of OAI 2025 revenue
--Lack of adoption
--AI not getting more efficient (compared to 10-40x reduction in inference costs YoY)
--Repeating "95% of pilots fail" but ignoring massive bottoms-up adoption
Christensen got this right when he had the insight to judge a technology by its rate of improvement relative to what a market needs, not by whether it's good enough today.
Put Zitron back at the early days of the integrated circuit and imagine what he would have written about the technology.
Some open predictions:
--Bubble is going to burst -- I tend to agree with the thesis that "there is some marginal capacity being planned/built today that will never pay back its capex"
--Losses mean there's no viable business here -- the recent moves by all of the players to some form of usage based pricing show there's price discovery occurring. I don't know of anyone who is stopping using LLMs because of the pricing changes...it's just changing their behavior.
That isn’t to say that he’s right, that’s just to say that while he gets creative in his phrasing numbers don’t lie and I haven’t seen anyone else present competing numbers that make sense.
If anyone has them to the degree with which he provides them then please, by all means, I’m interested.
Implying there is a "gushing torrent" of pro AI narrative is bizarrely out of touch. We both know this isn't true.
the pro-AI side on the other hand has poured billions into ads and marketing and CEOs are forcing it on people due to a combo of FOMO, personal investments in AI (CEO, board, investor), etc
folks in the US have been pretty aware that bubbles based on political opinions -- we're all surrounded by news from "our side" yet we know the other side exists in some other bubble. but for AI, we're all either surrounded by pro- or anti- opinions and its a little shocking to learn that theres a parallel internet with the opposite stance. especially shocking when you find somebody that lives in teh same political bubble but opposite AI bubble.
Happy to read why is he wrong, and change my mind, as long as the arguments provide the same level of analysis he provides.
My professional opinion is that LLM technology doesn't work very well for software development and I'm not interested in hemming and hawing over its supposed benefits any longer.
And ATMs can be free still, if you use a bank that refunds all worldwide ATM fees, such as Charles Schwab.
I'm biased because I don't use AI, but I don't understand why you wouldn't fire these LLM-addicted people and replace them with people who can code or write an email without the use of an AI assistant.
Besides that, I actually disagree that the Vision Pro is not good for gaming. It's not good for traditional VR gaming, but using apps like Portal it's phenomenal for, e.g., playing existing PS5 games on a massive virtual screen, which in many ways I actually enjoy more than VR gaming, which is far too limited in comparison.
Conversely; the grandparent comment is still right, Apple could have made this a real VR headset without compromising any part of the AR experience. Apple knew that Hololens was a total failure, they saw the pushback on Google Glass, they should have known that people would feel like an idiot buying one of those headsets. Instead of trying to recoup the missed opportunity, Apple doubles down and pretends that those experiences are beneath them.
Anyone with a $300 headset can stream games to it. AVP isn't unique for that, it's was already outsold by the Index and it's eschatologically destined to be outsold by user-respecting headsets like the Steam Frame. It's downright impossible to imagine a world where Vision Pro outsells the Quest.
Apple had more money and influence than combination of all of them, yet failed with car and VR. Why would anyone think they can't fail much much bigger with AI and LLMs ?
FOMO, someone else being successful in VR (and now in glasses) is a nightmare scenario for Apple because it's not their platform. They've got a lot to fear from someone else owning the platform, having written the playbook on how that platform would be controlled and exploited at everyone else's expense.
Imagine if Apple doesn't make their glasses: Meta keeps selling truckloads of them, as they get more powerful the smartphone becomes an optional accessory and then an unnecessary one, eventually they become an alternative to a smartphone and an app platform for third party devs.
It's the same story with VR, Apple does nothing and risks Meta or Steam building a viable platform where people want to use software instead of on their iOS devices, until the hardware can replace iOS devices entirely. (although VR is certainly more of a moonshot)
Something to consider: would your description also apply to the dot-com boom of the late 1990s? The internet was real, the ideas for internet business were real, and we were not going to the previous reality. But the valuations weren't quite right and a "correction" happened at some point.
When people talk about AI crash, that's what they mean. Not that AI is a hoax, but that the correction could be quite violent and have effects on the broader economy.
Also I remember reading that Anthropic is on its way to be profitable in 2028
Access to a centralized copy of the web/literature/media, scraped and indexed/integrated, seems another non-local center of gravity. Versus a local model's last minute reaching out to "manually" search and browse.
Also mass parallelism for large ensembles. Perhaps unless/until we get those local models as 10k+ tok/s chips. Local can follow frontier because frontier is still sort of "expensive rack local". If STOA becomes massive burst-parallel ensembles, that following may get harder.
That’s the crash… that’s pretty much exactly Ed Zitron’s thesis
I think the flaw in this logic is thinking about how AI is currently used only. Yes, Opus is good enough for the task you are asking it to do, but that doesn't mean that is all you will ever need.
As AI gets better and better, it will open up new use cases that require the better performance.
And there are many such moonshot startups.
AI on GPUs is an efficient as gaming on CPUs.
All that math where perfect precision is not required means that you can’t tell do things in different ways.
In addition to on-device, Apple is making efforts to secure computations that need to occur off-device, see: https://security.apple.com/documentation/private-cloud-compu...
In this case, it’s really irresponsible.
Onboard decent LLM performance thats _power efficient_ is at least two/three hardware generations away. (assuming linear performance)
but, the valuations, with debt trade and private credit obscuring exposure is a recipe for disaster.
Depends on what you mean by early.
WAs it released before the market was proven? yes.
Was the whole user experience up to scratch before it was released? no.
As someone who worked on the Quest pro/3 ecosystem, We were shitting our selves, because we knew that apple would only release something when the user experience was right.
Oculus just shat out features as and when performance cycles needed juicing.
Sure the UX was good, but it was less good than I expected. The headset was bulky and surprisingly off balance. It wasn't an all day wearable (i mean the quest pro was actually comfortable by comparison)
The resolution wasn't that great and the only thing you could do with it was either look like a twat recording video or have spreadsheets pasted everywhere. I was expecting an OS where we could put things on work tops, create augmentations to my room, decorate or be creative. Instead we got OSX XR.
If XR companies had been as "reckless" with their product strategies as mobile companies in the 90s and 2000s, AVP would have had its iPod moment 6 years ago and we'd be hurtling towards the facephone. The market suffered because of the entrenchment of incumbents with all-too-clear memories.
Ultimately as a manager you are held accountable to your investors and the cost-benefit analysis is such that it was the right action to take.
People are starting to kinda hate tech. People don't seem to be looking for yet another "smart manacle". Time to move on.
National security is a canard for covering up national investment. The US doesn't need to have it's hand in the pockets of private businesses, but by making a murky NATSEC claim they can manufacture interest for investments and under-the-table cooperation. This is exactly how we saw companies like Google, Apple and Microsoft get roped into PRISM, in exchange for alignment on political objectives and upholding the security theater of obviously-backdoored shitware like BitLocker. Departments like the NRO have had their own in-house AIs for decades, they're not beholden to any of these firms for access to cutting-edge inference. SENTIENT predates the conception of OpenAI as an organization.
If you disagree; look at China. Why would China Open Source their frontier models if there was a genuine national security threat in doing so? Because there is no threat, it's a tacit refutation of the United States' fearmongering and a direct attack on the credibility (and profitability) of the America's frontier labs like OpenAI and Anthropic. While Anthropic and OpenAI send representatives to grovel in the Oval Office to release a model, China greenlights anything, just like a real neoliberal free market would. The US' federal interventionism is causing China's AI to recoup all of the ground that America loses in credibility.
Fact is, China's hosting is no less trustworthy than America's hosting. OpenAI and Anthropic are certainly just as backdoored and data-hungry as China's hosted services are, I'd be shocked if both weren't populating a vector database of my characteristics for "marketing" purposes or somesuch. But I still trust China more as an American citizen, because their ability to prosecute, blackmail or manipulate me is much weaker than my domestic government. There's no fear when I send a request to Z.ai or Qwen because neither of them are begging for federal interventionism on a weekly basis. They have nothing to ask my government for in exchange for my data.
Tbh, I hope it does. Prices are outrageous, and I'm a firm believer in personal computing/having access to powerful hardware, privately and locally, is hugely important.
On a more selfish note, I miss the days of being able to build an outrageously powerful desktop for myself for relatively cheap. For now I'm still holding onto my AM4 motherboard w/ 64GB DDR4 and an aging GPU, praying my 7 y/o GPU holds up long enough for prices to drop again.
Unfortunately, I don't think IBM Cloud will suffer enough for me to play Tux Racer on my very own Z17.
in fact, agri sees its share of impact now! all of this investment has to come from somewhere, which means that money is getting sucked up into AI from the entire economy. job cuts in tech are part that, part general "de-bloating" of post-COVID tech. retail investors are a lot more likely to invest into AI than any other industry, although there are signs that big players already consider risks of overexposure to AI.
not to mention, agri saw a "bubble" of its own, surrounding Agtech, which peaked at 2021 and has been cooling ever since. scare quotes here because it was nothing like AI bubble
For years he was all-in on "AI is useless, it has no business value at all" claims, and then when that was decisively disproven by Claude Code, he didn't spend one second on self-reflection on why he was wrong and whether this might mean he is wrong about other things. He just smoothly pivoted to "AI companies will never be profitable, tokens are hugely subsidized". And now that that's about to be disproven (word on the street is that Anthropic will soon report profitability, and both Anthropic and OAI have repeatedly said inference is profitable), he's pivoting again to "tokens are too expensive and the ROI isn't there for business". No correction, no reflection, just "we're at war with Eurasia, we've always been at war with Eurasia".
Even if he’s sometimes right, there are better analysts out there who are also right and have the honesty, humility and integrity that Zitron lacks.
No such disproval has happened. Claude Code does not provide business value, it provides the illusion of value. Just like everything else LLMs do.
>And yes, that sound you hear is the slow deflation of the bubble I've been warning you about since March...
>How does GPT – a transformer-based model that generates answers probabilistically (as in what the next part of the generation is most likely to be the correct one) based entirely on training data – do anything more than generate paragraphs of occasionally-accurate text?
etc.
Basically the essence is he's skeptical of AI / LLMs being any good so thinks all the investment is down the drain. Meanwhile AI progresses such that Fable can probably beat most humans and IQ tests, maths and the like. And the 'bubble' didn't deflate yet.
My take is that as a PR guy, used to people hyping stuff and not that up on comp-sci he fundamentally doesn't get what's going on. He assumes it's all hype rather than the steady progress in computing reaching brain equivalent levels and beyond.
Fable scoring OK on some benchmarks does not the mean the investment has financially paid off, because that's not how ROI works.
You're right that no bubble has deflated, but I think we can all see that a) there seems to be a bubble - I'm old enough to remember the dot-com era bubble, and this definitely feels like that, b) the only company clearly making a profit on AI right now is NVIDIA, and c) the net economic value of the sector as a whole (i.e. money out > money in), is still unproven
It's not even obvious to me as somebody who is up on comp-sci, that this isn't all hype, that the progress is/will be steady, and that reaching brain equivalent levels and beyond using these architectures will be economically viable.
I can create a perfectly good human brain in 9 months, train it in ~20 years, and have it pay off economically in a more proven way than [waves hands] all of this.
For some reason we've decided paying a hyperscaler the equivalent of a year's salary to do a job in a a day that a similarly paid and skilled human could do in a month is value for money.
You could probably give everyone in the US free healthcare for life and a free ride through college for less money than has been pumped into AI in the last 5 years, and have a more proven economic model.
Sure, time value is a thing, but given the error rates, the energy issues... this direction is not exactly a slam dunk as a net benefit, and I think that's all he's called out in any of the writing I've seen of his.
I also happen to agree with his take on Oracle, as it happens - they're only going to survive if it turns out they're a part of critical national infrastructure deep in the bowels of the US government. They look totally over-leveraged otherwise.
I'm not qualified enough myself to judge Zitron's reporting, nor this criticism. But George Pearkes is a reasonably respected financial analyst.
I actually think Zitron would be better off focusing on those deals where there is clearly something weird going on (SpaceX IPO, Oracle/OpenAI/NVIDIA triangle, the neoclouds, and so on), than picking a fight with the one company that actually seems to be trying to play it straight in this sector (Anthropic), and I can see how he's covering the whole field with the same cynicism which might not be appropriate.
That said, I'm not entirely convinced that some of Pearkes' analysis deserves us all to align to the rosy bullish view he takes either.
Time will tell, it always does, but at least you've put forward some data unlike anyone else, so thank you for that.
I mean you can say he’s not a “science guy” but he’s undeniably “smart”.
I am sympathetic to some of his criticisms that the enthusiasm and financial commitment has run far ahead what can be delivered. But I don't quite share his intensity over the doom and gloom. There probably will be a correction. It will probably sting. But I don't expect it to be the near wipe-out that Ed's passionate voice seems to steer towards.
A "correction" in this model isn't a reasoned and gradual re-evaluation of the market cap of every company. It's a broad pullback where people panic and no one wants to be left holding the bag. Money shifts into other assets for years and tech employment, incomes, and the availability of funding takes a big hit.
As to your comment about profitability... every unprofitable company is on a path to be profitable. Some even get there.
Plus, look at it this way: Kellogg's is a profitable company and a part of almost every person's life. Does this make them worth trillions of dollars? No, they just provide boring, commodity products, their valuation is basically a low multiple of the assets they hold and the revenues they bring. There's a future where OpenAI or Anthropic are more powerful than all the world's governments combined, but also a future where they're Kellog's.
Anthropic's offerings are vertically differentiated. However from an economical stand point, it also be true that they are not the preferred option for many.
He said his version of this observation is "your agents are only as good as you are." I think he's right, and the key is to practice and build the skill.
that doesn't sound like a net benefit at all. I think mostly people just enjoy playing with these toys, and good developers are still good with them, bad developers still bad
This comment is so obviously false (in my experience) I'm wondering what sort of niche environment you're working in
This is pretty rude!
This is less agreeable. You kind of just made statements without even having claims (even without requirement of evidence or details) to back them up.
This is like the illusion of contribution.
i have yet to have an LLM beat me. i am forced to try regularly so i don't look like an AI anti at a company that is very much SV pilled
Some developers work better with LLMs in their workflows than others. Some problems are easier for LLMs to generate a reasonable solution for than others. Some folks prompt minimally and see what the vibes bring. Others start with a detailed architecture and implementation plan.
The individual results will depend quite a lot on the person, the problem, and the approach. It's not a guaranteed winning formula for everyone.
And the price has to be half as much with all of those improvements. And last but not least the software needs to get even better all of that sounds like iteration over what? 3-4 years.
Fortunate for Apple, the competition is even further away…
If we're being actually honest with ourselves, Apple did not even try to make a VR headset. They tried to carve out a nonexistent niche with AR, and bashfully wrote it off as a Tim Cook folly like the Apple Car or Airpower.
Yes Meta spent 80 billion sure, but did you know Meta never had a loss-making quarter in their entire history as a public company?
My overall point is that its not as clear cut as Apple is smart for willingly not being good at AI and they will win in the end.
IMO - while nothing is a guarantee against a state-level actor that can compel both hardware-level bypasses and put such bypasses under a gag order, it's a well-thought-out system under those constraints. The idea of a self-auditing and self-healing fleet that requires simultaneous compromise of software and enclave hardware is a meaningful level of defense in depth against literally everything else.
The point isn't that his thesis is fucked, just that his analysis tends to be free with details in a way that shouldn't inspire confidence, e.g. mixing up EBIT and EBITDA.
Someone elsewhere in this thread referenced this guy [1]. His takes properly summarise the lack of care Zitron appears to display for getting detailed arguments right.
If you're deeply familiar with financial jargon, I think he's fine. But if you're not, it's easy to get whisked into woo-woo nonsense that's falsely precise due to mis-using (and in some cases, very clearly mis-understanding) core financial and economic concepts.
I use LLMs all the time, but I'm also a highly capable and experienced engineer that can definitely work without. I have just found that the LLM is a force multiplier, like I haven't had before.
I haven't personally seen a speed boost from LLMs, but my experience with them suggests that they could be effective in some domains and inadequate in others. E.G., I have observed that LLM tools are inadequate for generating C++ code.
- An electrical component "synthesizer" which takes natural language specifications in (e.g. from an RFQ) and uses material data sheets and core physics calculations to design new components. (Python)
- A transcript processing pipeline which reads what happened on a call, identifies participants from the CRM, drafts structured call notes, and posts them to the CRM with the proper associations after a human-in-the-loop confirmation. Fairly complex server & web app. (Go)
- Interaction design prototypes for a family of smart devices, some with screens. Brought together as an interactive web app to see how each device behaves (vanilla JS). Each prototype generated with the help of an LLM, including 3D animated examples (Claude driving Blender).
I have found it does good PHP code, and Swift code that definitely needs adult supervision.
Also, very good for things like code documentation, debugging, and writing copy for things.
I'm a web developer to myself who has done zero game or systems programming so I can't speak to that side of it, but you reminded me of it!
You should probably understand Thomas' comment in part as a reaction to Zitron attacking Thomas in a pretty low-brow fashion, essentially for saying "hey folks, AI coding works and you should be using it":
https://www.wheresyoured.at/how-to-argue-with-an-ai-booster/
The bottom line is that Ed Zitron, like Gary Marcus, built their entire brand on being AI contrarians; they can't say anything positive about it without qualifying it with a more damning negative. There's nothing wrong about having voices like that, but it makes them unreliable narrators. I am apprehensive about AI and its externalities, but I don't want to be caught citing either of them for that reason.
Sure, but an ad hominem is still an ad hominem regardless of one's personal feelings (it took me a good minute to figure out who "Thomas" is).
Otherwise, I agree, and I try not to cite extremists or bullies on either side. I don't even really care that much about defending Zitron, but I do put him on in the background because he does so many podcasts as I find it soothing to listen to someone rant and rave against the tech bro overlords, even if some of the shit he says annoys me. But he does show his work sometimes and wanted to point that out.
Nobody has the patience to read a comprehensive point by point rebuttal to any of it, it's just too tedious.
You say "numbers don't lie". But they do, when the numbers being presented have been adversarially chosen. You just find numbers (no matter how low quality) that fit the chosen narrative, you throw out the numbers (no matter how high quality) that rebut it. If you're trying to predict the future, you can't afford that of bias. If you're an anti-AI grifter, you can't afford to not have that bias, your entire livelihood depends on only showing things that support the narrative.
This is the crux of my argument: Zitron provides numbers and he provides a lot of them sourced from reliable and trustworthy sources. He provides counterarguments to numbers-sparse arguments by using even more actual, real numbers with a logical theory formed from them.
This should make it rather easy to dispute his claims, no? So then where are the equally rational disputes?
Just this article has a dozen instances of numbers being used to prop up the argument. They're mostly irrelevant to whatever argument Zitron is trying to make at the time. Some others are straight up misrepresentations. HN really is not a good forum for point by point rebuttals for that many cases. Do you think there's a particularly high impact use of numbers in this article?
Really? I think we've heard C-suite blabbermouths and borderline nontechnical tech CEOs being amplified by social media making these assertions, but boy did the weakly efficient market deliver a relatively swift correction to that mindset, no?
Q: "What would it take to change your point of view?" A: "[AI] would have to solve all hallucinations forever, which they are completely incapable [of]."
Can you produce a human that is infallible? I'll wait.
Finally: https://martinalderson.com/posts/no-it-doesnt-cost-anthropic...
My hot take is AI is increasingly less unprofitable as the cost of serving tokens drops and Nvidia's ongoing offers to guarantee profitability is a sign that it isn't stopping anytime soon.
https://newsletter.semianalysis.com/p/nvidia-gpu-debt-backst...
You either believe in the underlying science and technology or you don't. But in a world where AI is a fad like cabbage patch kids and beanie babies, what's next?
AI will almost certainly not be a fad for software developers, or at the very least for web developers, but it's certainly possible that other industries will look back on "experimental AI days" and chuckle.
Even if there is a financial bubble (that pops), the technology is still not going away.
As for what's next, I dunno. I've heard people talking about similar tech that doesn't use neural nets/etc, but that is out of my wheelhouse.
But I think AI is here to stay and it has already proven useful. And sure, we will look back on today's models like we look back upon Gordon Gekko and his giant early cell phone. And I cannot fathom how one can not separate AI the science and technology from AI tech bros and CEOs. I agree the latter are going to go through some things as the AGI fails to arrive on their schedule, but IMO there's no turning back on the technology nor should there be.
He has zero numbers and is completely making stuff up to a gullible audience.
The real numbers show high demand for all these services to the point where the companies are rate limiting the services because demand is TOO high.
The main argument of Ed Zitron is just that. Huge huge amounts of debt that's due "soonish". Every SPV is different but we are fast approaching the point where this all starts to become real.
Now I dunno why Ed Zitron feels the need to write 10,000++++ word articles that only says "more leverage than people expect", but that's really the crux of things.
You’re thinking of a different person.
Most recent free article: https://wheresyoured.at/the-subprime-data-center-crisis/
Number after number, e.g.:
/article-new/2026/07/iphone-18-pro-max-counterpoint-bom.png){kind=link}
“Per my own analysis, NVIDIA’s predicted $1 trillion in Blackwell and Vera Rubin GPU sales (by the end of 2027) represents around 40GW of data center capacity, which will, assuming a PUE of 1.35, result in around 30GW of usable capacity. At a cost of around $12 million a megawatt, that works out to around *$435 billion in global annual compute revenue to make these data centers necessary.*“Additionally, as another commenter has pointed out, his thoughts and criticisms are being echoed by people who actually have the money.
Another point against your gym analogy. When people don't have that much disposable income why would they subscribe to something they semi regularly use? And if it becomes more expensive why would they not cancel it?
Having a subscription might make people run agents overnight, or casually generate images for fun, or start more coding projects. The reason gym memberships work is that people like the idea of going to the gym a lot but don't really enjoy actually doing it very much and fall off over time, leaving the gym for the rare person who's really into fitness to use it cheaply.
is AI like that where most users will get bored? or is it like movies where they'll try to get their money's worth?
User's getting bored will also spell trouble for the subscription model. I don't think we'll see a world where the average normie is going to be running agents in a loop overnight to make software. They'll get bored use it to cheat on their homework and as a Google replacement, generate silly images for a week then get bored of that, and then realize the have no actual reason to be spending $100/month and drop down to a lower tier or just cancel all together and live within the free-tier limits.
So if that happens, the labs are now left only with power users that can actually generate asymmetric cost. The market will bifurcate into free-tier users and "get their money's worth" users. So the risk isn't that the top 5% of users use more tokens than the subscription has any right to give you, it's that everyone else cancels their plans and only those top 5% of users remain.
The labs will have to put stricter rate limiting and token limits on the subscription plans to avoid MoviePass style burn.
Gyms get away with it because most don't let you cancel easily. You pay monthly, but are locked in for 6 months or a 1 year at a time (Adobe subscription style). Maybe the labs will start to do the same? Let you pay $20/month, but you are forced into an annual commitment?
Especially if part two of your theory comes to fruition and they have to raise prices. I might of forgotten my Netflix subscription when it was 10 dollars a month but certainly not if it was 200 you know?
I mean Zuckerberg is a drop out.
zitron has been around since before flash was dead, more over hes worked in PR so like anyone with heavy involvement with c-suite, sales and journalism, you get a good nose for bullshit.
That said, do I agree with his assertions? no not entirely. But. I suspect its a bit of an audience capture thing. He makes wild statements, then goes on a flight of fancy to back them up. I appreciate that, but it also lets me decide that he's going down a path that I feel is not backed up by his data.
My argument here is just that Zitron says things that are wrong all the time.
You could fairly rebut me by saying I didn't substantiate that argument. That's true! That doesn't make the argument fallacious; it just means it isn't facially dispositive.
And I agree that he says things that are wrong much of the time, but not all the time. I also don't like him enough to care, but I do care enough to bring it up when I see multiple random attacks.
Either way it looks like he didn't just "repeat the numbers". He updated them and provided more data to back them up.
Yeah he uses more words to say it and bad arguments as part of it. But so did Galileo's "The tides prove that the Earth is moving around the sun" argument. Some arguments are bad, even if the overall point was correct.
I think given the new Nikkei report from last week, we are seeing more and more of Ed Zitrons arguments bear fruit, after years (literally) of him calling wolf.
It all is crazy talk until real numbers come out and prove him correct. But it seems to take a long time for those numbers to materialize.
Yes, I said as much. I've said this is two other places in this thread, but when they are talking AI being useful, they aren't just talking about for software development. The software industry is seeing far more AI adoption, by a very large margin, than any other industry, and the success of the current companies hinges on it being heavily adopted everywhere. I don't really care to defend Zitron because he does speak out of his lane far too much, especially when it comes to software, and those are the points the HN crowd latches onto. But that's not the bulk of what he talks about.
I didn't have a lot sympathy for the guy who described the Internet as a series of tubes yet I believe Al Gore caught far more flack for that bit about inventing the Internet than he deserved given the inventor of the Internet defended that very claim.
https://en.wikipedia.org/wiki/Al_Gore_and_information_techno...
So if the bulk of what Zitron's talking about is valid, then what are the main points that aren't self-evident i.e. there's definitely either a buildout or a bubble in CapEx and only time will answer that question, not hot air and blatant market manipulation. What is he bringing to the conversation that we're all missing? Seriously, educate me.
> So then what is the deadline for AI reaching into all these new industries?
Obviously the "deadline" would before the US companies go bust (again, IF they do). And I'm confused about your use of "new" here but I'm going to assume you mean "other existing industries that haven't adopted yet." The discussions around this are that several industries already jumped on AI then walked it back, so it's more that we're already past the "reaching in" stage. I don't have links for you but there have been several articles about companies firing people "because AI" then hiring them back. I suppose AI drive-thru order takers would work as another example. Again, I don't have the numbers on this myself. And again-again, none of this is "in defence of Ed Zitron" or "this is why Ed Zitron is great" or any such thing, but these are the things he talks about (and once again, no, it's nothing groundbreaking).
This is a very personalised use case, but I think everyone can find these kind of use cases that saves a lot of time for you.
This is completely unrelated to the how useful AI is for programming or how much more efficient you can be with it.
I never said you would need AI to do things you can do today. When I said it will not be 'all you ever need', my point is that NEW uses will come that will need more powerful AI. By definition, you can't NEED a new technology to do something that is already being done, because the fact that it is being done already proves you can do it without the new tech. However, that doesn't mean the tech can't do something new that does need the tech.
For example, no one needed an airplane before they were invented. However, you do need an airplane if you want to get somewhere 6000 miles away in less than a day.
There hasn't been a single bit of technology that is 'needed' if you use your strict definition of the word, because obviously humans existed and survived before the technology existed. Being absolutely necessary is not what makes a technology persist or spread, that is an artificial bar to reach.
Some people have such a natural aversion to LLMs that they will make incoherent arguments as to why they should go away. There are plenty of legitimate concerns and critiques of LLMs, you don't need to articulate arguments that you would never make about any other piece of technology to argue against their usage.
The narrative that LLMs are essential to the future of the trade often feels like an assault on my professional expertise.
Whatever your experience is, mine is that LLMs don't work well for software development. I wish people who use AI would be more willing take that perspective seriously.
> People misunderstand the point of the tokenmaxxing time period, it was to force people to use AI so as to not have them stuck in their way, as some people are, and then to evaluate how it can help the company.
The obvious second effect is that if you have a leaderboard people will compete to top it, by running dumb loops or whatever.
That impacts how much you can learn, and costs you more money than you wanted to spend.