That it is so absurdly ambitious and so likely to run up against reality strikes me as really indicative of the quality of the envelopes these calculations are being sketched on.
I don't think people will be doing business with the labs directly. "Enterprise AI" will be distilled down into purpose built products, with the model just basically being a generic commodity, and nearly irrelevant to the enterprises buying whatever these products are much like how I don't care if whatever SaaS was built in React, Vue, or some other framework as long as it works.
Ironically, for as much shit as they get about Copilot, Microsoft I think has the right idea for the long game they just suck at execution. Copilot is the tool, integrated into the rest of their enterprise stack, it doesn't care what the model is behind the scenes (they already offer you the ability to choose between different models).
That doesn't really bode well for the labs and their trillion dollar IPOs, because they are effectively reduced down to being a developer framework.
Even within the Fortune 5 of the US if be surprised if any of them are paying more than $1B annually currently in total.
And then you can take the parent context into account. If they can just equip users with a slightly more expensive Mac and call their Dell rep to order a few thousand DGX Spark to handle the rest... Why would they risk their trade secrets and intimate details flowing into models that may or may not be trustworthy long term?
Most large enterprise have been burned by SaaS over the years in some way. I can't imagine there aren't architects in the large organizations that are truly weighing how to effectively use AI. And beyond that we're seeing more and more progress in SLMs and orchestration agents which become easier to run at scale on-prem.
AWS billing.
Probably the same reason the Gemini app is still well behind ChatGPT in consumer usage and adoption despite being preinstalled on android phones worldwide ? Why are people using GPT on Windows. There's even a copilot button on new keyboards!
Or maybe its the same reason Microsoft Edge is not the most popular Windows browser ? Maybe its the same reason Instagram threads did not even dent Tiktok ?
You are asking the question the wrong way around. People use and like what they like and have a strong preference to continue doing so.
This is just human behaviour. You don't need mind blowing moat. You begin to have problems only when:
- Users are constantly using your product unsatisifed.
- There's a competitor(s) with a significantly better offering that people are talking about.
Will Apple's offering be providing any meaningful/significant benefit over just using GPT ? If not, don't expect any miracles.
Judging by the announcements today about its integration into the OSes? They are offering useful things ChatGPT cannot offer unless they write an "everything app".
One can (maybe should) make the argument that this is the browser monopoly again, but given that the USA has seemingly no intention of ever litigating that question again even if the EU does, there are clearly features here that OpenAI is effectively locked out of offering.
Why do I care if AI is integrated into my OS when I can choose my preferred AI and it can use the OS directly?
The other day I wrote up some notes for a presentation. Opened Google Slides, clicked the gemini button, pasted the notes in and asked it to make the slides. Nope; gemini can only modify a single slide at a time in Google Slides.
Pasted the same notes into claude, it wrote a pptx file, I imported that into Slides, job done.
Being integrated into the product doesn't always mean a better result.
Not being integrated can be an advantage because it gives you the freedom to think outside the box.
Meanwhile an AI engineer embedded into an incumbent slide app team has to ask permission and get cross functional alignment for every little feature. And deal with neckbeard tech leads lecturing them on what the right architecture is
Open AI can match it but at what price?
Well, it can mean a lot of things, which is why Apple outlined plenty of specific use-cases and details of what they meant by "integrating into the OS" here.
If AI use via Apple represents 10% of the total that vaguely implies that the total AI market is worth around $10 billion per year (which admittedly seems a bit low), and if it is just 1% (which also seems low) then we get $100 billion per year upper-end estimate.
Which just is not enough to justify the current valuations of AI companies.
This is a bizarre way to try and estimate market size.
Maybe Apple’s AI represents 100% of the market! Maybe it represents 0.000001%! We can make up any numbers we want!
Anthropic subscriptions alone already dwarf your $10 billion number.
OpenAI's commitment to privacy is absymal relative to the sensitivity of the data people are dumping onto the platform. The CEO also has a reputation for being untrustworthy.
The biggest threat to ChatGPT's moat may be a brilliant marketing campaign by Apple that really gets people thinking about what platform they want to be upload their secrets to.
But not because anybody asked for that. It is part of the force feeding so execs can make this exact argument, and pretend that demand is there.
Now that you can get Gemini, operated by Apple (with the Apple privacy features that come along with that), why would you ever consider going Android/Pixel (outside of running GrapheneOS, but I'm talking regular consumers here)?
Google isn't even making anything on the deal with Apple. They pay $20B/year to be the default search engine. This is Apple just giving a $1B a year discount to that to be able to license Gemini.
If you're in the Google ecosystem like Gmail and Calendar, it is exceptionally refreshing to be able to use an assistant that uses that ecosystem, instead of iOS requiring you to use Mail or its own Calendar app.
I don't think there's any real gap between Pixel and iPhone on the things that matter: UX jank, battery life, camera. Even the messaging issue in the US has closed with encrytped RCS support between them launching. So now it's just an ecosystem question, which might be why Gemini is mentioned so much with Pixel.
But given how dependent OpenAI are on Samsung, it's hard to believe they will see a radically better deal in material terms.
I've been using Kagi Assistant for my AI needs, and have to say, Siri will probably replace it in the fall. The question will be, will I still want to keep Kagi for search, or will this new Siri get me where I need to be on all fronts? I need to start paying more attention to how often I actually use the search results vs just the AI summary.
There are things I didn't see Apple show and I wonder how Siri will handle it. One example would be basic coding. They mentioned LLMs in Xcode and Siri with the Shortcuts app and Safari Extensions, but I just had Kagi write up a webpage as a means to display a bunch of data it gave me. Gemini could also do this, so maybe it's not a problem for Siri, but it remains to be seen. There is also a question of what the experience will be like. ChatGPT, for example, handles writing up this code is a much nicer way than Kagi Assistant. Kagi feels more like the results I would have had from ChatGPT a couple years ago where it just dumps out the code in a block and any change is an entirely new code dump, meanwhile ChatGPT goes into a coding interface with a live editor. Going to Xcode feels like overkill, Siri will probably be not enough... so that's a gap in the market Apple may not serve. I assume there will be several things like this. The prosumer level of AI usage, if you will.
I could see the same thing being useful for the ultimate output of a lot of chats. For example, they showed Siri comparing specs for few different products. I used an LLM to do this once as well, but it was comparing 12 things with about 50 attributes. The table was fine, but what was better was asking for a webpage that let me click on the attribute rows I cared about so it could total up each column, which allowed me to easily rank them and better make a decision.
Once it can make html files, it’s a small step to have Siri throw it into iCloud, and make it web accessible. This would be more of a feature than something it would just do, but I could see this being used in the same way Google talked about making dynamic widgets to help explain concepts within Google Search. That’s dynamic coding with an LLM as well, even if people don’t know it. Apple wouldn’t even need to show the code, they could just save it directly to a file and open Safari. That’s essentially what their extension builder will do… write some JavaScript and load it into Safari.
I'm not saying that this is what really happens. I'm saying that believing a CEO is as foolish and as grounded in reality as believing Ed Zitron.
I don't, and that's the point, isn't it?
It's the keys to a substantial chunk of the kingdom for $1B a year. Literally they are getting, for a very small price, the right to distill their own models from Gemini.
Is there money in this for someone with a data centre? Possibly. Is there money in it for NVIDIA? Possibly.
But either way, that's not OpenAI or Anthropic, is it?
Here is a different interpretation: Apple bought the rights to distill and use a smaller version of one unspecified model in the Gemini family (there are many such models).
The distillation will be carried out at Google's data centres so that the original weights never leave Google premises.
For this to be keys to be kingdom it would need to cover all current and future models and would need to be very permissive with regards to distillation parameters and allowed uses of the distilled model.
I expect the reality to be somewhere between these two extremes.
Apple themselves have said there is usage limits, with a subscription upgrade for more usage. So clearly AI Labs are directly competing on that front, it's just a normal default/chosen decision. Considering there are defaults and still successful competitors (eg. safari v chrome), there's no reason to think that competition can't handle this too.
Edit: I want to add that Google is also probably willing to give the model away at a discount to its true value in exchange for guaranteeing that their primary competition (who has tons of cash) won’t have an economic incentive to enter the foundation model training arms race.
Most users who actually want these features for anything more serious than summarization and style updates will probably find value in a modest subscription or ad-supported tier of higher quality models, even if just for occasional usage. Apple can provide this, but once you're comparing features, for many Gemini/Claude/ChatGPT may be a better fit.
Oh, and I think there is an unfortunate but real risk that once again, apple totally over-promises here, and their AI models that they ship end up being pretty poor, and that drives users further into subscriptions.
OK, that I would concede is a possibility. Though Gemini is clearly capable, and the (alleged) story is that they have licensed a one-trillion parameter form of Gemini. I don't think they are making the same mistake.
ETA: I also concede they could make a different mistake ;-)
Specifically for image generation. They haven't indicated you have limits for Siri interactions.
Start at 1:07:00 in their announcement video. Craig is absolutely talking about "Apple Intelligence" as a whole in this segment.
Pragmatically, of course they'd need to add metering to any cloud available APIs that rely on large models. There's no way they will eat the cost of serving unlimited access to a cloud LLM to end users if they won't eat the cost of an image generation model.
Who makes consumer devices? Google
Who makes operating systems? Google
Who makes browsers? Google
Who makes the world’s most popular websites? Google
By the time 90% of average internet users get to chatgpt.com or whatever, they already went through several Google chokepoints, each layer is one more place Google can answer their questions.
And that’s not even getting into the chips, the data centers, the data, the talent, the consumer apps, the enterprise apps, the cloud platform, the brand, and of course the biggest cash printing machine in human history.
You would honestly have to be insane to bet against G.
Nah this is just Googler cope
Google missed the AI boat. Period.
Isn’t that more damming for Google?
Invent the boat, don’t know how to use it, abandon it, then someone else comes along and steals your boat.
But it isn't like this hasn't been the long-running strategy for Google as well - provide more results on search so that people don't go to the site with ads, provide paid product results for shopping, to offer more services to keep people providing personal/behavioral queues to Google and more opportunities for ad placement.
If anything, AI turned up the heat such that the frog noticed what temperature the pot was. But that doesn't really put them in a better position to execute than Google.
Unless you are confident that Google will be antitrust forced to weakening their hold in some of those markets. A crack is all you need for the water to rush in and widen the crack.
gobby ... British rabble-rouser. "Gob" is the Dick van Dyke approved word for mouth.
Would you care to wager on that?
Because I would gladly take the other side at even odds.
> consumer AI is worth, to Anthropic
Anthropic does not really care about consumer AI. I expect consumer is where their least profitably customers are.
My primary expectation is that Apple will mostly increase usage of AI by general consumers. To me, this reads like Instagram adding stories. Did it stop Snapchat's growth? Sure. But I would be cautious about claiming it will take too many users away from OpenAI. I think it will be a fairly different product offering.
If you're paying to use ChatGPT right now, you might be using it for hobby coding, projects, or image generation. If you're paying a lot for ChatGPT, you're almost certainly using it for personal programming projects.
The $100/month (and up) subscribers aren't going to churn because of this, and I would be extremely surprised if the $20/month users do in any meaningful way.
I don't gamble. Though you might not be alone taking the bet:
https://www.notus.org/technology/trump-blindsided-ai-compani...
"OpenAI CEO Sam Altman pitched the idea of turning over shares in his company to Trump in early 2025 and discussed the matter again with senior officials in recent weeks"
If you’re only giving even odds you’re not very confident in openAI at all. $15 billion is peanuts.
It's a steal at even odds
It's still positive EV at much worse odds but not worth my time
2. You can’t know whether it’s worth your time or not since you don’t know how much the wager was for.
If you’re 99% confident, 10:1 odds with a 10x larger bet is still a steal. My guess is your confidence is somewhere just north of 50%.
I just don't see how being the "premium" provider really works if much cheaper models are basically good enough.
I actively hate it when it brings in some nonsense it thinks it knows about me. I told it my income once in an attempt to use it to find the perfect rewards credit card mix. Now anytime I try to get it to search for a deal it brings up some nonsense about “as a high income individual you don’t worry about saving $X, you care more about reliability, so you don’t need to look for the lowest cost” or something similar.
I have iterated through different option configurations to reach a level of 'customization' that more or less conforms to my own use case, and this does include opting out of any and all lasting memory between instances and across chat sessions; and adds a selection of single initialization prompts which shape the chatbot's behavior to my requirements for that session's objective. these trim most if not all af the sycophantic interactions, reduce outputs to the specific formats and contours as defined and omits any of the 'explanations of the underlying reasons behind...' which is just noise. This also has enabled some pretty useful results without ever spending a dime on a paid account: the premium behavior presented to 'potential customers' as a lure continues to work for me, and for iteration across instances and accounts is possible with machine-ready yaml context file when a single sessions hits the 90% wall : one emit and ingest cycle rotation across account profiles in firefox and i pick right up with a fresh limit.
Bouncing between ChatGPT and Claude, and between models for discrete subsets of larger tasks has really been impactful for my particular needs; but as i am not working in regions of knowledge that are beyond my own expertise and because I require the model to limit responses to very specific parameters, the logic space for unchecked hallucinations is low (but not zero).
The most useful project results for me have been in developing an air-gapped private menagerie of multi-domain models which uses an operating structure not dissimilar to OpenMythos; but then my background includes HPC environment development for NUMA, unikernels, MPI and bare metal hypervisor design - so getting a design plan and functional code without requiring a team of programmers and months of time in order to even start using models under my control which have zero public facing risk for the projects i'm working on is a much better place to spend limited budget on. Last gen hardware in the V100 class is perfectly capable of running and delivering the physics calculation optimizations as required and I would rather buy and/or install solar+storage to supply the electricity for token generation than rent the same from any of the frontier models AND trust that "don't train and learn from me" preferences are and continue to be followed.
If your use-case is a a 'lifestyle shopping assistant' then just turning off customization might be sufficient to stop it from telling you how to live your best life.
Web SaaS gonna end up being seen as another failed play at slim clients and entirely centralized sources of pay to play access to eyeballs; more AOL-ification of networks
Honestly, I don't think I need anything from AI at all. It's a convenience but it doesn't really enable anything I wasn't doing before. That's probably their biggest problem. The biggest thing is it enables non-coders to write code, but it's very debatable that that's a good thing excluding personal projects
If you are a consumer and you have a Mac or an iPhone, what do you need from AI that Apple's new offering won't provide? Why would you pay for ChatGPT, or even tolerate its inevitably increasingly desperate ad placements?
Assume Google will have similar tools in their phones, and Google search will continue to have the offering it does.
In short, where is the evidence that once Apple's tech exists, consumer AI is worth, to Anthropic or OpenAI, anything noticeably more than that $1B a year?
Maybe OpenAI strikes a deal to put something in Samsung phones. Let's say Samsung is ten times as desperate as Apple (which is how it looks, often). Still only $10B a year?
2026 consumer revenue projections from OpenAI are pitched at $14-15 billion, apparently. If they get that, it's the only year they will get that, because by late this year, everyone with an iPhone will have something useful built in.
Ed Zitron is a mouthy British rabble-rouser, but I think he is probably mostly on the money.