Apple, to me, is the winner in the end of this AI chapter.
Distillation works: we can get models leaner, have better performance with lesser and lesser parameters, more efficient, running locally. Apple gets this and we'll see the result with M7 Max/ultra chips and I reckon this will be one of the titular apps when M7 line of chips are introduced.
I guess it explains the haste Anthropic and OpenAI are in to IPO, cash out and leave the the retail investor holding the bag.
I think the question to me is whether Apple can ship affordable consumer devices with enough memory to run sufficiently useful models sufficiently fast.
Right now 128GB MacBook is the top of the line and you could do Qwen 3.8 Flash or similar.
There is a rumor that the M7 chip can use 1.5TB of memory which would imply Apple is aiming for more memory on devices.[1]
Would that imply the entry-level MacBook will have 128GB and top of the line models could run larger reasoning models? Then the question is price and if any of this will translate to iPhone for on-phone models.
With how expensive memory is, IDK if they can pull this off but I hope they do
Even if it's 2030 - that's a mere 3 years from now - it will take at least that long for the greater economy to figure out how to use LLMs. And probably a memory boom/bust cycle in that time too.
If Apple, in 2030 has a compute product that a worker can BYO(X) to work that has 2TB RAM, with an encrypted personalized context that can float between agents, then they will be in a very interesting place.
Why is there a desire to have this stuff on the edge on the engineers lap instead of just building some arbitrarily scalable cluster for I'm guessing cheaper than apple's usual premium? Habitual practice from the engineers who would compile on the device getting hot in their lap? Lack of widespread hpc experience compared to running docker?
Because why not? If you can get unlimited inference with roughly Opus 5.5 intelligence for $10k that easily beats current API pricing. Like not even a question.
Many users would prefer to use a Sonnet-level (or better) LLM without sharing their data with others.
And with a device like this, you pay one single time to buy the product and then never have to think about paying for tokens ever again. For ordinary users that's silly, but for users or businesses chewing through lots of tokens, this means they can create their own turnkey replacement for "the cloud" where the only cost after purchasing the machine is the electricity to run it (which with these machines is very little). Mix that with the privacy benefits and suddenly it makes a lot of sense.
Seems everyone in this thread is assuming I am saying go to private third party compute. No. Build your own cluster on metal you own on property you own, and have your team use that and just some thin clients to write with. Your compute goes way further. Your cluster can be going 100% of the time with jobs queued up by your team. All that memory you buy for your team in macbooks is idle and unavailable when your employee closes the lid and tosses it into the bag into the car on the way home. It could be used 24/7. On hardware you own. For cheaper to build I'm sure than apple hardware tax. On easier OS to manage and no doubt deal with anything involving any sort of heavy compute work. Are we really going to waste resources with docker when we could be using a node completely? Come on.
Like I said, seems HPC is lost art. All four sub replies missed that in my original comment.
I agree with you, and i would like to set up a server at work for local inference but we simply do not have the knowledge, it's much easier to get a couple of AI Max machines that are good enough for me and the boss which i convinced on the good points on running clankers locally and be done with it (though my plan is to use the AI MAX machines to figure out what the infrastructure and workflow will look like, and at some point ask for the investment)
It's investors making these decisions. They would rather have predictable cloud spend than trying any experiments with high cap-ex. Has been this way since AWS is a thing.
Your main argument is about better allocation for that RAM, which only makes sense in the conditions of a shortage. If RAM is plentiful, like it’s bound to be once the datacenters craze is over, that argument becomes “why does every apartment need a washer/dryer when we can have shared washers/dryers in a common area that are being used 100% of the time”. The answer is “because we can and it’s not that huge of a difference” plus you can have both. Like you can have a server cluster for large batch jobs, and personal ones for personal use.
For individuals though, it’s the same reason plenty of people pay for the 2tb or 4tb SSD for their laptops or desktops even though NAS has been a thing for 40 years. Maintaining a “server” of any sort is a completely separate endeavor for most. Just like a NAS is a no-brainer for many and non-starter for many more.
RAM was dirt cheap before and yet until M3 Apple made laptops starting with 8 GB of RAM. Why do you think Apple will suddenly become generous later?
The problem with LLMs outside of server is that people typically have many devices. So you can run LLM on your laptop. You want to make a query from your phone. Start your laptop? You want to do something from your smart home? Start your laptop?
What about your partner? Will they need to buy another top of the line laptop for themselves?
Your kids?
Alternative to server is of course that everything will have over 100 GB of memory, eat battery like crazy, be dumber and slower than single server in your household.
4 TB ssd was like $300 until recently. Even if RAM prices were to go to previous values, I don't think AI capability will be any comparable. Besides you also need proper compute, unless you want your prefill to be snail slow.
It has nothing to do with generosity. Apple (and other laptop and PC makers) just have different segments of the market that they catered for. Plenty of people use laptops just to browse the web. They didn’t need a lot of RAM. Those who needed the RAM (usually branded as pro-users or “creators” or engineers) always had the option to get more RAM both from Apple and others at a premium. It’s just that ~32GB was the upper amount for almost any use case. Why would Apple put 256GB or 512GB in every laptop back when there was no use case for it?
Every scenario you described applies the same to a NAS. What if you need a file from your phone? Open your laptop? What if your partner needs a file? Open your laptop? The answer to all of them is either yes, or they have it on their laptop/phone too.
Apple (and other laptop makers) weren’t charging $300 for 4tb even before the shortage. They realized that not everyone needed that and they way overcharged for that but still less than today’s prices. People who needed it bought it even though it would have been a lot cheaper for them to build a NAS. A current-gen M5 is perfectly capable of doing decent tk/s given enough RAM. In 2 or 3 years it would be even more capable, it just needs RAM.
Plenty of people will use clou hosted models like they do with storage today. Plenty of people will build home servers for AI. Just like plenty of people have home NASs. But the number of people who just want a large SSD on their laptop make the previous group a rounding error and those people would rather have a beefy laptop or desktop with enough RAM in it. If anyone can solve the Phone<->Desktop communication seamlessly it’s Apple after all.
Unlike storage however, a lot of people are now aware of how much data collection tech and AI companies are doing and the idea that AI models are training on their personal photos and prompts is deeply unappealing to many. With just storage, what could Google or Apple do? Certainly they wouldn’t just print your photos and put them in an ad or on their website. But with AI models, it’s a different equation
> Every scenario you described applies the same to a NAS.
No. Files are your property. Inference is a service.
There's also the fact that working with files over network is often ill-advised and borderline impossible for some workloads. Even when that NAS is connected over wire and not wireless. Meanwhile even if you're on so-so connection, LLMs work great.
I don’t want it in my laptop, I want it in my mini server rack at home. I’m not running any of my personal data through an external service that is certainly collecting it as training data.
I want it in neither my laptop nor my home server, which I don't even have, I want it in my phone or something even smaller. I understand that it's currently not possible, but this is the direction I want the technology to be moving into.
Highly likely it will happen. But at that point the phone will likely just be a slab of display and not much of an interface. Everything is done through an LLM and showed on the display. Not sure if touch screen will even be required at that point.
First of all, it seems to be possible to build a scalable cluster from TB5 Macs. They don't have to be on individual desks, they could be hosted in a data center.
But on the other side, equipping data centers was never Apples main business. Their approach to markets is to design products for the individuals. They also usually have very long time plans. Look at where the Apple Silicon today is and how long a path it took. What about the day, where at least a medium sized Mac becomes a great platform for running large models locally? Very easy to use, no setup, no privacy concerns. Also never worry about "tokens". You can then use AI as you use other local resources of the hardware you own today.
The Mac Studio hardware can do 512GB. That might be a hardware limit of the current architecture.
But the family shouldn’t be limited by that. We’re not going to see M6 Max hardware, because the M7 line is being fast-tracked and pulled forward. And I wouldn’t be at all surprised to see M7 capable of 1-2TB or more.
To constrain supply? Or to punish unreasonable demands?
If Apple can't place DDR5 orders that outbid the HBM demand, then yeah, Apple Silicon is going to get fucked over by every vendor. They'll be laughing their way to the bank while Apple begs Californian courts to give them free RAM.
I dunno, I reckon they are targeting 64GB and will develop a smaller model thats built in by default that handles a lot of simple tasks quickly.
Probably sitting under a router, so that it can hand off bigger jobs to an Apple DC somewhere, probably with a certain number of credits based on your iCloud sub.
Apple’s position down the road. When the Chinese company CMXT takes a much bigger share of the memory market worldwide, and they will. Their dominance in memory will probably be pretty close to what BYD is today in the worldwide car industry.
I don’t think Apple can be in a position where all of the Chinese device-making companies can get memory at a significantly cheaper price from CMXT, and Apple is stuck with buying memory from the three hyenas of memory at a significantly higher price.
That’s not going to help Apple sell their devices at a competitive price worldwide over time against the Chinese computer makers (Lenovo, Asus, and Motorola) and make no mistake about it: their share of the market worldwide will get bigger because of this change in the memory market; their focus will be on the long term, while the lawyers/MBA/finance types in the West will have their heads stuck in the short term.
It’s the (Chinese companies’) opportunity, and they’re not going to let it pass; I wouldn’t. In addition, Apple is in competition with Nvidia, Microsoft, Intel, AMD, Qualcomm, Google, Meta, and indirectly with Anthropic and OpenAI (whose CEO, Sam Altman, recently declared that computer ram is wasted on low-level regular computer users), along with a whole legion of other 3D model makers on the procurement of memory.
The combination of all these elements going into the future means that if Apple wants to stay ahead, I believe Apple has no choice but to take memory in house, they have the design and engineering capability in-house, and they have the money because they didn’t burn it on AI models and data centers.
This scenario is no different than replacing Intel or Qualcomm; you can see the writing/destination and the ultimate train wreck up ahead if Apple doesn’t roll up its sleeves and get busy, the market has changed and again Apple has to change and adapt to the new conditions.
You're from hackernews, hackernews is terrible and you're a terrible commenter. You should be ashamed of yourself.
Why are you talking about memory? Nobody cares about memory. You know what people care about? People care about ballrooms, and you know what, the US has the best!
People who know ballrooms say the US is hot right now, hotter than anywhere else in the world, because we're building the best ballroom in the world. Can you imagine that? China is nowhere on ballrooms.
If I recall Apple was planning on buying it pending permission from the US Government, and when the memory maker got wind their price sheet went up to match Samsung, making it pretty pointless as they had supply, it was price that was a problem for Apple.
> The combination of all these elements going into the future means that if Apple wants to stay ahead, I believe Apple has no choice but to take memory in house, they have the design and engineering capability in-house, and they have the money because they didn’t burn it on AI models and data centers.
I wouldn't be surprised if that was happening already -- they plan five years ahead, after all, and they've been slowly taking other systems in-house (e.g. baseband/modem) already.
> Would that imply the entry-level MacBook will have 128GB
Most people neither need nor want to run large LLMs locally.
The entry level MacBook Neo costs $599. Even if RAM prices revert to the pre-AI mean, the entry level models are not going to be that stacked with RAM.
Eventually we'll get to a doubling where entry level means 16GB of RAM, but most people won't be running anything more than simple models for light local work.
Have you ever tried running large LLMs on a MacBook Pro? It's hot, loud, and makes your battery level disappear at rates you didn't know were possible if you're not plugged in. It's a very niche hobby.
I'm old enough to remember in NYC when people knew the number of their local car service and that tools like Uber were 'dumb' because how could they compete with something that was already so simple ...
Don't all models block that, and also not train on pr0n data?
First someone would need to invest and train a model on pr0n, and then they'd likely first want to recoup that money - perhaps by selling a subscription to pr0n addicts.
The first time my family bought standalone RAM to upgrade our PC, it cost over $100 for 2 MB. Not GB, but MB. Consumers will get much larger stacks of RAM at some point, likely sooner than later.
If you select with care you can get fully-functional abliterated models. There are good abliterated Qwen 3.8 27B models and quantized versions thereof. Heretic Grimoire models are a good starting point.
> whether Apple can ship affordable consumer devices with enough memory to run sufficiently useful models sufficiently fast.
i suspect that they will at least do offloading for casual usage as well. if they can ship 24GB in a homeview or appletv style product, it's game over.
> Would that imply the entry-level MacBook will have 128GB
why would a company who locks iPad Pro entry level literally out of ram that exists on its own chips.
I get their high end 10k-20k laptops will be able to reach these specs and gatekept behind also upgrading the cpu to the ultra but no way will it be an option in entry level.
Cloud compute has been completely commoditized. Yes, local models will get better and better, but I think for the vast majority, it will continue to be both cheaper and more productive to use cloud based models, far into the future, where bulk purchasing, oversubscription, and discounted industrial electric rates can be taken advantage of.
Where I am, it's cheaper to purchase multiple claude subscriptions than it is for the power bill from running a local model. Factor in the hardware costs, and you're looking at a break over a decade in the future, for comparable cloud models. Yes, there's privacy, etc, but now you're down into fractions of a percent of users who care about the government reading the memory of their computer.
Take a look at how qwen3.8 27B compares on benchmarks to much, much larger models. A while back I called it the crow of AI models: paradoxically strong given its tiny size, like the brains of crows and other intelligent birds.
It's not comparable to huge models of course, but given the power/size ratio here it shows that serious gains are on the table for models that can run fast on a high-end laptop.
I agree with you, even if many of the pundits say that Apple was left in the dust.
At some point, the real cost of inference/intelligence is going to come due, and the existing open source models will be good enough for 95% of tasks users want (or training their own model(s) will be inexpensive for Apple).
Apple will be there, as usual, to provide secure, on-device, faster inference that the vast majority of their customers will be fine with. Let everyone incur the R&D costs.
You don't have to sell it as a local model. Sell it as offline capabilities. Sell it as private background processing while your phone is idle and plugged in. Sell it as un-enshittified LLM once Anthropic and OpenAI inevitably screw everybody.
Sometime I wonder if other people on this website lived through the same era of web applications as I did, where people built whole successful businesses on 3rd party APIs, that were destroyed overnight when the terms changed.
Yes, I promise you, there are many, many people who are hesitant to build anything real on the assumption that either Anthropic or OpenAI will continue maintaining their services as they are today in perpetuity.
Both the OpenAI and Anthropic core APIs have become widely supported standards, so any application you ship today on their models can be trivially ported to other hosted providers or to your own infrastructure if you want to own the whole stack.
I agree that it pays to be wary of the additional abstraction layers they offer that purport to make your life easier, but for anything running standard completion calls or the like, I don’t see why this needs to be a concern.
> I guess it explains the haste Anthropic and OpenAI are in to IPO, cash out and leave the the retail investor holding the bag.
I don't think so. LLM inference enjoys huge economies of scale. At no point in the foreseeable future will a local model be able to compete with a cloud model in terms of price. Local models will have the privacy angle, but very few people will be willing to pay top dollar for that. There will be some people, for sure, but the vast majority will just use cloud models.
Most people just Google something or use the free ChatGPT. They have no clue how the answers get there and frankly would be just as pleased with GPT 4o as they would be with Astra. Questions like “what does this light mean on my dashboard” or “which bidet is best”.
For inference alone they get far better utilization of the hardware because they can keep it fully loaded with maximum concurrency. Other than that they get to collect a lot of data for training.
The larger Gemma4 and Qwen models are competattive with GPT3 / 3.5 maybe 4o, so you can run the original ChatGPT experience on a nice consumer device now.
I am using everyday Gemma4 31b and Qwen 3.8 27b + 3.6 35b MoE.
Beside the 35b, the others are too slow to do anything vs cloud models, unless you have a local B200+. And the Qwen 35b has the reasoning capacity of a 2yo doing maths.
I wasn't "impressed" with any models for coding until 1year ago, only then they started to be "usuable" with the fault that we know and still exists.
You'll still need a cloud AI subscription for the remaining 10%, so you might as well enjoy the superior performance. If you're cost sensitive, models like Deepseek have you covered.
Small models inherently have less pre-trained knowledge, which will hurt their performance on queries like this. Moreover a lot of queries benefit from doing web searches, especially if they need to respond to changing conditions such as your bike example. Now you have to send data through a data center anyway for good performance, and probably need to use an API if you don’t want your computer blocked. So what’s the point, when GPT Luna exists if you really want to get a decent model at a rock-bottom price?
I don’t need or want “pre-trained knowledge”, I need a computer program capable of searching the web, sifting out the garbage returned in the search results, and giving me a summary of the useful information it found.
Unless cloud PCs take off and kill "AI PCs" and the future of the end-user hardware market, I think it'll be a bit of both.
It'll become easier and more affordable to distill/train/optimize smaller local models to meet your specific work needs. Multi-agent efforts will orchestrate the balance between using on-prem hardware for most tasks, and then defer to cloud models for more demanding/edge-case tasks. Especially if an organization can manage what data is shared with which model.
I also expect cloud providers to eventually start charging different rates for on-peak and off-peak usage, which would incentivize running what you can with the local compute you have otherwise sitting idle across dozens to hundreds of machines in your fleet.
I could be wrong and all your future hardware purchases will be glorified thin clients, with less local processing capabilities than today's laptops, servers, and towers.
My guess is it will probably be both. There was a time when no one could envision computing without the mainframe.
In the mainframe's heyday, it was difficult to grasp a world in which compute became so dense and so cheap that it made sense for each individual to have all processing and memory on their desk, talk less of on their lap and in their hand.
Of course this time could be different. But, when we look at previous relentless progress in compute and efficiency, then also couple that with the breakthroughs AI itself unlocks, it's hard for me to imagine that centralized AI will meet both the scale and flexibility desired for all use cases.
Just as centralized compute never fully disappeared -- it took the form of client-server computing and the cloud, there's no question that centralized inference will continue to exist.
But it wouldn't surprise me if, ten years from now, 90% of global inference happens at the client.
While simultaneously, the 10% of inference that takes place in the datacenter is wildly lucrative and orders of magnitude greater than today.
Put simply: the inference pie may become so large that both client and datacenter win.
> Put simply: the inference pie may become so large that both client and datacenter win.
Beyond inference, even local HPC or large in-memory workloads. Steve Jobs dreamed that the NeXT Computer could be used to cure cancer in a dorm room (whatever tha tmeans), and I guess a bunch of Mac Studios chained together wouldn't be bad for a lot of computationally tough biomedical "basic science" applications.
Eh, I don’t know about that. With current Sonnet API pricing, an M5 Max w/ 128 GB can break even in a few months.
Qwen flash-next is at least as good as Sonnet for my uses, and the speed is perfectly acceptable.
I really think the big labs are going to soon start feeling pricing pressure for everything smaller than Opus-tier models, and it will come for those next.
> Eh, I don’t know about that. With current Sonnet API pricing, an M5 Max w/ 128 GB can break even in a few months.
Only if you're comparing apples with oranges. There are no local LLMs at 128GB RAM which compete with Sonnet 5.5 intelligence. And the gap is a chasm when comparing against Opus 5.5, which is fast and cheap and plentiful on event he $20 subscription.
Don’t say it too loud, I’m planning some arbitrage moves in the future based on what I consider is the absolute wrong vibe on here that thinks local llms are going to dominate in the near future
Does it actually? Sure, it is operating at huge scale and this design choice allows enormous models (presumably enormous MoE models). But is it achieving economies of scale?
Economies of scale are rooted more in the idea of a profitable thing becoming cheaper as the market grows, allowing cheaper manufacturing and further growing the market.
The big two or three LLMs are cheap because they are substantially subsidised by investor cash, and they are not profitable at all in any sustainable sense: the current pricing can only be sustained for as long as they can raise money, and the market will (and does!) shrink when it is asked to pay a less-subsidised price.
There appears to me to be no route to true profitability for this design strategy at all. Even the open weights router companies are downstream of largesse from state-affiliated companies.
There is simply not adequate proof of this for Anthropic, certainly not for OpenAI. All of their figures are distorted by ignoring the costs of training, free tier usage; even seem to ignore the spent cost of training the current models, let alone the future models.
In the end, I think companies like OpenAI and Anthropic will be the nameless and faceless backend provider, and they will get decoupled from the users. At which point, if they haven’t already had their multi-trillion dollar IPO and everyone cashed out, I don’t think they’re going to.
Apple and Microsoft and other companies will own that front-end experience with the customer, and it’s not going to matter to most customers whose backend gets used for what.
> Apple gets this and we'll see the result with M7 Max/ultra chips
I'm not sure which chip it'll be, but I now for sure I'm not buying new devices until the AI boom has somewhat settled and we kinda understand what we'll be using it for. Until then, every manufacturer will just be guessing what device to build and will be outdated within a few months.
>Apple, to me, is the winner in the end of this AI chapter.
Could be the winner, in your opinion. They are extremely well positioned with hardware, devices, ecosystem, privacy, brand value & trust, and a significant active base. So far, they have not yet delivered on any useful AI application. There are so many opportunities in both MacOS and iOS, I guess they just need a new software VP.
NVIDIA has not delivered useful AI applications either, yet it clearly is one of the main AI companies, and while many are struggling to turn their AI applications into reliable revenue streams, NVIDIA has a viable business model.
So does Apple even need to create applications (that nowadays will get copied and opensourced in a day), or the hardware is the winning strategy in the end?
NVIDIA GR00T, Omniverse, Cosmos. Their local AI stuff for streaming is good and free, and while not a revenue stream, it's all just more reasons to go with NVIDIA cards over others when building a gaming PC.
Right. So far hardware has been the real winner. The rest is still a promise.
I think the more interesting question for Apple is the split between cloud and local. Economies of scale and aggregation vs privacy and ownership. I’d say so far it’s 2-1.
Tbh this applies to more or less the whole industry. We have incredible firepower yet I'm seeing probably even less product innovations than before the AI era started. Mostly R&D, benchmaxing and slop.
> So far, they have not yet delivered on any useful AI application.
Who has? Google, Anthropic, OpenAI and Microsoft have spent hundreds of billions and produced nothing but chat apps that Apple could build in a few days, they’re all just interfaces to models that are available to Apple via an API. Hard to imagine Apple as anything but the winner of this era unless OpenAI or Anthropic pull something out of the bag.
Google have spent hundreds of billions of dollars on destroying their golden goose, Apple’s inaction is absolutely a winning move that they deserve credit for.
The issue being is Apple should be spending money in the right places. Just because Google's product is poor, we shouldn't pretend Apple is being inactive due to some genius move, they are being inactive because they don't have the skillsets to make the correct moves.
This is literally Apple's strategy, and has been for decades. They watch other players discover what works, then Apple makes a well-polished consumer product for the mass market. They have run this playbook to great success in at least music players, smartphones, smart watches, wireless headphones, tablets. Apple is literally never (since the '80s) among the first to market.
> they are being inactive because they don't have the skillsets
Rare to see an argument that Apple can't hire talent.
Apple were regularly first to market with WiFi, USB, FireWire, etc. The Apple of late are not first to the market and also have mixed success regarding quality and usability. They sell plenty due to to name but I would say the watch has been greatly surpassed, Vision Pro is a clunky solution for a portable headset, and the airpods max are a rather poor product all told not being remotely usable in day to day life in the outdoors. Apple have had no success with Machine Learning, Siri getting worse as it goes on is one key example.
They need a significant team creating their own model, (the talent of which has already gone elsewhere) and they also need to be working towards the future of photonic computing for running those models with success long term, again which will require a huge hiring effort. Considering it's clear they are not putting in that hiring effort it's obvious what work they have been doing.
This is extremely revisionist. For ex with WiFi, ThinkPad + a Lucent WiFi card was standard issue for years before any Apple product had WiFi. Apple shipped integrated WiFi, which is wholly consistent with the notion of improving what the industry is doing.
Connectors? Yes, Apple has a history of improving connectors. Other computers have had connectors since forever. In this case, Firewire was Apple's semi-proprietary alternative to USB. USB, which Apple put into their computers years after others had shipped it.
> They need a significant team creating their own model
Your judgement. It's entirely reasonable at this point to license other models and look for a better/cheaper entry point. Apple tends to enter markets when it is possible to make profit. It is not clear yet that the business of creating and running advanced models is a profitable venture. By which I mean no company has yet demonstrated the ability to earn a positive ROI by developing and running an advanced model.
> working towards the future of photonic computing
Again, nothing in Apple's history would suggest that they would even aim to do this.
This isn’t true they’ve literally launched Apple Intelligence with multiple apps and surfaces three times now and it’s flopped each time so badly they regroup and pretend it wasn’t the real launch then launch it again at the next WWDC.
Their intelligence was more of a support guru rather than a like for like. I would agree Siri was a failure even in comparison with Alexa etc but it's not a comparable service to what people are referring to. There is no Apple model to compete, there is no hiring of staff that could implement a long term strategy of using that model on alternative computing platforms such as photonic.
They provide picks and shovels to all the gold rushes. Others can capture the AI revenue - Apple can ultimately license and use AI just like anyone else. They certainly seem to be capitalizing on it already.
I mean personally I also think the Codex computer use is brilliant, as far as harnesses go though it’s okay and I’d say Anthropic is a tad nicer otherwise.
But this is all based on marketing not facts. Apple is as privacy invasive as the next guy. Obviously this shtick will work as long as marketing continues do to good job.
Apple is the most privacy-conscious tech company, period. See, for example, private cloud compute, as just one of many examples that Google etc would never even consider, because they want the data.
They also don't have 100s of billions of dollars of debt obligations they have to pay off in two years while not being profitable. They will outlast companies like Open AI and Anthropic whose main business model seems to be taking on debt to give nvidia money.
The bigger question is whether consumer normies give a toss about AI. If they don't, and that's how it looks today, then Meta, Apple and OpenAI are all losing badly.
They definitely don't have AI use cases that exceed a $20/month subscription. Hard to justify the amount of RAM needed for local AI for a non-developer's level of AI usage
> I guess it explains the haste Anthropic and OpenAI are in to IPO, cash out and leave the the retail investor holding the bag.
Anthropic and OpenAI want to exit because they know they have no moat... and personally, I believe that the sociopaths running the US at the moment want to provoke a great economic crash. The Evangelical ones because they want the armageddon [1], the rich fucks because they took the "you shall own nothing" part for real and want to be the new neo-feudal overlords. And the best way to achieve that is by tying everyone's retirement savings up into their bubbly assets and then crashing down the economy, leading them to buy up everyone's asset at fire-sale prices with the liquid cash they made during the exit.
Been using both DGXOS and MacOS for my local inference needs, and by far, NVIDIA's DGXOS is superior in every way. Apple have a long way to go to catch up on the tooling extravaganza that keeps the NVIDIA ecosystem vibrant and rewarding.
Not everyone can afford a top-of-the-line Mac and the electricity cost to run it is high. It will never run Frontier models, which you can use for the price of a Netflix sub.
Also a lot of people work from laptops and running inference on data centres for 10th of the price is a no-brainer. Local models are more of a niche and always will be.
MacBook Air + Claude max sub for 20 years would still cost less before a maxed-out Mac Studio makes sense (not taking into account running costs), even then it’s not a Frontier or as fast.
Apple, to me, is the winner in the end of this AI chapter.
I might be one of only few people who think Apple is a loser in the AI chapter.
I don't think it shows yet but I think the dominance of the App Store will end. I think a ton of iOS apps will be replaced by ChatGPT/Claude. I think there is a big opening for a company to redesign the phone from the ground up purely around an AI assistant instead of apps.
I think AI will severely weaken iOS, App Store, and the iPhone in the next few years. It was nearly impossible to compete against iOS and Android until recently because no one is going to redesign their apps for your new phone. Today, that's different. Many productivity apps are just being replaced by GenAI/ChatGPT/etc and services are making it easy for AI agents to access directly without an app.
Not to mention higher chip prices will eat into hardware margins.
> I think a ton of iOS apps will be replaced by ChatGPT/Claude
Sure, but I don't think it's usually the apps that people pay much money for. And on top of that, app store revenue is (from a rough estimation) 6-7% of total revenue (although it's high-margin). Apple would be entirely fine.
And if you're right that the phone of the future will be reimagined as centred around being an AI assistant: innovative device design, user experience, hiding technology, integrating software/hardware/cloud are all things that Apple is historically pretty good at (with hiccups and bad periods for sure, but other companies certainly aren't better in general).
Im not sure of the numbers, but I'd imagine a large portion of that are from their in-app-purchase fee, like taking cut from Spotify subscriptions and such. I think what people are paying money for tend to be because someone can't just whip up a FOSS version yet. Music licenses, branded games, cloud storage...
An LLM can probably whip up a music player app, but won't have the licenses to provide you the music.
An LLM can probably whip up someone akin to a free and single player Genshin Impact, but the whales want attention from other players.
An LLM can probably whip up some sort of SFTP from phone <-> Desktop for pseudo-cloud storage, but you still need to provide disk space
Their ability to profit "one way or another" is contingent on their ability to keep third party payments out entirely or make them so expensive nobody can use them - and that tactic doesn't seem likely to survive much longer.
> Their ability to profit "one way or another" is contingent on their ability to keep third party payments out entirely or make them so expensive nobody can use them
Why? Off the top of my mind (far from an expert and not thinking about it very hard), they can profit from AI in the following ways that have nothing at all to do with the app store:
- selling the best devices for local AI (which they kind of are with apple silicon and mac minis)
- selling devices (mostly phones probably) with excellent AI integration that are superior to the competition (remains to be seen, but that's the sort of thing that Apple tends to be better at than the competition)
- selling AI-powered services (not super convinced, I don't think Apple is excellent at services)
For example if they capture 10% more of the global phone market because iPhones became that much better than alternatives thanks to AI integrations, that'd probably offset decline in app store revenue.
Just throwing random ideas and numbers here, just making the point that shifting environment can weaken a revenue stream while strengthening (or enabling) others
Sure they can keep selling hardware and try to sell their own services, but none of these are as high-margin or scalable as fees on everyone else's services.
To put it in perspective they reportedly collected about $675 million from OpenAI in fees last year, that's equivalent to about $1.5 billion revenue in hardware sales and TBD for AI services but certainly over $1 billion.
The catch here is this revenue is already being factored-out of their stock price because of the current status quo in America where apps can use third party payments without commission for at least ~8-9 more months and possibly for good, and they have been rushing to establish alternative services revenue to plug that hole.
The other avenue that AI is hitting their services revenue is the Google Search deal, it's a revenue sharing agreement and AI is eroding that revenue on Google's side.
all things that Apple is historically pretty good at
But do they have a better, more efficient, smarter AI model? Can they keep up with Anthropic and OpenAI's models?
Not to mention, Apple is going to want to not disrupt itself so they're going to deliberately slow down which opens up the opportunity for someone else.
Does it matter whether or not they have a better model? Is frontier-level intelligence needed for the kind of day-to-day tasks people use their phones for? I’m not convinced this will be a big differentiator in this area.
They don't necessarily need to keep up. Apple's been working on building the APIs and underlying system that will allow AI models to be selected much like you select a default search engine in your browser. They've also been more focussed on developing small, special purpose models for on device processing of images, text, etc. and the silicon to accelerate them.
OpenAI, Anthropic, and others can spend billions on the infrastructure to train the models at a time when the technology is still immature and largely inefficient and then Apple can simply license the models as needed, giving the user the choice while potentially protecting their privacy. Once there's relative maturity in the industry, they can either continue to license the models, distill their own from them, or even train their own using others' infrastructure. To me, it's a smarter long term strategy even if it temporarily gives up their control of key technology.
Personally, it seems to be the smarter play and aligned with how Apple usually does things. They're usually not first to market but they're first to introduce a refined product that make people forget about everything that came before it.
> services are making it easy for AI agents to access directly without an app.
This is what we all want, this is what kicked off the open web back when the internet first started. This is what made everybody excited for user agents. Now we have user agents once more and we are getting excited about the new agentic web, but temper your expectations. This time, rent-seeking corporations are in it from the start, see the exact same promises that excited you about the open web, and plan to take full advantage right away. This is why everyone is trying to sell their own agent instead of giving access to your own.
I’m right here with you. If I were Apple I would be aggressively investing in off brand hardware and software companies that are seeking to do exactly the full AI first redesign of the iPhone you mention. This is how Microsoft pulled off a master stroke with OpenAI and ChatGPT, and this full redesign will never happen internally in Apple
You’re right, I guess MS would be doing a lot better, and staying even more relevant in this new age, if they never had anything to do with OpenAI (and the latter found their earlier funding somewhere else).
For the last 6 months, I've been telling my company that I don't think our app will matter much soon. I think most of our customers will want to interact with our service via their AI agent.
This is coming from my own experience as a software developer. I do not want to leave Claude Code. I'm hooking up everything into Claude Code and have it do what I want for me. I do not want to log into my cloud provider. I do not want to log into my Github. I don't want to log into any 3rd party services. I just want to stay in Claude Code.
I think developers tend to be ahead of the general public by 1-2 years in AI use. I imagine a future where most people simply want to stay within their ChatGPT/Claude/Muse/Grok/Gemini agent to do just about everything.
If my prediction is correct, I think Apple is going to be in trouble because it weakens the iPhone ecosystem severely and they are so far behind in AI everything that it's hard for me to see them ever catching up. Not to mention they do not have access to compute since they've not invested in compute.
> I think developers tend to be ahead of the general public by 1-2 years in AI use.
I'm pushing the perennial 'developers (and engineers) think differently' trope here, but I think you're wildly optimistic.
The inability of the average person to write a detailed enough spec for implementation, compared to how trivial it is for an experienced developer, is mind boggling.
Imho, this is going to extend to agentic interactions with app-as-service-for-agent.
You don't magically increase customer experience by multiplying (garbage average user prompt) * (whatever agent they use) * (MCP, etc).
Have people forgotten why software exists? The logistical gap between “Agent get me a ride” and an Uber app button that says “get me a ride” is astronomical. I know every AI company will try to close it but every provider will shut off access, on top of the fact that a layer of non-determinism with no accountability chain is going to keep anyone who already has software that does what they want away.
This raises two issues already, now you've already installed an app AND you've given an AI agent free rein of your entire phone? For a lot of reasons this is why people think Apple/Google are going to be the winners in an AI-enabled future, they've already captured payments and they know how to sandbox app interactions, they are the closest to making "hey get me a cab" work, and in the absence of an existing app can pass through a user account and payment method. Anthropic / OpenAI are not (in my opinion) anywhere near closing that gap.
Maybe I'm in the minority, but I would hate a future where I have to talk or type everything to an agent. It's difficult enough to communicate with people. Language is imprecise. For that reason, among many others, I think you'll always need traditional UI/UX, potentially some new concepts to adapt to new technology like AR, and supplemented by an AI Agent that can assist and augment rather than replace.
Apple's use of Siri in Vision Pro is a good start, but there would need to be tighter integration w/ the UI. That's probably on the roadmap... assuming they continue to develop it.
I think an AI can just generate a UI that you like. You can go back to that UI anytime. You can change it instantly. Your AI Agent still can interact with it.
Doesn't apply to most consumption based/scroll based apps like Instagram, Tiktok, Youtube, Hinge, etc. Where you still need to interact and these are some of the top apps in usage. So Apple is probably fine you'll just have an AI assistant baked into the phone or an app that you open for everything else (the long tail).
I think an AI-agent based dating service is a good idea actually. I might start one. :)
Instagram, Tiktok, Youtube, you're right. So hopefully OpenAI can convince those companies to convert their apps over. Wouldn't be too hard given how easy it is to duplicate an app nowadays.
> This is coming from my own experience as a software developer. I do not want to leave Claude Code. I'm hooking up everything into Claude Code and have it do what I want for me. I do not want to log into my cloud provider. I do not want to log into my Github. I don't want to log into any 3rd party services. I just want to stay in Claude Code.
How much longer do you give until Claude Code replaces you entirely? Product or CEO dictate their wishes and Claude returns a freshly compiled binary and offers to deploy it to production with a simple "yes". A SWE salary seems like an lot of money for a middleman to pull levers on Claude Code.
I invest most of my money at the top layer right now. TSMC a big one. I don't know if Nvidia will win, TPUs, Jalepenos, Cerebras or whatever is coming up. I just know they all have to go through TSMC. I have a bit in Intel as well just for geopoltical reasons.
I'll definitely be buying Anthropic.
Funny enough, I do have substantial amount of Apple stocks I accumulated pre-ChatGPT but I've been selling here and there as it reaches all time highs due it being an "AI hedge".
I think Wallstreet is getting it wrong and I think the narrative around Apple will change fast in the next few months.
What apps exactly do you think people actually use that could be replaced or reimplemented with LLMs on demand?
Essentially every single non-builtin app on my phone fundamentally requires the internet to work, and is backed by some sort of SaaS with a proprietary API. And even if/when the APIs are open, I doubt any normal person is going to decide that from now on they want to somehow use YouTube or Tiktok or Feeld or Spotify by talking to ChatGPT.
They're limiting their growth by somehow missing the initial AI train (even after buying Siri, etc.) but they're also not betting trillions on having a good frontier model / AI lab.
If a new mobile device form-factor proves to be usable, they'll come out with one eventually like they always do- like the iPhone duo being years and years behind?
When the bubble pops they could still snap up an AI lab on the cheap later.
I would say it was that there were structural reasons why they didn't want to expand to fit the new wider internet connected device category - not that they couldn't keep up.
They wanted to keep everyone on their proprietary platform instead of enabling their devices to be used on the wider internet / being a more open mobile platform.
As long as apple allows some AI integration, then I think they'll be fine- and in fact I trust their privacy and security model more than if I bought a Google device. (not to say I trust anyone's security model, but it's clear that Apple has a more vested interest in security and privacy than Google does)
I'm imagining a world where an AI-first phone is presented for the first time. Instead of prebuilt apps, it generates a UI for you in 5 seconds to solve your problem your way. And it works. It's fast. It's reliable.
It'll make the iPhone look like Blackberry.
And let's say that Apple wants to respond with their own AI-first phone. But wait, they don't have the model. They don't have the compute. Most of their employees only know the old way of doing things. It'll be like GM trying to compete against Tesla or BYD in EVs. Their whole workforce and skills are in ICE.
I just asked Claude to use a specific timestamp in a deployment configuration equivalent to 12 hours before 13.5 hours from now, i.e. 1.5 hours from now, and it got it catastrophically wrong. I don't trust LLMs with ordering food for me.
Sometimes even DoorDash gets it wrong, but at least I trust DoorDash to deterministically improve their systems so it doesn't happen frequently. With LLMs I would have to think that sometimes it just gets it wrong due to nondeterminism and insufficient reinforcement learning in the area of "reliable food ordering".
> I think a ton of iOS apps will be replaced by ChatGPT/Claude. I think there is a big opening for a company to redesign the phone from the ground up purely around an AI assistant instead of apps
We've already seen several attempts to make devices purely around assistants like Humane Pin or Rabbit. In 99% of cases users need a reliable screen with clear buttons not a conversational barrier between them and action
They mean it'd look like a screen like a regular phone but the OS is purely a chatbot. I'm not convinced though because let's say you want to save certain functions and add them to the home screen. Well, you've reinvented apps.
I think the direction was correct, but it was too early, execution was poor and the teams were inexperienced and underfunded.
I personally didn't like the voice-approach of Humane Pin. Rabbit R1 was more interesting but execution was poor, screen was not big enough, models weren't smart enough, apps and services weren't as open via MCP as they are now and in the future.
I want to see a company like OpenAI or a very well funded veteran company give it a try.
I'm waiting for lip-reading device, it has to happen. Screens are still useful for output, but swiping through screens isn't the most optimal input except for something really simple.
So here’s a gross generalisation: people don’t know what they want, and when they do know they can’t explain it.
I think folks who live in spec-driven worlds just don’t appreciate how poorly understood the capabilities and processes of computing and networks are, to the point where even if you gave people hallucination free LLMs you’d probably find people just frustrating themselves because they can’t figure out what they want, what it should do, and how it should be used.
Quite apart from the huge chunk of the app ecosystem that relies on common protocols, centralised services and so on.
people don’t know what they want, and when they do know they can’t explain it.
* Book me an uber or Waymo in 15 minutes. Which ever is cheapest.
* Based on our music discussions, make a playlist for me in Spotify. No songs older than 15 years.
* Look through my email, find all my receipts, show me a graph of my spend and categorize them.
* My business needs warehouse space. Find warehouse services within 15 miles of San Francisco. Ask them for their pricing and capacity. Here are pics of my items. Send these pics to them for quote.
* Find a hotel for me in Los Angeles on October 1st to Oct 5th. 4 starts. Max $200/night. Gym is a must. Don't care about pool. Close to airport. I have Platinum membership with Accor so try to find Accor hotels but not a must.
Only the fourth of those is not more easily accomplished using existing tools; the things you're describing as explainable are such because they've already been achieved.
> * Based on our music discussions, make a playlist for me in Spotify. No songs older than 15 years.
It is easy to express yes, but is this how people think about what they want to listen to?
Someone can already do this in a chat bot, see the list of things and then find it on Spotify or whatever their favorite music service is.
What the chat bot won't know is what they have listened to, which is probably a much higher predictor of what you'd like than the discussions you've had about music. You've probably had those discussions because you are already listening to that music!
That’s a surprising requirement. Is it a proxy for “less likely that I’ve already heard it”? For “the band probably still tours”? For “I can discuss it with friends without sounding like I’ve been living in a cave”?
> rustrating themselves because they can’t figure out what they want
I have seen this with yet "AI-unexperienced" users - for simple questions or textreformating, they are absolutely fine with the Chatbots. But thats all they can get out of it: Above that level their experience is very often frustrating due to too less understanding while having high expectancy towards the model in reverse (like "its AI so it should know...")
And this people end up then not using any of the AI tools beyond that because they do not "believe" / trust that this thingy can get work done.
> no one is going to redesign their apps for your new phone.
What about doing something like REA on a whole app store collection, rewriting the app's reversed description set into code according the new devices' requirements? Perhaps we can't see the future as there are so many variations yet imagined.
For instance; I don't think LLMs will be replacing Spotify, Facebook, Insta, or Youtube for my parents. They absolutely do not want to purely interact with a blackbox. They want to scroll through songs/shows/etc. they want to tap on it, and have the app do the thing.
Even as a SWE, I have absolutely zero desire to interact with an LLM for everything, but I'm not the litmus. I have found use with these tools for taking on some nested set of tasks, but at no point do I want to give up the app entirely.
I watched someone shop on ChatGPT, and it's far from just typing into a black box. This was not a technical person, they were just using NLP to create a curated feed in real time, with links to "do the thing".
For people who like shopping (not me!), most of it is actually browsing and there is a joy in sifting through options. The OpenAI designers seem to understand this.
It's still terrible or misleading for certain things though.
Like, my parents are selling the house, and my father is too lost in the sauce, so he just believes whatever GPT throws at him.
Apparently, GPT told him that there were 4 house listings in the area and so he asked my why the agency took down our house listing, even though he could have just went to <house_listing_website> and searched for it. Took me 3 minutes to open my phone and disprove him.
I think they are designing an addiction of knowledge. I think people want to feel like they possess all the information at a moment's notice (mixed with misinformation served as fact), without having to spend the effort to look for it
I don't think LLMs will be replacing Spotify, Facebook, Insta, or Youtube for my parents.
That's already happened. The curated feed you see is a machine-learning driven output of an AI model that Spotify calls a "Large Taste Model", supplimented by a host of models trained using Spotify's NEO framework. It might look pretty but it absolutely is the output of AI. I'd bet a fair sum that FB, Insta, and YouTube are also showing you the output of a model now rather than a traditional application.
That's not what I mean, I mean they don't want to pull up chatgpt and type "show me a list of songs that I might like" -> then have chatgpt generate a playback interface for those songs.
(unrelated; but this sort of paradigm would be heinously more expensive)
"I'll show you that when you're finished driving."
Please, no. Just TELL me; that's why I'm using voice to talk to you in the first place. Summarize and tell me, don't show me a list of links from a web search.....
Why wouldn't Apple catch this trend/want, through their enourmous QA and R&D departments, use their top-of-the-line dev and design department to build it and ship it on their state of the art hardware and use their world leading logistics and marketisng operation to get it out to everyone before anyone else even trying this finishes their R&D step?
Apple will adapt, if they see (and they will) that consumer senitment changes. Right now, 90% of regular "consumers" seem to despise AI and anything made by it.
They don't have the AI culture to develop models that can compete against OpenAI/Anthropic. They don't have the compute. They have a lot of baggage from legacy business and people.
1. OpenAI/Anthropic must purposefully not sell their service to Apple and/or users using Apple hardware.
2. Local models must not get any better or slimmer than they are today, such that they can not be ran by Apple or locally on Apple hardware. This is arguably not even the case today, there are open-weight models that rival the frontier models.
None of these two things are going to happen. Apple does not need their own models.
Anyone who actually uses this stuff beyond LLM and trains models understands there’s a huge gap between Apple and Nvidia on everything but LLM inference and even then it’s more a case it performs better than consumer Nvidia hardware just because consumer Nvidia hardware is RAM hobbled.
If Nvidia wanted they could end it overnight with a 60GB 6090.
Maybe I’m not the largest consumer of apps, but the only time I use an app is if I want to interface with a service or with other people. Apps are just the man in the middle.
Do I want to set up my scale to sync its data to my vibe coded app? Not really. Do I want to set up the dozens of visualizations and hundreds of different actions and thousands of different items for calorie tracking, where even a 5% error is extremely significant? No way. Am I going to convince any of my friends to download my vibe coded chat application? Absolutely not. Do I want to store my photos in my own vibe coded solution and trust it to run in 20 years? Not a chance.
While a segment of apps will surely become obsolete, I don’t think the majority of users time will be spent on their own vibe coded replacements. Even as a developer I have no interest in turning all my conveniences into maintenance burdens, with risk of catastrophic failure
It’s a mistake to think vibe coded apps will kill the App Store.
The issue is there will be less demand for apps. I don’t need to pay for a dulingo subscription when Claude can write a 5 minute language quiz for me. I don’t need to pay for Turbo Tax when Claude can prepare my tax documents for me.
Sure, it won’t replace all apps, but people will spend less time in apps and more time interacting with llms which hurts App Store revenue (Apples golden goose)
Replacing a good app is still extremely hard. Let me explain.
Commodities like todo apps and calculators rarely compete on features, they compete on product design. This includes look and feel, relation between elements, design choices on what exact common features are available.
All products are, at their essence, just a set of design decisions stitched and shipped together.
Now, making these design decisions is hard. It requires knowledge, consistency, and insights into how real people operate real products. Even as a sole user of the app you build you rarely know what you actually need.
LLMs are 'design-less' in their essence. Building your own todo app takes literal seconds and it all looks simple because LLMs will default most of these design decisions for you - until you try to use your new app daily and notice it somehow sucks. Building a good todo app or good calculator requires tweaking, publishing, polishing, and solving bugs, testing, rinse and repeat.
Another tricky side of this is that you might know logically what features you want, but you are not selecting one product over another logically. You need to try and see if that works for you, all the vibes and good parts and bad parts and design choices and colors, till you find something that "clicks" for you. Using pure Claude as a todo app only works as long as you want to have more Claude in your daily life.
In the meantime app store provides dozens of well-reviewed alternative solutions which look good and require zero maintenance.
And that's just a calculator or todo app. Non-commodity apps are even harder than that.
I don't see any future where programs follow the static nature of old paradigm programming. A nice intuitive UI? What about just telling the app what I need when I need it? The gold standard of consumer UI is...another person.
If I have a team of god tier light speed SWE's at my disposal, why the fuck am I fumbling around in an actions panel? The program can just be dynamically modified in near real time to be whatever I need it to be in that moment.
If you extrapolate LLMs out to their extreme, you get a shape-shifting liquid array of bits in memory that manifest anything at the behest of a completely natural language interface.
> The gold standard of consumer UI is...another person.
Absolutely not. Constraining the set of possible actions instead of having you verbalise what you want every time is often a feature, not just a technical limitation. A camera with a shutter button is better design than having to tell the camera “please take a photo”.
At the supermarket, I personally also prefer the self-checkout machines over going to a human cashier.
It wouldn't be just verbalizing, you could type or gesture just as well. Nor would you need to communicate each command, the machine would recognize that you want a button for your camera, the checkout would understand that you want to checkout and understand each intended action as you do so.
The UI is totally fluid at all times, because an "intelligence" is actively manifesting it at all times. Nothing needs to be hard coded.
You are assuming that you'll always have the time and energy to determine what you need at the moment you need it, which doesn't really sound true.
You're also assuming that there are no negatives to saying what you want compared to just acting on things, but since the beginning of computing many people have proven this wrong time and again. I suggest you read about HCI as a research field and its history, it doesn't take much to realise how limited the average HNer's understanding of UI/UX design is, and how far current LLMs are from the skills required to be able to carefully design an app for productive human use.
This sounds like the car industry touch screen craze all over again.
Good design rarely calls for a “you can do everything” approach. Good design is often context specific and subtly guides users to better outcomes.
Doctor appointments would absolutely suck if I had to navigate decision trees and click buttons to communicate my symptoms. Car blinkers would absolutely suck if I had to type “blink left” into a chat window before every turn.
>I don't see any future where programs follow the static nature of old paradigm programming. A nice intuitive UI? What about just telling the app what I need when I need it? The gold standard of consumer UI is...another person.
Maybe it’s just the ‘tism, but I hard disagree that I would ever say another person is a superior interface into… anything? It’s slow, they may misinterpret you, they have their own ideas and motivations, they’re (mostly) expensive because the only reason I need them is because there’s information / skill asymmetry. I don’t know that many or even most of those problems are resolved with ai, in some ways the problems are amplified.
Unless AI will be ar the level of saying "I want a scheduling app that fits me" I still see value in buying some apps.
I like to customize my Neovim setup or terminal and a bunch of scripts that I keep making them nice and maybe my blog and so on.
I don't want to become the PM of every app I need. But I do agree changes will happen to the current apps and how we buy and use them. Not sure exactly where it will settle. But still unknown unknowns and domain knowledge and vocabulary is not yet allowing anyone to build their own apps unless for simple cases.
Or maybe it is just me: I want someone else to think about my accounting app. I dont want to spend time to learn and understand and keep myself up to date or setup a series of LLM sessions that navigate and watch laws and propose changes.
Why do you need a scheduling app when you have a secretary?
Why do you need an accounting app when you have an accountant?
Programs of the future will in no way resemble what they are today. I don't even want to call it a "program" because in my head it's more like an intelligent digital liquid that can rapidly flow into any state. This flash-in-the-pan 50 year archaic era of software will look like what the oil lamp looks like today. A crude intermediate tech during a time of much larger transition.
You ask your secretary to give you a visual representation of your schedule right now and presto - you got it.
Same thing for your accountant.
Imagine that both of them, while being good at their respective jobs, are also software engineers that can mint a custom throwaway app for you at a moment's notice.
I tend to agree, on the whole. LLMs force more attention toward the purpose of an app and how best to satisfy that purpose for a given audience. (Undisciplined SWEs can forget about end users and get lost in the details of some particular piece of tech.)
However, you could expect LLMs to train on existing successful apps and hardened design patterns that compose well into good UIs. In that case, you might expect the last mile polish to be the differentiating factor. However, the more marginal this last mile becomes, the more other social, political, business, etc. forces will determine which app rises and which falls. But this has always been the case, certainly in software.
To train for business and user outcomes LLMs need what they currently don't have: open insights into the real world app and business metrics.
If you meant training on UI and features of established apps then it gets even more complex, as success of B2C apps is obscured by their pricing strategies and marketing, and business models for successful B2B apps like Jira/Confluence/Office often revolve around sales, partnerships and other non-product things like that.
Any business is a set of design decision someone made, with their app just being a small subset of these.
A huge amount of apps I've downloaded are just to solve a problem I have for a moment. Then most of the rest are regular use apps.
I dont think llms are going to replace banking apps anytime soon. But I no longer look for an app to pull data from my Peloton app into another exercise app. LLMs can build an app or a script that will do that and I wont have to pay some annual or monthly fee. It feels like these types of apps are going to die off.
I rolled out my own to-do app yesterday, because I needed something fast as a companion to Obsidian, while it can do cloud syncing for me.
It works exactly the way I want and looks exactly the way I want. If there's something I want changed about it, I add it as a todo item in the very same app. This works today.
I don't think the companies have moat in experience fine tuning.
I feel the same about this. There are many task trackers out there and most are overkill for my own projects. I don't feel like onboarding through Linear, or setting up through Trello and think about how to have the agents connect, etc, so I just found myself motivated to look for my own, local solution track and manage my projects.
I think it's better when you are at the origin, and the software branches off of your taste. For so long, we had to adapt and accept the ways of established software because it would take so long to make something of your own. Now it's both cheap and fast to iterate.
Design is sometimes important, but what we might call product-market fit (for a "market" of one) probably beats it. Sure, a calculator app is generic so you might as well use one that's well-designed. If there's an existing app you like, it's probably not worth rebuilding it. But personal apps can do things that an off-the-shelf app won't do.
I recently used a coding agent to build a TV-like YouTube player for my mother, who is quite elderly. It's inspired by channelsurfer.tv, but runs on the Google TV Streamer in my living room. Unlike any set top box I've seen, it has on-screen diagrams that tell her which button to press on the remote to change the "channel." It doesn't have ads. The UI will never change unless I ask the AI to change it.
You can't say that about most apps in an app store. They will do a UI redesign and you're out of luck if you preferred the old version.
Do you really think Google or Apple or anyone else will build this app? Maybe some self-hosted media box out there would be similar, but I can't be bothered to search for one and I'm doubtful that any of them were designed for an elderly parent who isn't good with technology. It's still work in progress, but at least she can use a single remote to turn the TV on and watch TV without being bombarded with things she doesn't understand and doesn't care about.
There are downsides that make this app less compelling for other people, but I didn't build the app for them.
But they are not designless. A great deal of their utility comes from the harness itself.
There's a recent benchmark indicating Hermes Agent and, I think, Qwen's harness are far and away the most effective.
Pi isn't bad but it doesn't seem as talented as Hermes. Nor does it currently array of self-learning/adaptive plugins seem as capable. That said, it's difficult to keep track of the SotA with pi; it grows like the node ecosystem in its first year.
This is a very good point, as one can argue that harnesses will evolve, learn to iterate, and will make custom and on-the-fly app creation comparable in quality to human designed apps, with just few user inputs. Effectively killing the market of pre-made apps.
> Commodities like todo apps and calculators rarely compete on features, they compete on product design.
Whatever "good design" is (and it varies by user), it's very easy to copy. Commodity apps complete on marketing. Companies who charge $10/month for an app, are probably spending $8/month in marketing/advertising.
Great take. I agree with this. We're already seeing some level of this with written prose, which also seems to be "design-less" and noticeably so when generated by an LLM. So many claim to have solved this but I have yet to see it.
At the end of the day, it seems we'll be left with a decision of using something made by a person with real opinions or using something generated by AI with opinions that are a distillation of past documentation fed into it.
We're in this weird spot now where we are transitioning toward all ui feeling the same, and only the people really tuned in even notice this. Once the general public starts noticing, I believe there's going to be a search for fresh, organic, opinionated software. Not that the software won't be AI-generated, but it will have a human truly sculpting it, like art.
So the question becomes, if people still appreciate human-made art in a world where a better image can be generated in a small fraction of time via a prompt, will they feel the same way about the software they use?
> I think there is a big opening for a company to redesign the phone from the ground up purely around an AI assistant instead of apps.
This always puzzled me. Here's how that would go:
1. A new company, NeoPhone, launches that is redesigning a new "AI forward" phone.
2. The customers of that phone start noticing that they do many repetitive things with their phone, that they want the ability to design some repeat experiences with their chatbots. Call these repeat experiences "applets." Neophone ships some Applets on their phone that they created, a default set.
3. Customer A looks at customer B's phone, and notices their different applets that they designed. They work slightly better than customer A's applets, because customer B is slightly better at prompting.
4. NeoPhone notices this too. They create a new thing on their phone, the Applet store, where customers can share their applets and rate them and it surfaces the best ones to all customers.
5. Customers start complaining that they wish they had more creative control over Applets. Maybe the API is a bit more restrictive than they need, or the Applet needs to do special things with the phone's hardware, outside the context of the AI agent. Or they find that Claude Opus Fable 6.2 is better at making Applets even than NeoPhone's available chatbots.
6. NeoPhone allows 3rd party developers to develop on its phone, launching an "App Store" and a "Chat API" that is totally independent of the AI agent stuff.
This just isn’t going to happen because of one thing: trust.
Would you trust access to your bank to a vibecoded phone of unknown origin? Backs will likely limit access as best they can for customer security. We already have banks who block GrapheneOS.
Look at the hardware wallet company that got sold and a spy chip to steal the contents was added.
I do believe AI is going to destroy companies like Adobe however.
You would be surprised as to how many people actually give unattended access. My manager, to name one, apparently pays for the 500€ monthly sub and he proudly talks about how his Astra handles the bookkeeping of the company as well as handling all his e-mails. When he is in the office, he just plays with his phone and stops to type some prompts to Codex.
This is the new reality that we must prepare to accept. You and I may not trust the companies, but the circle is closing in, and people around you are already on the "hype train".
> I don't think it shows yet but I think the dominance of the App Store will end. I think a ton of iOS apps will be replaced by ChatGPT/Claude.
Huh? I just took a look at my phone -- I have about 100 apps on there, though probably only use 20 of them with any regularity. I don't see more than a small handful that could be replace with Claude (if it had access to the APIs that the app does),
And even then, a chat interface isn't the optimal way to interact with many services. Random example: A simple metronome / tuner. Could I ask Claude to play a tick at a certain frequency for X amount of time, or a E-flat? Sure. Would it be easier than using the app? No. Could Claude design a metronome app for me? Yes, but I don't want to have to deal with that process -- I just want a metronome app that works and I don't have to spend time designing or fiddling with, and that works offline, and all of that is well worth the $5 I might spend to buy it.
IMO App Store had been dead for a while, and iOS is dying.
Top paid apps outside of F2P gambling apps(all the "games" remaining on phones are gambling apps because they've repeatedly lured in and punished all traditional games, definitively burning bridges), making hundreds of dollars per paying users, had been like $3.50 ad blockers. iOS app development had been a lot less approachable than Android apps due to Mac host requirement, and with LLMs generating Android apps and HTMLs with ease, the fact that iOS can't execute .apk will make it successively less appealing.
Personally I don't think things like agent-first OS that don't rely on third party apps will come soon - that will have to be built from ground up as proper OS kernels, and I think it will run on desktops first.
The third OS always existed, the problem with those is that telcos won't carry it and people won't use such a phone with the nation's pick of social media - Fb Messenger for US, WhatsApp for EU, Telegram for Eastern Europe or whatever, MAX for Russia(presumably?), WeChat for China, LINE for Japan, KakaoTalk for Korea, ...
... thinking out loud, the third platform has to be the one endorsed by a social media. Like the old HTC ChaCha, or like, idk, "Startile" slab device running TeslaOS with Starlink DTC SIM built in. I wouldn't touch such a thing, but something spiritually like that. Or like a "Rakuten Star" with AST SpaceMobile eSIM baked in. Things like that.
But that is something Apple can not stop, models will get better and better, and generating and replacing software will get easier and easier. That is out of control for Apple. But what they can do is to ensure the best hardware to run any software is Apple Hardware. If they achieve in that, they win in the long run.
There is AI for devs and corps, and there is AI for end users, a more convenient search widget. "Who took gold * in 2007 in X". "How long should I air fry French fry and at what temp". I use my ChatGTP sub like that too, that's OS agnostic.
The Apple platform provides certain benefits to me. Take a look at Matthew Cassinelli' work (great speaker), if you have time (and a Mac I guess):
The bits of AI mixed the platform affordance to simplify daily paper cuts by clicking on an icon is worlds apart from a Google SWE spawning 48 "agents" to parse lidar data to discover pyramids in Mexico. That's also cool.
Because fabs take years to spin up, decades to become competitive, and are a constant financial drain throughout. If (with great prescience) they started planning their own fab at the beginning of the RAM crisis, we should see Apple in-house RAM sometime around... maybe 2040?
I feel that with all the trillions Apple have lying around, they should've been thinking about their own fabs already when designing the first Apple Silicon chips.
> they should've been thinking about their own fabs already when designing the first Apple Silicon chips
Why, though? They had a sufficient mountain of cash to outbid everyone else for TSMC capacity, and no realistic hope of catching up to TSMC's process on their own fab
Well, outbidding everyone else seems to have worked up to a point, hasn't it. Of course, we could say that we're living in unprecedented times with regards to the demands currently being placed on fabs, but I'd say that I would've been worrying about TSMC being a fragile bottleneck for more than one reason, and for more than a few years now.
Also, I'm not exactly sure we should be claiming what is and what isn't realistic for a trillion-dollar company. People could (and probably did) argue that they had no hope of catching up to Intel/AMD, and yet here we are, with Apple arguably having surpassed both, at least from an energy efficiency point of view.
Apple has alternatives to TSMC. Intel and Samsung both have comparable process nodes, albeit at a pretty high cost and relatively lower yield. Apple would prefer to avoid paying competitors like that, but in many cases they have no other option. The iPhone Duo display, for instance, is sold to Apple by Samsung at a considerable per-unit markup. It was that, or no folding phone.
Realistically speaking, "catch up to TSMC" is something that every single EUVL fab has failed to do despite tens of billions of dollars in investment. Spending money is the easy part, developing high-yield processes over a timetable of decades is the hard part. I think Nvidia and Apple both agree that paying TSMC to build American fabs is a smarter use of their money if they just want excess fab capacity.
It would be slightly ironic if they did win. They were incredibly slow to adopt AI, and even now what they shipped with Siri is incredibly poor. Having arrived late to the party they stumbled, lied to customers, mis sold a product, got sued for it and lost, and then ended up slapping a siri logo on gemini.
But the hardware. The hardware is fantastic for local LLM's, generative ai, etc. And they seem to maybe finally be waking up to that, in true slow sleepy Apple fashion.
If they'd been a bit quicker about it we could have a higher tier iCloud plan that included a chatgpt plus style setup with the "its private" marketing tag that cant be provided by anyone else. Instead we get Siri being slightly less useless than before.
Totally disagree. I do not think local models will displace cloud models any time soon, and I think distillation has been demonstrated to NOT have better performance and is LESS efficient. Chinese models use dramatically more tokens and time to achieve dramatically inferior results. Watching deepseek spit out 20M tokens when Opus solves in <1M is always an entertaining window into "efficiency" and "performance".
Local models literally cost more in electricity than using far more powerful cloud models cost in inference.
When your options are "dramatically drain your iPhone battery for substandard AI, or have instantaneous cloud AI with no battery drain" what do you think the average user will choose?
And that's before we discuss how ridiculous the pricing is for home hardware that has a chance of running a halfway decent model at a halfway decent speed. Residential electricity + overpriced underutilized hardware means you're going to pay 100X or more per token.
I think the future prospects for local AI are looking worse than ever. It will be far more expensive for a far worse experience for the foreseeable future. It will fill a privacy/control niche, but beyond that, it's not a serious option
I think I agree but it will take some time before their gamble on local LLMs start to pay off and I don't think it will completely kill off Anthropic (though. I would question their trillion dollar valuation if I was not already).
Running local models on desktop hardware will always be behind the SOTA, and thinking that cutting edge models will lose value because people can run good models locally is kind of odd.
I believe this as well. There is an inbuilt market for local models. A big example is law firms for whom cloud models violate attorney-client privilege.
NVidia aggressively segments the market on VRAM. Apple’s shared memory model is a competitive advantage. Well, it will be when they hand the raw PFLOPS, which is coming.
It was pretty fast to get it set up. It could be better for sure, as it's not a first class concept the same way that it is in linux, but an hour of google search ai mode will get you where you need to go.
I know you can set it up to make it seem to work, but from what I’ve seen people say over and over, it will eventually end up asking you some question through a dialog box. And you will have no idea that’s going on other than all work stops. And the machine seems locked up or something until you connect a display and find out it wants you to click OK.
Hopefully it won't be another preinstalled bloatware app like Image Playground. macOS enshittification has been greatly accelerating, even giving Win 11 Home a run for its money.
OMG after 25 years, Claris gets something beyond FileMaker (among the most legacy software still alive)? Claris would not have the engineering talent, so why would Apple use them as a vector?
Currently FileMaker is a really expensive but sophisticated developer deployment environment. They stayed pretty true to the original vision in the mid 80’s when they merged their DOS database and added a gui layer for the Macintosh. I started on version 2 and it was at its best at when it went relational in version 3. Since then they have been focusing on deployment of graphical data applications and being a better database.
"Harbor is currently in a private preview under Apple’s Claris brand"
There's a name I haven't heard in a long time! Claris (always an Apple subsidiary, since being founded in 1987) renamed themselves to FileMaker in 1998, but apparently they rebranded back to Claris in 2019.
This could actually be interesting for "regular" people. FileMaker was great in enabling non-IT people and SMBs to develop custom software for themselves. Think HyperCard with tables.
Some of that software has been in use since the 90s and still works. Adding AI to the equation will empower a lot of people.
However, FileMaker is not really targeted at software engineers. In fact using it as a developer will likely drive you up a wall. It's things that feel totally fine to non-devs but completely insane if you're used to something else. E.g. I remember having to click a thousand GUI buttons just to edit a few lines in a script once (regular users just "record" them through the UI). It did all work fine though, very reliable.
They added some more developer-specific features later, but it took a long time. E.g. editing scripts as text(-ish) only came in 2015 (FileMaker 14) and an actual programming language in 2020 (JS, FileMaker 19).
FileMaker really could have had a moment when iOS devices took off since you could get a basically free native app just by having a database, but (I'd argue) got bogged down by poor pricing models. I would have loved to run it, but it's always been prohibitively expensive for anything not an enterprise, and now it's far too easy to spin up a CRUD API and call it a day.
I believe Claris got rolled into Apple when Steve Jobs came back, but Filemaker was jettisoned to Filemaker Inc because it had a Windows version. Steve Jobs was apparently reluctant to have any Windows software within the walls, but lost that fight with Windows iTunes.
A few years ago Filemaker, Inc took on the Claris name.
https://getlocalcode.app/ runs local AI models (larger ones of course, DeepSeek 4.1 Flash 1017) easy. You can download and try it now. The problem is elsewhere: you cannot find M5 Ultra that you can purchase and set it up today.
> Cluster for scale: “Connect two or more Thunderbolt 5-capable Mac computers and Harbor pools their unified memory into a single inference resource. Run models that exceed any single computer’s capacity, or serve more concurrent requests across your cluster.”
This seems like the main way it would differentiate itself over existing solutions like LMStudio etc.
I really want capable models local but it’s impossible for me to afford the hardware. If there is a magical way to run glm 5.3 without quantization at 80-100tps with a total spend of less than 10K I’d be happy but that’s impossible now.
> $200 a month is only $2400 a year. That’s 4.16 years until you spend 10k.
I don’t find it reasonable at all. Subsidized Subscriptions or api for cheaper models, is the best value.
If you buy your own hardware, then after those four years you still have that hardware, and it likely still holds significant value. Plus you know the local model you're running today won't change or suddenly degrade in four months, let alone four years. Whereas after four years of paying subscriptions and API costs, you have literally nothing.
Oh? How come the four year old 4090 RTX cannot be purchased for less than (or even at) MSRP then? Can you buy me one? I'm willing to pay $500 even if it is worth zero in your eyes, that's a stellar deal.
Sure, it has value to you, I'm the same. My main workstation is 10 years old. But try to sell it.
GPUs are holding value longer, for obvious reasons. But even a 5 year old PC or server will be out of warranty, full of NLA components, and will get pennies on the dollar of its original cost on the salvage market. Most businesses will have fully depreciated them and expect maybe to sell them for enough to cover the costs of disposal.
Cheap remote is probably not going to last forever. Memory prices will eventually drop. Local models will be more than enough for many purposes when the hardware is available cheaply enough.
While I might imagine that there are customers who would store their data in the cloud but not necessarily run inference on it, there are also those who have strict rules against any cloud-based service. Those are the ones I’m talking about.
"Harbor is currently in a private preview under Apple’s Claris brand, but it seems likely that we’ll see it launch to the public in the weeks and months ahead."
Claris?? Ok, this sounds like it's just someone's personal project. Something like this that was actually going to ship would be in a different org.
If true, it is an astounding story. That said, I'm happy to see that Apple ships a graphing calculator with macOS 27 (albeit, a different one). I may not use it much these days, but I have fond memories of it when I was in university.
This is what I've wanted for the last few years. I have played around with local models and had some success, but I don't want to deal with the hassle of setting up, updating, etc.
If all of this were easy, I would happily invest in new dedicated hardware to run it. This could be a full-blown Mac or it could be a souped up HomePod-type device, loaded with RAM. Better yet, make it a revitalized Matter-enabled Airport!
Yeah, I might dive in with llama.cpp at some point, but unless Apple censors the models I can choose I'd just as soon go with Apple's solution.
I also want something that integrates super easily with my iPhone and other devices, and hopefully Apple will make it possible to use my server as an intermediate fallback between my iPhone and their cloud models.
After I think about it, this makes sense that Claris would tackle such a project under the auspices of Apple. They're a subsidiary so they can monkey around with this stuff without having to adhere to the draconian everyday operating restrictions of "Apple proper". Good on them if they can pivot into something that hews to AI.
It's fun to think about Claris as the hobby "well, we'll see if it works out I guess, but no worries if not" brand. Perhaps "Claris Vision Pro" would have been less embarrassing for Apple. Hell, if they want to bring their car back, "Clarismobile" would be hilarious.
(1) All the headaches of maintaining FileMaker Server on Apple hardware, (2) multiplied with the additional requirement to cluster devices, (3) but without users accustomed to the silliness of misusing consumer devices running a consumer operating system for a service that must be reliable, and (4) an Apple subsidiary unable to bring about meaningful change.
I really do not understand the value of running large models locally. I am absolutely fond of local models but for specific tasks. The agents need to do a lot of work in this case: orchestration, big or small model split of planning and work, split into small tasks, merging, etc.
LLMs with <30B parameter models seem like the sweet spot. I would rather focus on multiple tiny LLMs that are tuned to specific tasks (SQL queries, finance/accounting, backend/frontend dev, etc.).
For larger models, a $10-40 ish Neuralwatt/Opencode Go or any similar subscription goes so far for the average user that it is ridiculous. We are still in the "youth" era of agents. There are many patterns that work well with smaller LLMs. Or with a big-small LLM combination.
My context:
I own the tiny 16GB M4 Mac Mini and I use LLMs (coding agents) to build software and a few of my products use LLMs directly. Full-time user of LLMs, do not read or write code manually. I run the tiny Qwen series (<10B params) sometimes but a couple $20 subscriptions (Claude, Neuralwatt) take care of my full-time agent usage.
Model/harness that runs locally, which you can connect from your phone to do non-dev queries (looking stuff up, recommendations, planning and etc.) can save ~$300/year on various subscriptions.
The addressable market is enthusiasts. The cheapest and least loyal consumer that exists. But I guess it is just a small side project for apple which is fine.
Idk. Once they say our AI app is facilitating AI needs of big number of DAUs, the stock price will likely go up even if there’s no revenue path. It’s a pretty tulip.
When I hear Apple and AI, I think of Siri. And then of my attempts to use it in my smart home where it would turn off all smart sockets instead of turning on the lights in one room. Or its response latency. Or how I was excited about Apple Intelligence and then removed it completely and am hesitant to upgrade to OS 27.
I understand that we’re talking about local inference with an Apple inference engine for Apple Silicon uniform memory. And I know it’s a stretch of an analogy. But somehow I’ve grown skeptical of Apple’s AI capabilities. Not to say Microsoft’s or Google’s weren’t disappointing.
For me Siri hasn’t improved a bit since release and it’s just the fact that she can’t understand my voice/words. Constantly mishears or asks me to repeat things. I have a pretty “regular” or “generic” US English accent and other voice agents have no problem so it’s definitely Siri. So I have never really been able to ascertain if it’s any good without an acceptable interface. I bounce immediately when I encounter the friction of having to yell, try to enunciate differently, think how to word my statements, etc. It’s like ordering food at a drive thru that has a broken mic and speaker and the employee is distracted not even trying to pay attention to my order. All to say, the spirit of your analogy seems appropriate to me.
I have the same Mac M4 16gb. Do you have the Qwen model assist the larger models for menial tasks/tool calls during coding it can handle? As in have them work in tandem to save on tokens/calls on the large models or are your local agents just meant for their own specific side tasks?
A mix of both. I have been experimenting building agents that focus on small/tiny LLMs. For example, I built a web-app prototyping harness which used Qwen 4/9B models but separate agents (tool calls, prompts and actual code to parse) for CSS management, TSX management, Router, state, etc. It used SolidJS only.
There was a task manager which was given context of the tech stack so it could split the incoming requests. Then branch out the tasks to appropriate agents. Combine the generated code - this was a file/module/import manager.
The other way is let tiny models (again I would use agents for different use cases) do the busy work - tool calling, finding evidence, etc. I usually setup agents to re-write tool call history and create fresh API requests without using multi-turn chat with tool call history.
I really appreciate the response, that gives me good ideas on where to start going about all this. It seems so silly to have API models doing so much file exploration and tool calling.
I saw this a while ago which shows small models and focuses on the ones that make the least (or no) wrong tool calls and it got me intrigued on the possibilities in this realm, especially for lower end devices. https://github.com/MikeVeerman/tool-calling-benchmark
Yeah. I for example use local inference for tagging up my documents that I self-host in a local Paperless instance, and I don't really feel like sending my medical invoices to clammy Sammy.
I can imagine a future where we look back at these times of relying on big opaque hosted services to run inference and find it quaint. Perhaps the same way nowadays we find it quaint that you couldn't always just compile software on a fanless portable computer.
How far away that is, or whether it will ever happen, I don't know. I agree though that the tooling is still so young it's hard to say what shape this type of computing will look like and therefore what kinds of resources will be in-demand.
Same context as yourself. I run 8 - 12Bish models locally on a 24GB M4 Macbook Air and also have a couple subscriptions (OpenCode Go, Mistral Pro) for access to DeepSeek etc. to power through yucky big bloated codebases at work.
subagents use mostly the same allocation of memory, so you can do orchestration and planning easily with many subagents just like you do with Claude Code with anthropic's cloud models
Pi has a subagents package to enable in its harness
An Apple M5 Ultra running an app like that? Because Apple hardware is a monolith, unlike the PC space, "Harbor" will be configured out of the box with the ideal settings for many well-known open-weight models.
... Looks like I picked the wrong week to buy parts for a linux-based inference box.
If you ask who would pay for Apple local models, I would. A Mac Studio high end is $30k one-time versus my current run rate of $10k/week for Claude Code Sonnet (if I paid API rates rather than Ultra plan).
If/when Anthropic changes to pure API pricing, that's when I'd need to switch to much more local.
I currently run 128GB with qwen3.8:27b-mlx. It hits the heat limit of the Mac when working on a 200k LOC codebase for security auditing and hardening.
$10k for Sonnet per week? How did you arrive at that number? It’s quite a huge inference volume. No consumer hardware will give you that level of latency at comparable cost.
$6k is not the only cost - electricity adds up over time. And hardware can fail, redundancy is priced into the token prices and subscription plans. Maintenance is mostly negligible, but some updates/upgrades might regress. There’s a lot to SRE to keep it running at that scale.
I run "npx ccusage@latest weekly" and also spot-check Claude Code sessions via "/usage".
> No consumer hardware will give you that level of latency at comparable cost.
You're right. My belief is the frontier AI companies are currently subsidizing real costs for the purpose of rapid growth customer sign ups, and that in a couple years or so local costs will become reasonable, at least for full-throttle work that's queueable.
I did the math once, to run Kimi K3 on a Mac, you'd need to daisy chain 5 Mac Studio M3 Ultras. So $125K. And the latency would be bad.
For good latency, you would need a multi GPU server around $3-600K.
I agree the hardware will catch up and local is the future. For the average person who is sharing there privacy data out willy nilly all the time without much regard, i.e. most people, it makes no sense right now to go local vs. API / subscription given the price differential.
Even if one believes the price of models will only go up from frontier AI companies AND believe the local AI frontier will be fast enough at a good enough profile, it still doesn't explain how you're going to beat the price of a place hosting open weight models at scale on shared hardware always maxed out 100% in a place with cheap electricity by buying a single platform up front for 3 years.
Data sovereignty always makes sense in this case because it's explicitly not about cost, but I've not seen a convincing "it's actually cheaper to do it yourself" analysis that came to that conclusion by answering all of these things up front instead of just hoping it'll somehow get better by then.
I can easily see everyone on my team having a Claude subscription, and then we can orchestrate and federate to spend the unused limits. I don’t see how this breaks toc, given we don’t resell. And 1000€ (or dollars) in Anthropic subscription buys billions if not trillions of tokens, if used right.
Why is it that you don’t include costs for the engineer catering to 512gigs system, potentially well saturated in usage? Anyone who understands inference at this level is at least 5k a month.
I would be quite happy with an external box with no processor, no ports etc. (potentially HDMI out), just the GPU cores and plenty of memory. Then I can upgrade AI without the rest of the machine.
447 comments
[ 3.0 ms ] story [ 84.1 ms ] threadDistillation works: we can get models leaner, have better performance with lesser and lesser parameters, more efficient, running locally. Apple gets this and we'll see the result with M7 Max/ultra chips and I reckon this will be one of the titular apps when M7 line of chips are introduced.
I guess it explains the haste Anthropic and OpenAI are in to IPO, cash out and leave the the retail investor holding the bag.
Right now 128GB MacBook is the top of the line and you could do Qwen 3.8 Flash or similar.
There is a rumor that the M7 chip can use 1.5TB of memory which would imply Apple is aiming for more memory on devices.[1]
Would that imply the entry-level MacBook will have 128GB and top of the line models could run larger reasoning models? Then the question is price and if any of this will translate to iPhone for on-phone models.
With how expensive memory is, IDK if they can pull this off but I hope they do
[1] https://www.tomshardware.com/tech-industry/semiconductors/ap...
If Apple, in 2030 has a compute product that a worker can BYO(X) to work that has 2TB RAM, with an encrypted personalized context that can float between agents, then they will be in a very interesting place.
holy shit
And with a device like this, you pay one single time to buy the product and then never have to think about paying for tokens ever again. For ordinary users that's silly, but for users or businesses chewing through lots of tokens, this means they can create their own turnkey replacement for "the cloud" where the only cost after purchasing the machine is the electricity to run it (which with these machines is very little). Mix that with the privacy benefits and suddenly it makes a lot of sense.
Like I said, seems HPC is lost art. All four sub replies missed that in my original comment.
I agree with you, and i would like to set up a server at work for local inference but we simply do not have the knowledge, it's much easier to get a couple of AI Max machines that are good enough for me and the boss which i convinced on the good points on running clankers locally and be done with it (though my plan is to use the AI MAX machines to figure out what the infrastructure and workflow will look like, and at some point ask for the investment)
How is this literally even possible in 2026? In a thread about LLMs, no less?
Then there's testing the approaches and see whatever sticks best.
All while you're actually doing your actual job
Not saying they're right.
For individuals though, it’s the same reason plenty of people pay for the 2tb or 4tb SSD for their laptops or desktops even though NAS has been a thing for 40 years. Maintaining a “server” of any sort is a completely separate endeavor for most. Just like a NAS is a no-brainer for many and non-starter for many more.
The problem with LLMs outside of server is that people typically have many devices. So you can run LLM on your laptop. You want to make a query from your phone. Start your laptop? You want to do something from your smart home? Start your laptop? What about your partner? Will they need to buy another top of the line laptop for themselves? Your kids?
Alternative to server is of course that everything will have over 100 GB of memory, eat battery like crazy, be dumber and slower than single server in your household.
4 TB ssd was like $300 until recently. Even if RAM prices were to go to previous values, I don't think AI capability will be any comparable. Besides you also need proper compute, unless you want your prefill to be snail slow.
Every scenario you described applies the same to a NAS. What if you need a file from your phone? Open your laptop? What if your partner needs a file? Open your laptop? The answer to all of them is either yes, or they have it on their laptop/phone too.
Apple (and other laptop makers) weren’t charging $300 for 4tb even before the shortage. They realized that not everyone needed that and they way overcharged for that but still less than today’s prices. People who needed it bought it even though it would have been a lot cheaper for them to build a NAS. A current-gen M5 is perfectly capable of doing decent tk/s given enough RAM. In 2 or 3 years it would be even more capable, it just needs RAM.
Plenty of people will use clou hosted models like they do with storage today. Plenty of people will build home servers for AI. Just like plenty of people have home NASs. But the number of people who just want a large SSD on their laptop make the previous group a rounding error and those people would rather have a beefy laptop or desktop with enough RAM in it. If anyone can solve the Phone<->Desktop communication seamlessly it’s Apple after all.
Unlike storage however, a lot of people are now aware of how much data collection tech and AI companies are doing and the idea that AI models are training on their personal photos and prompts is deeply unappealing to many. With just storage, what could Google or Apple do? Certainly they wouldn’t just print your photos and put them in an ad or on their website. But with AI models, it’s a different equation
No. Files are your property. Inference is a service.
There's also the fact that working with files over network is often ill-advised and borderline impossible for some workloads. Even when that NAS is connected over wire and not wireless. Meanwhile even if you're on so-so connection, LLMs work great.
But on the other side, equipping data centers was never Apples main business. Their approach to markets is to design products for the individuals. They also usually have very long time plans. Look at where the Apple Silicon today is and how long a path it took. What about the day, where at least a medium sized Mac becomes a great platform for running large models locally? Very easy to use, no setup, no privacy concerns. Also never worry about "tokens". You can then use AI as you use other local resources of the hardware you own today.
But the family shouldn’t be limited by that. We’re not going to see M6 Max hardware, because the M7 line is being fast-tracked and pulled forward. And I wouldn’t be at all surprised to see M7 capable of 1-2TB or more.
Spoiler alert: they're supply-constrained too. Their customers will pay to outbid Apple, just like how Apple outbid the industry on TSMC's 5nm node.
If Apple can't place DDR5 orders that outbid the HBM demand, then yeah, Apple Silicon is going to get fucked over by every vendor. They'll be laughing their way to the bank while Apple begs Californian courts to give them free RAM.
Probably sitting under a router, so that it can hand off bigger jobs to an Apple DC somewhere, probably with a certain number of credits based on your iCloud sub.
I don’t think Apple can be in a position where all of the Chinese device-making companies can get memory at a significantly cheaper price from CMXT, and Apple is stuck with buying memory from the three hyenas of memory at a significantly higher price.
That’s not going to help Apple sell their devices at a competitive price worldwide over time against the Chinese computer makers (Lenovo, Asus, and Motorola) and make no mistake about it: their share of the market worldwide will get bigger because of this change in the memory market; their focus will be on the long term, while the lawyers/MBA/finance types in the West will have their heads stuck in the short term.
It’s the (Chinese companies’) opportunity, and they’re not going to let it pass; I wouldn’t. In addition, Apple is in competition with Nvidia, Microsoft, Intel, AMD, Qualcomm, Google, Meta, and indirectly with Anthropic and OpenAI (whose CEO, Sam Altman, recently declared that computer ram is wasted on low-level regular computer users), along with a whole legion of other 3D model makers on the procurement of memory.
The combination of all these elements going into the future means that if Apple wants to stay ahead, I believe Apple has no choice but to take memory in house, they have the design and engineering capability in-house, and they have the money because they didn’t burn it on AI models and data centers.
This scenario is no different than replacing Intel or Qualcomm; you can see the writing/destination and the ultimate train wreck up ahead if Apple doesn’t roll up its sleeves and get busy, the market has changed and again Apple has to change and adapt to the new conditions.
Why are you talking about memory? Nobody cares about memory. You know what people care about? People care about ballrooms, and you know what, the US has the best!
People who know ballrooms say the US is hot right now, hotter than anywhere else in the world, because we're building the best ballroom in the world. Can you imagine that? China is nowhere on ballrooms.
I wouldn't be surprised if that was happening already -- they plan five years ahead, after all, and they've been slowly taking other systems in-house (e.g. baseband/modem) already.
Most people neither need nor want to run large LLMs locally.
The entry level MacBook Neo costs $599. Even if RAM prices revert to the pre-AI mean, the entry level models are not going to be that stacked with RAM.
Eventually we'll get to a doubling where entry level means 16GB of RAM, but most people won't be running anything more than simple models for light local work.
Have you ever tried running large LLMs on a MacBook Pro? It's hot, loud, and makes your battery level disappear at rates you didn't know were possible if you're not plugged in. It's a very niche hobby.
I'm old enough to remember the time when most people neither needed nor wanted a phone with a large screen and no buttons
First someone would need to invest and train a model on pr0n, and then they'd likely first want to recoup that money - perhaps by selling a subscription to pr0n addicts.
Whatever local LLM device comes out of this, it certainly won't be a laptop.
(I envision a black box like a router you plug into a socket and let sit in a corner of some room.)
You can get uncensored versions, but they are less usable (lobotomized).
i suspect that they will at least do offloading for casual usage as well. if they can ship 24GB in a homeview or appletv style product, it's game over.
why would a company who locks iPad Pro entry level literally out of ram that exists on its own chips.
I get their high end 10k-20k laptops will be able to reach these specs and gatekept behind also upgrading the cpu to the ultra but no way will it be an option in entry level.
Where I am, it's cheaper to purchase multiple claude subscriptions than it is for the power bill from running a local model. Factor in the hardware costs, and you're looking at a break over a decade in the future, for comparable cloud models. Yes, there's privacy, etc, but now you're down into fractions of a percent of users who care about the government reading the memory of their computer.
If you are an investor, you should remember they are a fashion product more than a tech company.
If you are a user, you should remember they are a fashion company more than a tech company.
It's not comparable to huge models of course, but given the power/size ratio here it shows that serious gains are on the table for models that can run fast on a high-end laptop.
A few years back we never thought open source llm would get as good and even the small and cheap model
Its good to have a race and such competition
At some point, the real cost of inference/intelligence is going to come due, and the existing open source models will be good enough for 95% of tasks users want (or training their own model(s) will be inexpensive for Apple).
Apple will be there, as usual, to provide secure, on-device, faster inference that the vast majority of their customers will be fine with. Let everyone incur the R&D costs.
Look at what it costs to serve Kimi K3.1 Pro on a B300. It suggests margins of 98-99% at API prices for OpenAI and Anthropic.
Should the shortages for memory and GPUs ever improve, this will only become much cheaper, won't it?
That said, R&D is another story.
Yes, I promise you, there are many, many people who are hesitant to build anything real on the assumption that either Anthropic or OpenAI will continue maintaining their services as they are today in perpetuity.
I agree that it pays to be wary of the additional abstraction layers they offer that purport to make your life easier, but for anything running standard completion calls or the like, I don’t see why this needs to be a concern.
I don't think so. LLM inference enjoys huge economies of scale. At no point in the foreseeable future will a local model be able to compete with a cloud model in terms of price. Local models will have the privacy angle, but very few people will be willing to pay top dollar for that. There will be some people, for sure, but the vast majority will just use cloud models.
Maybe when chip would infer at 10-20x the speed of a 5090 RTX, otherwise it's really..bad.
Beside the 35b, the others are too slow to do anything vs cloud models, unless you have a local B200+. And the Qwen 35b has the reasoning capacity of a 2yo doing maths.
I wasn't "impressed" with any models for coding until 1year ago, only then they started to be "usuable" with the fault that we know and still exists.
What new bike should I get? How should I rephrase this email? What to do about my persistent headaches? Recipes based on these ingredients?
It'll become easier and more affordable to distill/train/optimize smaller local models to meet your specific work needs. Multi-agent efforts will orchestrate the balance between using on-prem hardware for most tasks, and then defer to cloud models for more demanding/edge-case tasks. Especially if an organization can manage what data is shared with which model.
I also expect cloud providers to eventually start charging different rates for on-peak and off-peak usage, which would incentivize running what you can with the local compute you have otherwise sitting idle across dozens to hundreds of machines in your fleet.
I could be wrong and all your future hardware purchases will be glorified thin clients, with less local processing capabilities than today's laptops, servers, and towers.
In the mainframe's heyday, it was difficult to grasp a world in which compute became so dense and so cheap that it made sense for each individual to have all processing and memory on their desk, talk less of on their lap and in their hand.
Of course this time could be different. But, when we look at previous relentless progress in compute and efficiency, then also couple that with the breakthroughs AI itself unlocks, it's hard for me to imagine that centralized AI will meet both the scale and flexibility desired for all use cases.
Just as centralized compute never fully disappeared -- it took the form of client-server computing and the cloud, there's no question that centralized inference will continue to exist.
But it wouldn't surprise me if, ten years from now, 90% of global inference happens at the client.
While simultaneously, the 10% of inference that takes place in the datacenter is wildly lucrative and orders of magnitude greater than today.
Put simply: the inference pie may become so large that both client and datacenter win.
Beyond inference, even local HPC or large in-memory workloads. Steve Jobs dreamed that the NeXT Computer could be used to cure cancer in a dorm room (whatever tha tmeans), and I guess a bunch of Mac Studios chained together wouldn't be bad for a lot of computationally tough biomedical "basic science" applications.
Qwen flash-next is at least as good as Sonnet for my uses, and the speed is perfectly acceptable.
I really think the big labs are going to soon start feeling pricing pressure for everything smaller than Opus-tier models, and it will come for those next.
Only if you're comparing apples with oranges. There are no local LLMs at 128GB RAM which compete with Sonnet 5.5 intelligence. And the gap is a chasm when comparing against Opus 5.5, which is fast and cheap and plentiful on event he $20 subscription.
People saying this actually know how to run local models.
People claiming otherwise are Claude customers who know nothing beyond its point-and-click interface.
Consider that fact.
Does it actually? Sure, it is operating at huge scale and this design choice allows enormous models (presumably enormous MoE models). But is it achieving economies of scale?
Economies of scale are rooted more in the idea of a profitable thing becoming cheaper as the market grows, allowing cheaper manufacturing and further growing the market.
The big two or three LLMs are cheap because they are substantially subsidised by investor cash, and they are not profitable at all in any sustainable sense: the current pricing can only be sustained for as long as they can raise money, and the market will (and does!) shrink when it is asked to pay a less-subsidised price.
There appears to me to be no route to true profitability for this design strategy at all. Even the open weights router companies are downstream of largesse from state-affiliated companies.
No, it's exactly the opposite.
The big money is in small(ish) but finely tuned for specific tasks models.
Whoever invents an app store for LoRA adapters will win this AI race.
Apple and Microsoft and other companies will own that front-end experience with the customer, and it’s not going to matter to most customers whose backend gets used for what.
And Apple really wants to sell people who have no cred in the space their stuff because THEY have no cred in the space.
Its basically a club for stupid f**s
It's not as if they are short of money.
I'm not sure which chip it'll be, but I now for sure I'm not buying new devices until the AI boom has somewhat settled and we kinda understand what we'll be using it for. Until then, every manufacturer will just be guessing what device to build and will be outdated within a few months.
Could be the winner, in your opinion. They are extremely well positioned with hardware, devices, ecosystem, privacy, brand value & trust, and a significant active base. So far, they have not yet delivered on any useful AI application. There are so many opportunities in both MacOS and iOS, I guess they just need a new software VP.
So does Apple even need to create applications (that nowadays will get copied and opensourced in a day), or the hardware is the winning strategy in the end?
I think the more interesting question for Apple is the split between cloud and local. Economies of scale and aggregation vs privacy and ownership. I’d say so far it’s 2-1.
Who has? Google, Anthropic, OpenAI and Microsoft have spent hundreds of billions and produced nothing but chat apps that Apple could build in a few days, they’re all just interfaces to models that are available to Apple via an API. Hard to imagine Apple as anything but the winner of this era unless OpenAI or Anthropic pull something out of the bag.
You could spend it on free universal healthcare.
> they are being inactive because they don't have the skillsets
Rare to see an argument that Apple can't hire talent.
They need a significant team creating their own model, (the talent of which has already gone elsewhere) and they also need to be working towards the future of photonic computing for running those models with success long term, again which will require a huge hiring effort. Considering it's clear they are not putting in that hiring effort it's obvious what work they have been doing.
This is extremely revisionist. For ex with WiFi, ThinkPad + a Lucent WiFi card was standard issue for years before any Apple product had WiFi. Apple shipped integrated WiFi, which is wholly consistent with the notion of improving what the industry is doing.
Connectors? Yes, Apple has a history of improving connectors. Other computers have had connectors since forever. In this case, Firewire was Apple's semi-proprietary alternative to USB. USB, which Apple put into their computers years after others had shipped it.
> They need a significant team creating their own model
Your judgement. It's entirely reasonable at this point to license other models and look for a better/cheaper entry point. Apple tends to enter markets when it is possible to make profit. It is not clear yet that the business of creating and running advanced models is a profitable venture. By which I mean no company has yet demonstrated the ability to earn a positive ROI by developing and running an advanced model.
> working towards the future of photonic computing
Again, nothing in Apple's history would suggest that they would even aim to do this.
None of that stuff has worked right for several years now apart from maybe Gmail.
This isn’t true they’ve literally launched Apple Intelligence with multiple apps and surfaces three times now and it’s flopped each time so badly they regroup and pretend it wasn’t the real launch then launch it again at the next WWDC.
AI doesn’t have to be full-blown conversational AI.
In Photos, text recognition in Photos works well, object recognition and face recognition are somewhat decent.
In Mail, I find message summarization useful.
Small steps? Yes, but I think these are useful AI applications.
I do not use them, but in accessibility, they have AI applications to describe scenes which, I think, are useful to some.
Siri just got AI-fied and it's worse and slower than ChatGPT in 2022. Also it doesn't work offline anymore.
I used to "describe a shortcut" maybe 5 times this week and today I was hit with "you reached the limit".
These are useful applications of AI - or would be if they actually worked.
But this is all based on marketing not facts. Apple is as privacy invasive as the next guy. Obviously this shtick will work as long as marketing continues do to good job.
Anthropic and OpenAI want to exit because they know they have no moat... and personally, I believe that the sociopaths running the US at the moment want to provoke a great economic crash. The Evangelical ones because they want the armageddon [1], the rich fucks because they took the "you shall own nothing" part for real and want to be the new neo-feudal overlords. And the best way to achieve that is by tying everyone's retirement savings up into their bubbly assets and then crashing down the economy, leading them to buy up everyone's asset at fire-sale prices with the liquid cash they made during the exit.
[1] https://www.theguardian.com/world/2023/oct/30/us-evangelical...
Soon enough: "Claude, please build me a MacOS clone. Hell, copy the entire Apple ecosystem the way I want it."
Also a lot of people work from laptops and running inference on data centres for 10th of the price is a no-brainer. Local models are more of a niche and always will be.
MacBook Air + Claude max sub for 20 years would still cost less before a maxed-out Mac Studio makes sense (not taking into account running costs), even then it’s not a Frontier or as fast.
Not if they can't sell anything because of ridiculous DRAM prices.
I don't think it shows yet but I think the dominance of the App Store will end. I think a ton of iOS apps will be replaced by ChatGPT/Claude. I think there is a big opening for a company to redesign the phone from the ground up purely around an AI assistant instead of apps.
I think AI will severely weaken iOS, App Store, and the iPhone in the next few years. It was nearly impossible to compete against iOS and Android until recently because no one is going to redesign their apps for your new phone. Today, that's different. Many productivity apps are just being replaced by GenAI/ChatGPT/etc and services are making it easy for AI agents to access directly without an app.
Not to mention higher chip prices will eat into hardware margins.
Sure, but I don't think it's usually the apps that people pay much money for. And on top of that, app store revenue is (from a rough estimation) 6-7% of total revenue (although it's high-margin). Apple would be entirely fine.
And if you're right that the phone of the future will be reimagined as centred around being an AI assistant: innovative device design, user experience, hiding technology, integrating software/hardware/cloud are all things that Apple is historically pretty good at (with hiccups and bad periods for sure, but other companies certainly aren't better in general).
~15-20% of Apple's total profits at ~75% margin.
Investors are not going to like if that suddenly starts shrinking.
An LLM can probably whip up a music player app, but won't have the licenses to provide you the music.
An LLM can probably whip up someone akin to a free and single player Genshin Impact, but the whales want attention from other players.
An LLM can probably whip up some sort of SFTP from phone <-> Desktop for pseudo-cloud storage, but you still need to provide disk space
Why? Off the top of my mind (far from an expert and not thinking about it very hard), they can profit from AI in the following ways that have nothing at all to do with the app store:
For example if they capture 10% more of the global phone market because iPhones became that much better than alternatives thanks to AI integrations, that'd probably offset decline in app store revenue.Just throwing random ideas and numbers here, just making the point that shifting environment can weaken a revenue stream while strengthening (or enabling) others
To put it in perspective they reportedly collected about $675 million from OpenAI in fees last year, that's equivalent to about $1.5 billion revenue in hardware sales and TBD for AI services but certainly over $1 billion.
https://www.macrumors.com/2026/03/20/apple-made-nearly-900m-...
The other avenue that AI is hitting their services revenue is the Google Search deal, it's a revenue sharing agreement and AI is eroding that revenue on Google's side.
Not to mention, Apple is going to want to not disrupt itself so they're going to deliberately slow down which opens up the opportunity for someone else.
OpenAI, Anthropic, and others can spend billions on the infrastructure to train the models at a time when the technology is still immature and largely inefficient and then Apple can simply license the models as needed, giving the user the choice while potentially protecting their privacy. Once there's relative maturity in the industry, they can either continue to license the models, distill their own from them, or even train their own using others' infrastructure. To me, it's a smarter long term strategy even if it temporarily gives up their control of key technology.
Personally, it seems to be the smarter play and aligned with how Apple usually does things. They're usually not first to market but they're first to introduce a refined product that make people forget about everything that came before it.
Why this question? I don't think that the way that anyone expects Apple to profit off AI is from selling models
This is what we all want, this is what kicked off the open web back when the internet first started. This is what made everybody excited for user agents. Now we have user agents once more and we are getting excited about the new agentic web, but temper your expectations. This time, rent-seeking corporations are in it from the start, see the exact same promises that excited you about the open web, and plan to take full advantage right away. This is why everyone is trying to sell their own agent instead of giving access to your own.
Is this master stroke in the room with us now?
Now it can at maximum profit from it going public but not even that is sure now.
Literally nobody in the world wants more AI slop in their lives.
This is coming from my own experience as a software developer. I do not want to leave Claude Code. I'm hooking up everything into Claude Code and have it do what I want for me. I do not want to log into my cloud provider. I do not want to log into my Github. I don't want to log into any 3rd party services. I just want to stay in Claude Code.
I think developers tend to be ahead of the general public by 1-2 years in AI use. I imagine a future where most people simply want to stay within their ChatGPT/Claude/Muse/Grok/Gemini agent to do just about everything.
If my prediction is correct, I think Apple is going to be in trouble because it weakens the iPhone ecosystem severely and they are so far behind in AI everything that it's hard for me to see them ever catching up. Not to mention they do not have access to compute since they've not invested in compute.
I'm pushing the perennial 'developers (and engineers) think differently' trope here, but I think you're wildly optimistic.
The inability of the average person to write a detailed enough spec for implementation, compared to how trivial it is for an experienced developer, is mind boggling.
Imho, this is going to extend to agentic interactions with app-as-service-for-agent.
You don't magically increase customer experience by multiplying (garbage average user prompt) * (whatever agent they use) * (MCP, etc).
The first two variables are too unbounded.
Apple's use of Siri in Vision Pro is a good start, but there would need to be tighter integration w/ the UI. That's probably on the roadmap... assuming they continue to develop it.
Instagram, Tiktok, Youtube, you're right. So hopefully OpenAI can convince those companies to convert their apps over. Wouldn't be too hard given how easy it is to duplicate an app nowadays.
How much longer do you give until Claude Code replaces you entirely? Product or CEO dictate their wishes and Claude returns a freshly compiled binary and offers to deploy it to production with a simple "yes". A SWE salary seems like an lot of money for a middleman to pull levers on Claude Code.
That's why my investment strategy is centered around AI replacing me and making me unemployable.
I'll definitely be buying Anthropic.
Funny enough, I do have substantial amount of Apple stocks I accumulated pre-ChatGPT but I've been selling here and there as it reaches all time highs due it being an "AI hedge".
I think Wallstreet is getting it wrong and I think the narrative around Apple will change fast in the next few months.
Failing to evaluate AI beyond "slop" is down to a lack of imagination.
Essentially every single non-builtin app on my phone fundamentally requires the internet to work, and is backed by some sort of SaaS with a proprietary API. And even if/when the APIs are open, I doubt any normal person is going to decide that from now on they want to somehow use YouTube or Tiktok or Feeld or Spotify by talking to ChatGPT.
If a new mobile device form-factor proves to be usable, they'll come out with one eventually like they always do- like the iPhone duo being years and years behind?
When the bubble pops they could still snap up an AI lab on the cheap later.
They wanted to keep everyone on their proprietary platform instead of enabling their devices to be used on the wider internet / being a more open mobile platform.
As long as apple allows some AI integration, then I think they'll be fine- and in fact I trust their privacy and security model more than if I bought a Google device. (not to say I trust anyone's security model, but it's clear that Apple has a more vested interest in security and privacy than Google does)
It'll make the iPhone look like Blackberry.
And let's say that Apple wants to respond with their own AI-first phone. But wait, they don't have the model. They don't have the compute. Most of their employees only know the old way of doing things. It'll be like GM trying to compete against Tesla or BYD in EVs. Their whole workforce and skills are in ICE.
Sometimes even DoorDash gets it wrong, but at least I trust DoorDash to deterministically improve their systems so it doesn't happen frequently. With LLMs I would have to think that sometimes it just gets it wrong due to nondeterminism and insufficient reinforcement learning in the area of "reliable food ordering".
They’ll be fine
We've already seen several attempts to make devices purely around assistants like Humane Pin or Rabbit. In 99% of cases users need a reliable screen with clear buttons not a conversational barrier between them and action
I personally didn't like the voice-approach of Humane Pin. Rabbit R1 was more interesting but execution was poor, screen was not big enough, models weren't smart enough, apps and services weren't as open via MCP as they are now and in the future.
I want to see a company like OpenAI or a very well funded veteran company give it a try.
To get something like this: https://www.youtube.com/watch?v=b9hxwgH2tN0
I think folks who live in spec-driven worlds just don’t appreciate how poorly understood the capabilities and processes of computing and networks are, to the point where even if you gave people hallucination free LLMs you’d probably find people just frustrating themselves because they can’t figure out what they want, what it should do, and how it should be used.
Quite apart from the huge chunk of the app ecosystem that relies on common protocols, centralised services and so on.
* Based on our music discussions, make a playlist for me in Spotify. No songs older than 15 years.
* Look through my email, find all my receipts, show me a graph of my spend and categorize them.
* My business needs warehouse space. Find warehouse services within 15 miles of San Francisco. Ask them for their pricing and capacity. Here are pics of my items. Send these pics to them for quote.
* Find a hotel for me in Los Angeles on October 1st to Oct 5th. 4 starts. Max $200/night. Gym is a must. Don't care about pool. Close to airport. I have Platinum membership with Accor so try to find Accor hotels but not a must.
People do know what they want.
I thought the parent's point is that the App Store will loose dominance (because you won't need apps).
It is easy to express yes, but is this how people think about what they want to listen to?
Someone can already do this in a chat bot, see the list of things and then find it on Spotify or whatever their favorite music service is.
What the chat bot won't know is what they have listened to, which is probably a much higher predictor of what you'd like than the discussions you've had about music. You've probably had those discussions because you are already listening to that music!
That’s a surprising requirement. Is it a proxy for “less likely that I’ve already heard it”? For “the band probably still tours”? For “I can discuss it with friends without sounding like I’ve been living in a cave”?
I have seen this with yet "AI-unexperienced" users - for simple questions or textreformating, they are absolutely fine with the Chatbots. But thats all they can get out of it: Above that level their experience is very often frustrating due to too less understanding while having high expectancy towards the model in reverse (like "its AI so it should know...")
And this people end up then not using any of the AI tools beyond that because they do not "believe" / trust that this thingy can get work done.
Please.
I want to see the projected numbers.
What about doing something like REA on a whole app store collection, rewriting the app's reversed description set into code according the new devices' requirements? Perhaps we can't see the future as there are so many variations yet imagined.
For instance; I don't think LLMs will be replacing Spotify, Facebook, Insta, or Youtube for my parents. They absolutely do not want to purely interact with a blackbox. They want to scroll through songs/shows/etc. they want to tap on it, and have the app do the thing.
Even as a SWE, I have absolutely zero desire to interact with an LLM for everything, but I'm not the litmus. I have found use with these tools for taking on some nested set of tasks, but at no point do I want to give up the app entirely.
For people who like shopping (not me!), most of it is actually browsing and there is a joy in sifting through options. The OpenAI designers seem to understand this.
Apparently, GPT told him that there were 4 house listings in the area and so he asked my why the agency took down our house listing, even though he could have just went to <house_listing_website> and searched for it. Took me 3 minutes to open my phone and disprove him.
I think they are designing an addiction of knowledge. I think people want to feel like they possess all the information at a moment's notice (mixed with misinformation served as fact), without having to spend the effort to look for it
That's already happened. The curated feed you see is a machine-learning driven output of an AI model that Spotify calls a "Large Taste Model", supplimented by a host of models trained using Spotify's NEO framework. It might look pretty but it absolutely is the output of AI. I'd bet a fair sum that FB, Insta, and YouTube are also showing you the output of a model now rather than a traditional application.
(unrelated; but this sort of paradigm would be heinously more expensive)
Please, no. Just TELL me; that's why I'm using voice to talk to you in the first place. Summarize and tell me, don't show me a list of links from a web search.....
Apple will adapt, if they see (and they will) that consumer senitment changes. Right now, 90% of regular "consumers" seem to despise AI and anything made by it.
1. OpenAI/Anthropic must purposefully not sell their service to Apple and/or users using Apple hardware.
2. Local models must not get any better or slimmer than they are today, such that they can not be ran by Apple or locally on Apple hardware. This is arguably not even the case today, there are open-weight models that rival the frontier models.
None of these two things are going to happen. Apple does not need their own models.
If Nvidia wanted they could end it overnight with a 60GB 6090.
However will they survive if they cede ground in commoditized consumer spaces to explore market fit in luxury goods?!
Do I want to set up my scale to sync its data to my vibe coded app? Not really. Do I want to set up the dozens of visualizations and hundreds of different actions and thousands of different items for calorie tracking, where even a 5% error is extremely significant? No way. Am I going to convince any of my friends to download my vibe coded chat application? Absolutely not. Do I want to store my photos in my own vibe coded solution and trust it to run in 20 years? Not a chance.
While a segment of apps will surely become obsolete, I don’t think the majority of users time will be spent on their own vibe coded replacements. Even as a developer I have no interest in turning all my conveniences into maintenance burdens, with risk of catastrophic failure
The issue is there will be less demand for apps. I don’t need to pay for a dulingo subscription when Claude can write a 5 minute language quiz for me. I don’t need to pay for Turbo Tax when Claude can prepare my tax documents for me.
Sure, it won’t replace all apps, but people will spend less time in apps and more time interacting with llms which hurts App Store revenue (Apples golden goose)
From what I have seen you belong to the vast majority. Though not here on HN.
Commodities like todo apps and calculators rarely compete on features, they compete on product design. This includes look and feel, relation between elements, design choices on what exact common features are available.
All products are, at their essence, just a set of design decisions stitched and shipped together.
Now, making these design decisions is hard. It requires knowledge, consistency, and insights into how real people operate real products. Even as a sole user of the app you build you rarely know what you actually need.
LLMs are 'design-less' in their essence. Building your own todo app takes literal seconds and it all looks simple because LLMs will default most of these design decisions for you - until you try to use your new app daily and notice it somehow sucks. Building a good todo app or good calculator requires tweaking, publishing, polishing, and solving bugs, testing, rinse and repeat.
Another tricky side of this is that you might know logically what features you want, but you are not selecting one product over another logically. You need to try and see if that works for you, all the vibes and good parts and bad parts and design choices and colors, till you find something that "clicks" for you. Using pure Claude as a todo app only works as long as you want to have more Claude in your daily life.
In the meantime app store provides dozens of well-reviewed alternative solutions which look good and require zero maintenance.
And that's just a calculator or todo app. Non-commodity apps are even harder than that.
I don't see any future where programs follow the static nature of old paradigm programming. A nice intuitive UI? What about just telling the app what I need when I need it? The gold standard of consumer UI is...another person.
If I have a team of god tier light speed SWE's at my disposal, why the fuck am I fumbling around in an actions panel? The program can just be dynamically modified in near real time to be whatever I need it to be in that moment.
If you extrapolate LLMs out to their extreme, you get a shape-shifting liquid array of bits in memory that manifest anything at the behest of a completely natural language interface.
Absolutely not. Constraining the set of possible actions instead of having you verbalise what you want every time is often a feature, not just a technical limitation. A camera with a shutter button is better design than having to tell the camera “please take a photo”.
At the supermarket, I personally also prefer the self-checkout machines over going to a human cashier.
The UI is totally fluid at all times, because an "intelligence" is actively manifesting it at all times. Nothing needs to be hard coded.
You're also assuming that there are no negatives to saying what you want compared to just acting on things, but since the beginning of computing many people have proven this wrong time and again. I suggest you read about HCI as a research field and its history, it doesn't take much to realise how limited the average HNer's understanding of UI/UX design is, and how far current LLMs are from the skills required to be able to carefully design an app for productive human use.
Good design rarely calls for a “you can do everything” approach. Good design is often context specific and subtly guides users to better outcomes.
Doctor appointments would absolutely suck if I had to navigate decision trees and click buttons to communicate my symptoms. Car blinkers would absolutely suck if I had to type “blink left” into a chat window before every turn.
Maybe it’s just the ‘tism, but I hard disagree that I would ever say another person is a superior interface into… anything? It’s slow, they may misinterpret you, they have their own ideas and motivations, they’re (mostly) expensive because the only reason I need them is because there’s information / skill asymmetry. I don’t know that many or even most of those problems are resolved with ai, in some ways the problems are amplified.
I like to customize my Neovim setup or terminal and a bunch of scripts that I keep making them nice and maybe my blog and so on.
I don't want to become the PM of every app I need. But I do agree changes will happen to the current apps and how we buy and use them. Not sure exactly where it will settle. But still unknown unknowns and domain knowledge and vocabulary is not yet allowing anyone to build their own apps unless for simple cases.
Or maybe it is just me: I want someone else to think about my accounting app. I dont want to spend time to learn and understand and keep myself up to date or setup a series of LLM sessions that navigate and watch laws and propose changes.
Why do you need an accounting app when you have an accountant?
Programs of the future will in no way resemble what they are today. I don't even want to call it a "program" because in my head it's more like an intelligent digital liquid that can rapidly flow into any state. This flash-in-the-pan 50 year archaic era of software will look like what the oil lamp looks like today. A crude intermediate tech during a time of much larger transition.
it's nice to see my financial situation at a glance, without asking my accountant.
speech/text is one modality. we humans are also visual and tactile. and we need consistency and recurring patterns of interaction to function well.
Same thing for your accountant.
Imagine that both of them, while being good at their respective jobs, are also software engineers that can mint a custom throwaway app for you at a moment's notice.
I'm sure your secretary will make a note of your preference.
Sounds cumbersome, slow to use, and prone to failure.
However, you could expect LLMs to train on existing successful apps and hardened design patterns that compose well into good UIs. In that case, you might expect the last mile polish to be the differentiating factor. However, the more marginal this last mile becomes, the more other social, political, business, etc. forces will determine which app rises and which falls. But this has always been the case, certainly in software.
If you meant training on UI and features of established apps then it gets even more complex, as success of B2C apps is obscured by their pricing strategies and marketing, and business models for successful B2B apps like Jira/Confluence/Office often revolve around sales, partnerships and other non-product things like that.
Any business is a set of design decision someone made, with their app just being a small subset of these.
I dont think llms are going to replace banking apps anytime soon. But I no longer look for an app to pull data from my Peloton app into another exercise app. LLMs can build an app or a script that will do that and I wont have to pay some annual or monthly fee. It feels like these types of apps are going to die off.
I think it's better when you are at the origin, and the software branches off of your taste. For so long, we had to adapt and accept the ways of established software because it would take so long to make something of your own. Now it's both cheap and fast to iterate.
I recently used a coding agent to build a TV-like YouTube player for my mother, who is quite elderly. It's inspired by channelsurfer.tv, but runs on the Google TV Streamer in my living room. Unlike any set top box I've seen, it has on-screen diagrams that tell her which button to press on the remote to change the "channel." It doesn't have ads. The UI will never change unless I ask the AI to change it.
You can't say that about most apps in an app store. They will do a UI redesign and you're out of luck if you preferred the old version.
Do you really think Google or Apple or anyone else will build this app? Maybe some self-hosted media box out there would be similar, but I can't be bothered to search for one and I'm doubtful that any of them were designed for an elderly parent who isn't good with technology. It's still work in progress, but at least she can use a single remote to turn the TV on and watch TV without being bombarded with things she doesn't understand and doesn't care about.
There are downsides that make this app less compelling for other people, but I didn't build the app for them.
There's a recent benchmark indicating Hermes Agent and, I think, Qwen's harness are far and away the most effective.
Pi isn't bad but it doesn't seem as talented as Hermes. Nor does it currently array of self-learning/adaptive plugins seem as capable. That said, it's difficult to keep track of the SotA with pi; it grows like the node ecosystem in its first year.
But we are not there yet.
Whatever "good design" is (and it varies by user), it's very easy to copy. Commodity apps complete on marketing. Companies who charge $10/month for an app, are probably spending $8/month in marketing/advertising.
At the end of the day, it seems we'll be left with a decision of using something made by a person with real opinions or using something generated by AI with opinions that are a distillation of past documentation fed into it.
We're in this weird spot now where we are transitioning toward all ui feeling the same, and only the people really tuned in even notice this. Once the general public starts noticing, I believe there's going to be a search for fresh, organic, opinionated software. Not that the software won't be AI-generated, but it will have a human truly sculpting it, like art.
So the question becomes, if people still appreciate human-made art in a world where a better image can be generated in a small fraction of time via a prompt, will they feel the same way about the software they use?
This always puzzled me. Here's how that would go:
1. A new company, NeoPhone, launches that is redesigning a new "AI forward" phone.
2. The customers of that phone start noticing that they do many repetitive things with their phone, that they want the ability to design some repeat experiences with their chatbots. Call these repeat experiences "applets." Neophone ships some Applets on their phone that they created, a default set.
3. Customer A looks at customer B's phone, and notices their different applets that they designed. They work slightly better than customer A's applets, because customer B is slightly better at prompting.
4. NeoPhone notices this too. They create a new thing on their phone, the Applet store, where customers can share their applets and rate them and it surfaces the best ones to all customers.
5. Customers start complaining that they wish they had more creative control over Applets. Maybe the API is a bit more restrictive than they need, or the Applet needs to do special things with the phone's hardware, outside the context of the AI agent. Or they find that Claude Opus Fable 6.2 is better at making Applets even than NeoPhone's available chatbots.
6. NeoPhone allows 3rd party developers to develop on its phone, launching an "App Store" and a "Chat API" that is totally independent of the AI agent stuff.
???
Of course, Apple eats their lunch at step 2 or 3.
Would you trust access to your bank to a vibecoded phone of unknown origin? Backs will likely limit access as best they can for customer security. We already have banks who block GrapheneOS.
Look at the hardware wallet company that got sold and a spy chip to steal the contents was added.
I do believe AI is going to destroy companies like Adobe however.
This is the new reality that we must prepare to accept. You and I may not trust the companies, but the circle is closing in, and people around you are already on the "hype train".
Huh? I just took a look at my phone -- I have about 100 apps on there, though probably only use 20 of them with any regularity. I don't see more than a small handful that could be replace with Claude (if it had access to the APIs that the app does),
And even then, a chat interface isn't the optimal way to interact with many services. Random example: A simple metronome / tuner. Could I ask Claude to play a tick at a certain frequency for X amount of time, or a E-flat? Sure. Would it be easier than using the app? No. Could Claude design a metronome app for me? Yes, but I don't want to have to deal with that process -- I just want a metronome app that works and I don't have to spend time designing or fiddling with, and that works offline, and all of that is well worth the $5 I might spend to buy it.
Apple is absurdly late to the game. The MLX community looks anemic compared to Linux.
So far, I prefer tailnet to my home Hermes agent inference system and giving it ssh into my Mac.
It's easier having a single powerful inference system.
A Claude Pro account and Openeouter credits are my backups.
You’re so DELUSIONAL. It’s hilarious.
Show us your portfolio of work as an innovator - just to ensure you have substance to support thes ridiculous claims.
Top paid apps outside of F2P gambling apps(all the "games" remaining on phones are gambling apps because they've repeatedly lured in and punished all traditional games, definitively burning bridges), making hundreds of dollars per paying users, had been like $3.50 ad blockers. iOS app development had been a lot less approachable than Android apps due to Mac host requirement, and with LLMs generating Android apps and HTMLs with ease, the fact that iOS can't execute .apk will make it successively less appealing.
Personally I don't think things like agent-first OS that don't rely on third party apps will come soon - that will have to be built from ground up as proper OS kernels, and I think it will run on desktops first.
The third OS always existed, the problem with those is that telcos won't carry it and people won't use such a phone with the nation's pick of social media - Fb Messenger for US, WhatsApp for EU, Telegram for Eastern Europe or whatever, MAX for Russia(presumably?), WeChat for China, LINE for Japan, KakaoTalk for Korea, ...
... thinking out loud, the third platform has to be the one endorsed by a social media. Like the old HTC ChaCha, or like, idk, "Startile" slab device running TeslaOS with Starlink DTC SIM built in. I wouldn't touch such a thing, but something spiritually like that. Or like a "Rakuten Star" with AST SpaceMobile eSIM baked in. Things like that.
Example - search iPhone Settings.
The Apple platform provides certain benefits to me. Take a look at Matthew Cassinelli' work (great speaker), if you have time (and a Mac I guess):
https://matthewcassinelli.com/sirishortcuts/library/#element...
His work is a library of Apple Shortcuts, AI, etc., that does stuff for consumers, I suppose pro-users. He was on a recent MacBreakWeekly podcast - https://twit.tv/shows/macbreak-weekly/episodes/1050?autostar...
The bits of AI mixed the platform affordance to simplify daily paper cuts by clicking on an icon is worlds apart from a Google SWE spawning 48 "agents" to parse lidar data to discover pyramids in Mexico. That's also cool.
Why, though? They had a sufficient mountain of cash to outbid everyone else for TSMC capacity, and no realistic hope of catching up to TSMC's process on their own fab
Also, I'm not exactly sure we should be claiming what is and what isn't realistic for a trillion-dollar company. People could (and probably did) argue that they had no hope of catching up to Intel/AMD, and yet here we are, with Apple arguably having surpassed both, at least from an energy efficiency point of view.
Realistically speaking, "catch up to TSMC" is something that every single EUVL fab has failed to do despite tens of billions of dollars in investment. Spending money is the easy part, developing high-yield processes over a timetable of decades is the hard part. I think Nvidia and Apple both agree that paying TSMC to build American fabs is a smarter use of their money if they just want excess fab capacity.
But the hardware. The hardware is fantastic for local LLM's, generative ai, etc. And they seem to maybe finally be waking up to that, in true slow sleepy Apple fashion.
If they'd been a bit quicker about it we could have a higher tier iCloud plan that included a chatgpt plus style setup with the "its private" marketing tag that cant be provided by anyone else. Instead we get Siri being slightly less useless than before.
Local models literally cost more in electricity than using far more powerful cloud models cost in inference.
When your options are "dramatically drain your iPhone battery for substandard AI, or have instantaneous cloud AI with no battery drain" what do you think the average user will choose?
And that's before we discuss how ridiculous the pricing is for home hardware that has a chance of running a halfway decent model at a halfway decent speed. Residential electricity + overpriced underutilized hardware means you're going to pay 100X or more per token.
I think the future prospects for local AI are looking worse than ever. It will be far more expensive for a far worse experience for the foreseeable future. It will fill a privacy/control niche, but beyond that, it's not a serious option
NVidia aggressively segments the market on VRAM. Apple’s shared memory model is a competitive advantage. Well, it will be when they hand the raw PFLOPS, which is coming.
They desperately need to fix that.
It SHOULD be totally possible. But it fights you.
It’s too bad I cannot trust it to put it in my rack.
https://en.wikipedia.org/wiki/FileMaker
There's a name I haven't heard in a long time! Claris (always an Apple subsidiary, since being founded in 1987) renamed themselves to FileMaker in 1998, but apparently they rebranded back to Claris in 2019.
Looks like they've mostly just released annual upgrades to FileMaker for the past decade, but they did add MCP to it last year: https://www.claris.com/blog/2025/why-mcp-matters-for-claris-...
Some of that software has been in use since the 90s and still works. Adding AI to the equation will empower a lot of people.
However, FileMaker is not really targeted at software engineers. In fact using it as a developer will likely drive you up a wall. It's things that feel totally fine to non-devs but completely insane if you're used to something else. E.g. I remember having to click a thousand GUI buttons just to edit a few lines in a script once (regular users just "record" them through the UI). It did all work fine though, very reliable.
They added some more developer-specific features later, but it took a long time. E.g. editing scripts as text(-ish) only came in 2015 (FileMaker 14) and an actual programming language in 2020 (JS, FileMaker 19).
A few years ago Filemaker, Inc took on the Claris name.
Users will praise it still. What's the phrase, Apple distortion field?
This seems like the main way it would differentiate itself over existing solutions like LMStudio etc.
https://exolabs.net
Although this might be the shortest timeline as-yet that Apple has "Sherlock'd" 3rd party apps
https://en.wikipedia.org/wiki/Sherlock_(software)
I don’t find it reasonable at all. Subsidized Subscriptions or api for cheaper models, is the best value.
But the future of AI is so uncertain, regarding model sizes. I don’t think making huge hardware investments is reasonable for individuals.
If you buy your own hardware, then after those four years you still have that hardware, and it likely still holds significant value. Plus you know the local model you're running today won't change or suddenly degrade in four months, let alone four years. Whereas after four years of paying subscriptions and API costs, you have literally nothing.
GPUs are holding value longer, for obvious reasons. But even a 5 year old PC or server will be out of warranty, full of NLA components, and will get pennies on the dollar of its original cost on the salvage market. Most businesses will have fully depreciated them and expect maybe to sell them for enough to cover the costs of disposal.
There are ways to salvage old machines even when there is 0 support left/available.
That statement made sense ten years ago, easily. Less now, though.
There is no ZDR at that price.
I would not let a non-local AI have access to my financial or health data.
Compute & storage were cheap, so renting them was expensive for many applications.
Inference is expensive. Just getting started is 4-5 figures. Why not rent it for pennies.
Claris?? Ok, this sounds like it's just someone's personal project. Something like this that was actually going to ship would be in a different org.
https://www.pacifict.com/Story/
If true, it is an astounding story. That said, I'm happy to see that Apple ships a graphing calculator with macOS 27 (albeit, a different one). I may not use it much these days, but I have fond memories of it when I was in university.
If all of this were easy, I would happily invest in new dedicated hardware to run it. This could be a full-blown Mac or it could be a souped up HomePod-type device, loaded with RAM. Better yet, make it a revitalized Matter-enabled Airport!
They are going to sell sure 10,000$ and you won’t need anything else for the rest of your life (if you are over 90).
Btw, llama.cpp with llama-swap ain’t that hard to configure if you have access to a potential LLM and have 5h of your life available to that task.
I also want something that integrates super easily with my iPhone and other devices, and hopefully Apple will make it possible to use my server as an intermediate fallback between my iPhone and their cloud models.
LLMs with <30B parameter models seem like the sweet spot. I would rather focus on multiple tiny LLMs that are tuned to specific tasks (SQL queries, finance/accounting, backend/frontend dev, etc.).
For larger models, a $10-40 ish Neuralwatt/Opencode Go or any similar subscription goes so far for the average user that it is ridiculous. We are still in the "youth" era of agents. There are many patterns that work well with smaller LLMs. Or with a big-small LLM combination.
My context:
I own the tiny 16GB M4 Mac Mini and I use LLMs (coding agents) to build software and a few of my products use LLMs directly. Full-time user of LLMs, do not read or write code manually. I run the tiny Qwen series (<10B params) sometimes but a couple $20 subscriptions (Claude, Neuralwatt) take care of my full-time agent usage.
I understand that we’re talking about local inference with an Apple inference engine for Apple Silicon uniform memory. And I know it’s a stretch of an analogy. But somehow I’ve grown skeptical of Apple’s AI capabilities. Not to say Microsoft’s or Google’s weren’t disappointing.
There was a task manager which was given context of the tech stack so it could split the incoming requests. Then branch out the tasks to appropriate agents. Combine the generated code - this was a file/module/import manager.
The other way is let tiny models (again I would use agents for different use cases) do the busy work - tool calling, finding evidence, etc. I usually setup agents to re-write tool call history and create fresh API requests without using multi-turn chat with tool call history.
I saw this a while ago which shows small models and focuses on the ones that make the least (or no) wrong tool calls and it got me intrigued on the possibilities in this realm, especially for lower end devices. https://github.com/MikeVeerman/tool-calling-benchmark
How far away that is, or whether it will ever happen, I don't know. I agree though that the tooling is still so young it's hard to say what shape this type of computing will look like and therefore what kinds of resources will be in-demand.
Same context as yourself. I run 8 - 12Bish models locally on a 24GB M4 Macbook Air and also have a couple subscriptions (OpenCode Go, Mistral Pro) for access to DeepSeek etc. to power through yucky big bloated codebases at work.
Pi has a subagents package to enable in its harness
... Looks like I picked the wrong week to buy parts for a linux-based inference box.
If/when Anthropic changes to pure API pricing, that's when I'd need to switch to much more local.
I currently run 128GB with qwen3.8:27b-mlx. It hits the heat limit of the Mac when working on a 200k LOC codebase for security auditing and hardening.
$6k is not the only cost - electricity adds up over time. And hardware can fail, redundancy is priced into the token prices and subscription plans. Maintenance is mostly negligible, but some updates/upgrades might regress. There’s a lot to SRE to keep it running at that scale.
I run "npx ccusage@latest weekly" and also spot-check Claude Code sessions via "/usage".
> No consumer hardware will give you that level of latency at comparable cost.
You're right. My belief is the frontier AI companies are currently subsidizing real costs for the purpose of rapid growth customer sign ups, and that in a couple years or so local costs will become reasonable, at least for full-throttle work that's queueable.
For good latency, you would need a multi GPU server around $3-600K.
I agree the hardware will catch up and local is the future. For the average person who is sharing there privacy data out willy nilly all the time without much regard, i.e. most people, it makes no sense right now to go local vs. API / subscription given the price differential.
Data sovereignty always makes sense in this case because it's explicitly not about cost, but I've not seen a convincing "it's actually cheaper to do it yourself" analysis that came to that conclusion by answering all of these things up front instead of just hoping it'll somehow get better by then.
Of course Apple will not offer this.