It might as well be the visualization of the two strategies:
- Everyone else: "We mainly build huge AI compute clusters to process large amount of data and create value, at high cost for ramp-up and operation."
- Apple: "We mainly build small closed-down AI compute-chips we can control, sell them for-profit to individual consumers and then orchestrate data-processing on those chips, with setup and operational cost all paid by the consumer."
I can't think of any company which has comparable know-how and, most of all, a comparable sell-out scale to even consider Apple's strategy.
No matter what they do, they will sell hundreds of millions compute devices for the foreseeable future. They use this to build out AI infrastructure they control, pre-paid by the future consumers.
> We mainly build small closed-down AI compute-chips we can control, sell them for-profit to individual consumers and then orchestrate data-processing on those chips, with setup and operational cost all paid by the consumer
I wish they did but they don't. They have been for decade so stingy on RAM for iPhone and iPad. There are at current point that only small percent of their userbase have iPhone or iPad with 8GB RAM that somehow can run any AI models even open source and be of any use. Not mentioning they don't compare to big Models.
They don't even provide option to sell iPhone with bigger RAM. iPad can have max 16GB RAM. Those mainstream macbook air also can have max 32 GB RAM.
And for the current price of cheap online AI where e.g. perplexity provides so many promo for PRO version for like less $10 per year and all ai providers give good free models with enough rate limit for many users I don't see apple hardware like particularly bought because of AI compute-chips - at least not non-pro users.
If the loose AI though and because of that won't have good AI integrations they will loose also eventually in hardware. e.g. Polish language in Siri still not supported so my mum cannot use it. OSS Whisper v3 turbo was available ages ago but apple still support only few languages. 3rd party keyboard cannot integrate so well with audio input and all sux in this case because platform limitation.
Some lot of good that's done them. The Neural Engine is dark silicon on most devices I've seen, and now we're getting another product segment with M5's matmul GPUs.
To me, it feels like Apple should have supported CUDA from the start. Sell the ARM-hungry datacenter some rackmount Macs with properly fast GPUs, and Apple can eventually bring the successful inference technology to cheaper devices. Apple's current all-or-nothing strategy has produced nothing but redundant hardware accelerators, while Nvidia's vertical integration only gets stronger.
I have a little rust script that uses the built in vision toolkit to do ocr of pdfs, it spins up the ANE to a full 1W compared to 0 as measured by the power profiler. So it is used!
IMO, It’s a very apple strategy, stuff just works and is slowly more accelerated/lower power.
Maybe. But Apple tried the server business and found that they can't compete there.
Not because of Engineering deficiencies, but because datacenters buy based on facts, not fluff.
Now their ARM silicon is top-notch, no doubt about that. But will they earn a higher margin if they put it in a datacenter instead of a consumer device which is then used to consume Apple Services? I don't think so.
> But will they earn a higher margin if they put it in a datacenter
Nvidia is a five trillion dollar business right now. The total sum of Apple's profits from services, hardware and servicing/repair costs all fail to crest Nvidia's total addressable market. We've been past the point of theorizing for almost two years now.
Apple has the means to break into that market, too. They don't need the silicon (iPhone/iPad are way overpowered, Vision Pro and Mac are low-volume), they have thousands of engineers with UNIX experience, and hundreds of billions of dollars in liquid cash waiting to be spent. If the China divestment and monopoly case happen, Apple needs a game plan that guarantees them protection from US politicians and secures an easy cash flow.
From the consumer perspective, it seems simple; stop shipping the latest silicon in the iPhone. Nobody uses it. They're not playing AAA-games or inferencing the latest AI models, and the efficiency gains haven't been noticable for a decade. You don't need TSMC 2nm to browse the App Store, or watch AppleTV. The only opportunity cost comes from selling consumers hardware they can't appreciate.
> From the consumer perspective, it seems simple; stop shipping the latest silicon in the iPhone. Nobody uses it.
From a vendor-perspective, ~200mn iPhones are sold each year, the end-user will pay for it. The scale of this is financing the entire development and supply-chain for the silicon itself, and it contributes not only to hardware but also service revenue of the entire company.
nVidia owns 94% of the GPU market and shipped 11.6mn GPU's in Q2/2025, let's say they ship 60mn GPUs in 2025 total.
--> Why should I stop shipping the latest silicon in the iPhone?
Even without stopping production, why should I enter and compete in a market that is currently dominated by a single player, has a total size of ~60mn units/year, with each product deprecating almost instantly as soon as a more efficient product is announced?
Apple's silicon is not magically more efficient than everything else, their products are efficient because they are vertically integrated.
I doubt that Apple Silicon is competitive to nVidia in a datacenter setting
The existential hope that all the other players have is that AI will drive adoption of a form factor that replaces the phone. Because if in 5 years the dominant device is still the phone, Apple wins.
Consumer hardware chips will be plenty powerful to run “good enough” models.
If I’m an application dev, do I want to develop something on top of OpenAI, or Apple’s on device model that I can use as much as a I want for free? On device is the future
In 5 years, the dominant form-factor will still be a phone. This is not the risk.
The existential FEAR of the smartphone ecosystem players (Apple, Google) is, that another ecosystem (!) may come along, one that is tighter integrated into the daily lives, is more predictive of the users' needs, requires less interaction and is not under THEIR control.
Because this is not about devices, it's about owning the total userbase of that OS-ecosystem.
Replacing the Smartphone has been attempted numerous times in the past decade, but no device was able to replace it as a consumption device. Now technology has reached a level of maturity that Smart Glasses may have a shot at this. AND they come along with their own ecosystem as well.
Whatever happens, they won't replace all phones within 5 years. But it's possible that such a device would become a companion to an iOS/Android phone and within 5 years gradually eases off users of their phones into that other ecosystem.
And that's scary for Apple and Google.
Because this is not a device-war, this is an ecosystem-war.
> How late do you think Apple can come to that party and still wind up winning in the end?
In my view Apple is positioning themselves (once more) to win without the need of competing on fair grounds. They are late to this party, but their biggest asset is the control over the data and spending of their users.
The users WANT to use those services, and Apple is not ready to offer anything. But as long as they can be the "broker" between the user and such services (and most of all the deciding party!), they can sell the consumption of their entire userbase for revenue-share to the service-providers.
Their biggest risk (beside of stock-market impacts) is, that Apple users start to engage directly with such services without Apple being an intermediary party (using a browser or another device).
So their highest priority will be to keep the user entertained so they can continue profiting from their consumption until they themselves have arrived at the party.
Once they have arrived, they will start diverting profitable AI-tasks from 3rd parties back to their own services, leaving unprofitable ones to the then-integrated 3rd party providers
...Nvidia? Did you just step out of a cryogenic chamber from 2008?
The datacenter business is booming right now, cutting-edge and efficient hardware is needed more than ever. Nvidia and Apple are the only two companies in the world with the design chops and TSMC inroads to address that market. Nvidia's fully committed and making money hand over fist; Apple is putting 2nm silicon in the iPad Pro and asking fucking consumers to pay $1,500 for it. Do you not see the issue with this business model?
People will say Apple can't crack the datacenter market, I say bullshit. Apple drafted OpenCL. Every dollar Nvidia makes is money Apple pissed away on trinkets like smartwatches and TikTok tablets.
Yes, as I said in another thread a few days ago: Apple's strength is in making personal computing endpoint devices for consumers. That's what's in their DNA. They have not done well at anything else.
While that’s definitely true, I think it’s maybe more fair to say that their actual strength has always been to take a personal computing technology that’s just about “ready-for-prime-time” and make it as accessible and fashionable as possible. Almost all of their failed products have been errors in judging how close a tech is to being ready for mass adoption.
"Monopoly abuse"? Apple Pay was a huge leap forward in preventing credit card info theft, to the point that I only buy gas at stations that use Apple Pay to avoid having my bank account emptied by someone running a pump skimmer.
Yeah and part of that specifically came by sacrificing a personal computing endpoint product they used to sell, networked storage, at the sacrificial alter.
The funny thing is that Time Machine still works, and works better than any local backup solution for Windows that I'm aware of (let alone what comes with Windows itself).
Not to mention, they are generous enough to allow it to work with a non-apple NAS setup. I feel like that would be a different story if they were still in the NAS business.
I'm not sure how Apple is enabling anything interesting around AI right now.
That's what this bland article is not even touching on. Yes, having missed the boat is great if the boat ends up sinking. That doesn't make missing boats a great strategy.
Building huge models and huge data centers is not the only thing they could have done.
They had some interesting early ideas on letting AI tap app functionality client-side. But that has gone nowhere, and now everything of relevance is happening on servers.
Apple's devices are not even remotely the best dumb terminals to tap into that. Even that crown goes to Android.
Although I am an Android user, I am not enough of a narcissist to need to remove the crowds from my tourism photos. So, not all Android users have photos without crowds.
> They use this to build out AI infrastructure they control, pre-paid by the future consumers.
I'm not following. What infrastructure? Pre-paid how?
Apple pays for materials and chips before it sells the finished product to consumers. Nothing is pre-paid.
And what infrastructure? The inference chips on iPhones aren't part of any Apple AI infrastructure. Apple's not using them as distributed computing for LLM training or anything, or for relaying web queries to a complete stranger's device -- nor would they.
> Apple pays for materials and chips before it sells the finished product to consumers. Nothing is pre-paid.
The AI-capabilities of the devices will be pre-paid, as they will come with the product without delivering any significant value yet.
The end-user will bear the cost for that before he is getting anything meaningful in return, because Apple's production volume is at such a scale that they can offset those investments without risking to lose any meaningful sales volume.
Other players can't do that because they don't sell 200mn units per year. If they would add on-device inference chips, they would have to significantly increase the device-price, risking to not sell any product
That's not "prepaid", that's just normal building of hardware. And Apple's AI hardware is pretty general-purpose -- it's already widely used for things like dictation, image recognition, all sorts of stuff. Sure they've upgraded their "neural engine" into a "neural accelerator" to optimize it more for LLM's, but that's still incremental.
What Apple is also doing is investing server-side, just like everybody else, precisely because phones can't handle the serious stuff:
> Magic Cue - Magic Cue proactively surfaces relevant info and suggests actions, similar to how Apple's personalized Siri features were supposed to work. It can display flight information when you call an airline, or cue up a photo if a friend asks for an image.
Likewise Daily Hub didn't work but was shipped anyway.
> In our testing, Daily Hub rarely showed anything beyond the weather, suggested videos, and AI search prompts. When it did integrate calendar data, it seemed unable to differentiate between the user’s own calendar and data from shared calendars. This largely useless report was pushed to the At a Glance widget multiple times per day, making it more of a nuisance than helpful.
They roll out hardware to consumers they can use for AI once their service is ready, with users paying for that rollout until then.
Meanwhile they have started to deploy a marketplace ecosystem for AI tasks on iOS, where Apple has the first right-to-refuse, allowing the user to select a (revenue-share-vetted) 3rd party provider to complete the task.
So until Apple is ready, the user can select OpenAI (or soon other providers) to fulfill an AI-task, and Apple will collect metrics on the demand of each type of task.
This will help them prioritize for development of own models, to finally make use of their own marketplace rules to direct the business away from third parties to themselves.
My guess is that they will offer a mixed on-device/cloud AI-service that will use the end-users hardware where possible, offloading compute from their clouds to the end-users hardware and energy-bill, with a "cheap" subscription price undercutting others on that AI-marketplace.
You are just making things up in this grand AI strategy you have imagined for Apple. I cannot "fulfill an AI-task" with my phone because the overpaid idiots building it in Cupertino have years ago bought into the trainwreck that is Siri. So now I cannot "select my favorite AI provider" from the "marketplace ecosystem for AI tasks" to "fulfill an AI-task" nor will a meddling middle manager in the Loop collect metrics on the demand for "my AI tasks".
And now they are converting Siri into an orchestrator to "broker" between the user and the AI-providers for a revenue-share, because they are not ready to compete in that space themselves...
It isn’t clear to me that Apple will ever pursue their own chatbot like Gemini, ChatGPT, etc. There’s lots of potential for on device AI functions without it ever being a general purpose agent that tries to do everything. AI and LLMs are not synonymous.
Assuming that Apple take 30% rev-share from other AI-service providers on their AI-marketplace, once they are ready they can easily offer a lower pricing than anyone else and still retain a higher profit-margin.
But for this to make economic sense, the "AI-bubble" may need to burst first, forcing the competitors to actually provide their services for-profit.
Until then it might be more profitable to just forward AI-tasks to OpenAI and others and let them burn more money.
> once they are ready they can easily offer a lower pricing than anyone else
Do you have any evidence whatsoever that could back-up this claim? It feels like you're just saying this because you want it to be true, not because you have any concrete proof that Apple can sell competitive inference.
> Do you have any evidence whatsoever that could back-up this claim? It feels like you're just saying this because you want it to be true, not because you have any concrete proof that Apple can sell competitive inference.
Sorry, I didn't mean to state that Apple A/M-series will be competitive on inference performance compared to other solutions. There is no sufficient data for this at the moment. But this is not the competition I expect to happen.
I expect them to stiffle competition and setting themselves up as the primary player in the Apple ecosystem for AI services, simply because they are making "Apple Intelligence" an ecosystem orchestration layer (and thus themselves the gatekeeper).
1. They made a deal with OpenAI to close Apple's competitive gap on consumer AI, allowing users to upgrade to paid ChatGPT subscriptions from within the iOS menu. OpenAI has to pay at least (!) the usual revenue share for this, but considering that Apple integrated them directly into iOS I'm sure OpenAI has to pay MORE than that. (also supported by the fact that OpenAI doesn't allow users to upgrade to the 200USD PRO tier using this path, but only the 20USD Plus tier) [1]
2. Apple's integration is set up to collect data from this AI digital market they created: Their legal text for the initial release with OpenAI already states that all requests sent to ChatGPT are first evaluated by "Apple Intelligence & Siri" and "your request is analyzed to determine whether ChatGPT might have useful results" [2]. This architecture requires(!) them to not only collect and analyze data about the type of requests, but also gives them first-right-to-refuse for all tasks.
3. Developers are "encouraged" to integrate Apple Intelligence right into their apps [3]. This will have AI-tasks first evaluated by Apple
4. Apple has confirmed that they are interested to enable other AI-providers using the same path [4]
--> Apple will be the gatekeeper to decide whether they can fulfill a task by themselves or offer the user to hand it off to a 3rd party service provider.
--> Apple will be in control of the "Neural Engine" on the device, and I expect them to use it to run inference models they created based on statistics of step#2 above
--> I expect that AI orchestration, including training those models and distributing/maintaining them on the devices will be a significant part of Apple's AI strategy. This could cover alot of text and image processing and already significantly reduce their datacenter cost for cloud-based AI-services. For the remaining, more compute-intensive AI-services they will be able to closely monitor (via above step#2) when it will be most economic to in-source a service instead of "just" getting revenue-share for it (via above step#1).
> I can't think of any company which has comparable know-how and, most of all, a comparable sell-out scale to even consider Apple's strategy.
I'm not sure where you position Samsung or Xiaomi, Oppo etc. They're competitive on price with chipsets that can handle AI loads in the same ballpark, as attested by Google's features running on them.
They're not vertically integrated and don't have the same business structure, but does it matter regarding on-device AI ?
Vertical integration matters for sure, but people often underestimate the scale in which this market is already skewed.
- Apple owns more than 50% of this market-segment, the annual sales of iPhones is roughly 200 Million units. In comparison, Samsung Galaxy S-series sits at roughly 20-25 Millions.
- Apple's is alone in the iOS ecosystem, while Samsung, Xiaomi and Oppo have to compete within the Android space every year. iOS is extremely sticky, which makes a certain volume of iPhones almost guaranteed to sell every year, at a lofty profit margin.
In comparison, Samsung always has to consider that the next BAD Galaxy-S might only sell a fraction of the previous one, because users might move horizontally to another Android brand (even to Pixel, a first-party product of their ecosystem provider). So Samsung cannot even make bets based on the sale of 20 million units, they are already at risk to make bets on the initial shipment-volume (~5 millions) because if the device doesn't sell they will have to PAY money to the carriers to get them into the market.
Apple has a much lower risk here. If the next iPhone is not catching on, Apple will likely still sell 200mn iPhones in that year, because the ecosystem lock-in is so strong that there is little risk of losing customers to anything else than ANOTHER (then more-profitable) iPhone.
So even when assuming a MASSIVE annual drop of 25% in Sales, Apple can still make development bets based on a production forecast of 150 MILLION units.
For their supply-chain that's still an average production output of ~400k units per DAY for each component. With that volume you can get entire factories to only produce for you.
That's why I can't think of any company in a comparable position. Apple can add hardware to their device and sell the resulting product to the consumer for profit before delivering any actual value with it.
If any competitor in the Android space attempts that, just the component costs alone will risk the device to be dead-on-arrival just because "some other Android device" delivers the same experience at lower cost.
I'm with you in that no other company is in the exact same position as Apple.
I read the original comment as positing no other company is positionned to forgot building their own AI platform and instead sell pricey terminals that can run second party local models.
From that POV, sur Samsung and others have competition, and theirargins won't be as large as Apple, but they also have a larger market and can work with less money (Samsung as a whole will never struggle to find more)
> 50% of this market-segment
If we're limiting the segment to phones that have enough RAM to run decent models, Apple users who haven't updated to the upper models (no SE, no 16e etc) recent years are all out of the picture. I didn't check the numbers, but wouldn't have expect them to be much ahead of Samsung and the beasties Android devices.
I agree that this is a reasonable perspective, but from my cursory understanding of the “shakeup” at Apple, I am not sure it is seen that way by the Board and Cook.
I don't want to imply that this is their only play or that it will even work out.
The EU (and others) already identified this general scheme of stiffling competition by "brokering" between the consumer and the free market, so outside of the US I'm not even sure how much Apple will be able to rely on such a strategy (again)...
I recently tried to figure out what their offerings currently are. I'm hoping for `efficent but performant AI compute-chips` by Apple ever since they kicked out Nvidia in 2015 (for the ML Models / Exploration parts bellow). It will be interesting to see how good their products will feel in this fast-paced environment and how much legroom (RAM + Compute) will be left non-platform offerings.
To my understanding, they market their ML stack as four layers [1]:
- Platform Intelligence: ready-made OS features (e.g., Writing Tools, Genmoji, Image Playground) that apps can adopt with minimal customization.
- ML-powered APIs: higher-level frameworks for common tasks—on-device Foundation Models (LLM), plus Vision, Natural Language, Translation, Sound Analysis, and Speech; with optional customization via Create ML.
- ML Models (Core ML): ship your own models on-device in Core ML format; convert/optimize from PyTorch/TF via coremltools, and run efficiently across CPU/GPU/Neural Engine (optionally paired with Metal/Accelerate for more control).
- Exploration/Training: Metal-backed PyTorch/JAX for experimentation, plus Apple’s MLX for training/fine-tuning on Apple Silicon using unified memory, with multi-language bindings and models commonly sourced from Hugging Face.
- Everyone else: "We mainly build huge AI compute clusters to process large amount of data and create value, at high cost for ramp-up and operation."
- Apple: "We mainly build small closed-down AI compute-chips we can control, sell them for-profit to individual consumers and then orchestrate data-processing on those chips, with setup and operational cost all paid by the consumer."
I can't think of any company which has comparable know-how and, most of all, a comparable sell-out scale to even consider Apple's strategy.
No matter what they do, they will sell hundreds of millions compute devices for the foreseeable future. They use this to build out AI infrastructure they control, pre-paid by the future consumers.
THIS is their unique strength.