ChatGPT Desktop for Apple Silicon Only: A Ploy by Apple to Boost Hardware Sales?

1388b68ca8dc79bebac26b75db37f12c0373e7af_2_690x143

Apple is on the verge of completing a significant partnership with OpenAI to incorporate ChatGPT into iOS 18.

‘Every. to’ writer Evan Armstrong yesterday highlighted how ChatGPT will likely “…act as a meta-layer on top of all your applications” after OpenAI’s Spring Update on Monday. The concept is familiar to anyone using Alexa or Siri, but what stands out is how Apple seems eager to monetise this new relationship quickly.

The just-announced ChatGPT desktop app is already raising eyebrows. Inexplicably, it appears that it will require macOS 14+ with Apple Silicon and won’t work on Intel-based Macs, effectively forcing desktop users to upgrade their hardware to access the native app. While the web version will remain accessible on Intel Macs, the push towards Apple Silicon is hard to ignore.

This limitation is puzzling, considering most Windows PCs use the same x86 chip architecture as Intel Macs. If the ChatGPT desktop app were available to both Mac and PC users simultaneously (with Intel support), it would reveal that Intel can handle the app just fine. Apple users might then question why they need to upgrade their hardware.

It doesn’t take a hardened cynic to suspect this artificial Intel lockout was part of the deal to bring OpenAI further into the Apple ecosystem. Apple is leveraging a much-hyped new technology to drive unnecessary hardware upgrades and corral more users in its walled garden. It’s a lovely garden, of course, but one with increasingly pricey entry.

OpenAI has suggested that a Windows version will come later, hinting that Intel Mac support may eventually arrive. But this begs the question: why limit the initial release to Apple Silicon? The app is likely simple, mainly just calling an API and handling audio/screenshot input. This points to Apple’s strategic move rather than a technical necessity.

By first rolling out the ChatGPT desktop app to Mac users, Apple ensures an initial surge in hardware sales as dedicated users rush to upgrade. Later, an ‘Intel-ready’ version for PC users will likely emerge, almost identical to the current Mac app. This staggered release strategy boosts short-term sales and reinforces Apple’s control over its user base.

Apple’s approach underscores its confidence in its monopolistic position, showing little concern for keeping up appearances. While this strategy might yield quick profits, it risks alienating a user base already wary of being pushed into unnecessary upgrades. But Apple are masters of calculating commercial risk, and this move suggests they’ve already done the maths. Apple has every right to pursue profits, but rushing to squeeze more cash out of its customers feels opportunistic and myopic, especially before the ink has dried on the OpenAI deal.

It’s a clear example of strategic hardware enforcement under the guise of development.

2 Likes

Intel chips in Apple products are 4 years old. I personally do not develop for Intel anymore when I make macOS applications as most people have switched to a “newer” computer.

Most ChatGPT users probably use Apple computers.

Their partnership is recent while they have been working on this app for 3-5 months.

You can use chat.openai.com and some browsers allow you to turn websites into apps which work basically the same way.

Yes.

I don’t see how this hints at anything. Developing for Windows and Mac Intel is much different. They would have to rewrite all the code, so if they were planning on doing an Intel release they probably would have done that immediately.

I am going to ignore this and assume you have never developed applications for macOS.

One more thing, have you ever seen their ad videos? They all use Apple Silicon Macs.

2 Likes

Intel chips in Apple products are 4 years old. I personally do not develop for Intel anymore when I make macOS applications as most people have switched to a “newer” computer.

Thanks for taking the time to go through this! Understood. My observation was more around the seemingly strategic absence of a companion Windows desktop app from the outset.

Most ChatGPT users probably use Apple computers.

That may well be the case. But limiting it to Apple, even in the short-term, is unaligned with the rest of the commercial strategy announced in the Spring Update.

Their partnership is recent while they have been working on this app for 3-5 months.

Sure - but, your previous point notwithstanding, that doesn’t mean that the strategy to defer developing for Windows is any less unusual. Arguably more so, unless I’m misunderstanding you (which I suspect is likely).

You can use chat.openai.com and some browsers allow you to turn websites into apps which work basically the same way.

One of the most attractive aspects to me in Monday’s demo (and arguably one of the most glossed-over) was the screen-sharing feature of the desktop app. That will have a huge impact on my workflow. If this feature is introduced to the web version - none of what I wrote matters much. Perhaps I should have led with that!

I don’t see how this hints at anything. Developing for Windows and Mac Intel is much different. They would have to rewrite all the code, so if they were planning on doing an Intel release they probably would have done that immediately.

You’re absolutely right. Of course it won’t. I hadn’t had my coffee yet. Retracted.

I am going to ignore this and assume you have never developed applications for macOS.

I haven’t. I’m completely non-technical, so apologies if this came across as wildly ignorant or arrogant.

One more thing, have you ever seen their ad videos? They all use Apple Silicon Macs.

I have, and also observed that. But it would be weird if one of the the most future-forward companies in the world was demoing gear on anything that wasn’t the most current tech.

Thanks again for taking the time to share your knowledge.

2 Likes

First off, thanks for the response.

Ah, I understand why you would be frustrated, but this is actually not out yet and will probably not be out for a bit. Currently, the desktop application can do nothing more than the phone app. The only difference from a computer is that you can talk to ChatGPT, but then again, all the new voices and features that were shown are not out yet.

1 Like

I received this notice yesterday. However, it failed to dunload. (I may have lost signal in the process).
Can someone please provide me with the URL to the download site?

Here is the download link: https://persistent.oaistatic.com/sidekick/public/ChatGPT_Desktop_public_latest.dmg

2 Likes

Thank you very much! It went well.
I will now try to use it to its fullest.

as most people have switched to a “newer” computer.

I’d be interested to see stats on this. I doubt they would agree. A lot of big app publishers - e.g. Adobe - still support Intel Macs and have no plans to stop supporting them.

Their partnership is recent while they have been working on this app for 3-5 months.

While the news of their partnership is recent, I would be surprised if they haven’t been having conversations for much longer than 3-5 months.

Windows and Mac Intel is much different. They would have to rewrite all the code

There are plenty of examples of cross-platform apps. In fact, Apple moving to Intel opened those floodgates, there is a lot of tooling for this purpose.

I suspect the developers ported the iOS app and built from there, in which case their motive to only support Apple Silicon is clearer.

OpenAI want ChatGPT in the hands of everyone, I think it would be a misstep to build a Windows app and not support Intel-based Macs at the same time.

This did not happen.

OpenAI and Adobe are very different so it’s hard to compare.

As someone who owned a Silicon mac right after it came out most of this tooling had warnings that said it was temporary until developers could fully migrate their applications,

Yes, everyone has a web browser.


The official answer is that there are libraries that are exclusive to Silicon macs that they need.

It doesn’t make any sense because Intel processors, which equip all Windows computers, are indeed based on the x86 architecture, right? And the application for Windows is coming soon (let me remind you that Microsoft is a major shareholder in OpenAI…) Anyway, the version reserved for the M1 processor is a real scam…

Because by targeting Apple Silicon they’re able to ensure that no matter which generation of M-series chip it is, it has dedicated neural processing hardware.
It’s only the next generation of intel and amd processors that are launching with any kind of NPU capability.
Windows users definitely have far, far more powerful rigs when they invest the money, but unless you go RTX GPU, or specifically buy a new CPU with an NPU attached to it, you have very very little AI processing power on your PC. If any at all. It’s not ubiquitous. Deploying on Apple M-series hardware guarantees some degree of neural processing capability. The M1 could handle 11 trillion AI ops/sec. The M2 could handle 15.8 trillion AI ops/sec. The M3 could handle 25.3 trillion AI ops/sec, and the M4 can handle 38.5 trillion AI ops/sec.
That’s in addition to them being essentially the best performance per watt cpus in existence right now, absolutely curbstomping intel equivalents which use over 10x more power and require active cooling…

Idk. It just seems like when you’re working on the back-end of an app, and you’re testing in the public (where you can’t necessarily control the distribution of hardware you might have to run on), it’s probably best to do that test with a userbase which has relatively ubiquitous hardware.