/Tech2h ago

Jason Calacanis urges Apple to build AI workstations optimized for local open-source models with cloud routing

The strategy aims to counter Chinese open-source LLM dominance.

58384343227K
Original post
@jason@Jason#416inTech

Given token economics, we really need @apple’s new ceo to go all in on workstations that can run local, open source models

Ideally, with a router that can flip between local models and frontier models when the former gets stuck.

And America needs an open source champion — we really should not be comfortable with the Chinese owning the open source LLM market to the extent they do

5:13 PM · Jun 11, 2026 · 27.2K Views
Sentiment

Many users endorsed Apple building workstations for local open-source AI models due to environmental benefits and strong current hardware performance, while some dismissed Apple as unsuitable or behind competitors like Nvidia.

Pos
72.2%
Neg
27.8%
19 comments with sentiment.
Cluster Engagement
Posts from X
Most Activity
Most Activity
VIEWS811LIKES12REPLIES1
clem 🤗@ClementDelangue

@Jason @Apple I endorse this message!

@jason@Jason

Given token economics, we really need @apple’s new ceo to go all in on workstations that can run local, open source models

Ideally, with a router that can flip between local models and frontier models when the former gets stuck.

And America needs an open source champion — we really should not be comfortable with the Chinese owning the open source LLM market to the extent they do

2hViews 811Likes 12Bookmarks 0
RETWEETS1
Rehat@R3H8T

@ClementDelangue @Jason @Apple Genuinely curious: without all the wishy washy nationalist language, why are you worried about it? Open is open. It doesn't matter where it originates from.

1hViews 26
🇺🇸 Santore@santoretech

@Jason @Apple More local, open-source models equals less energy and data center impact. Local LLM's change the web entirely so much less has to be fetched.

2hViews 85
Evergreen@evrgn11112231

@Jason @Apple Some interesting technical feedback on this topic here:

1hViews 714Likes 1
Alexander Heinz@BigAl_PHD

@Jason @Apple I just want to buy a 128 unified memory m4 Max rn. It’s literally impossible to find

49mViews 8
Ed@Eduardopto

@Jason @Apple pretty sure apple cares more about thermal constraints and keeping user data inside the walled garden than building hardware just to appease the open source crowd. they definitely prefer a locked ecosystem over us having full control over the weights.

1hViews 97Likes 1
PJ van Dyk@vanDykPJ

@Jason @Apple macOS 27 includes a System Orchestrator that routes between on-device Apple foundation models and Private Cloud Compute. You can also build your own smart router with the new Foundation Models framework + Core AI + MLX to mix local and frontier models.

1hViews 24Likes 2

@Jason @Apple Arguably both Microsoft & Apple were built on "But what if you run this at home on your own device?" to begin with.

That was the Personal Computer revolution.

Round 2, Go!

1hViews 44Likes 1
Matt Torchia@MattTorchia

@Jason @Apple We had one. Then they decided they’d rather make money instead. The American Way.

2hViews 67
Berke Demiralp@bedemiralp

@Jason @Apple They already built the workstation people wanted: the 512GB M3 Ultra Mac Studio. Then they killed it. I hope they see the opportunity staring them in the face.

1hViews 59
Aaron@dicemanorama

@Jason @Apple My vibe-coded harness does this

1hViews 12Likes 1
Steven Pu@reedvoid

@Jason @Apple I'd love to run a local model on iPhone to enable an AI dungeon-master, waiting for next-gen hardware to come out, hopefully soon. 12gb memory is a great direction.

2hViews 40
Chad Brewbaker@SMT_Solvers

@Jason @Apple Just load @Alibaba_Qwen 3.6 27B and Hermes Agent. It works now on Mac Studio 64GB.

1hViews 11Likes 1
Macro Bombastic@MacroBombastic

@Jason @Apple Apple's never gonna give us that kind of freedom with their hardware, but the push for local models is the right bet. Just don't sleep on the market sorting it out without government help.

1hViews 10Likes 1
Eric Stewart@erixstewart

@Jason @Apple Yes! 🙌

1hViews 9Likes 1
doge@realuniquedoge

@Jason @Apple Fate loves irony.

2hViews 30
Lenny Ray Martinez@realLennyRay

@Jason @Apple They should own cusumer hardware

1hViews 7Likes 1
Neal@nepomo0

@Jason @Apple Yeah but how? We will probably need a RAM Act (à la the Chips Act) to get us producing enough RAM and VRAM to accomplish this endgame. 128 GB of RAM isn’t gonna cut it. Need several TBs on each work station.

45mViews 19
muz@ozjaus

@Jason @Apple No money in open source. Who is going to build it for free?

1hViews 18
Arun Rao@sudoraohacker

@Jason @niubi @Apple Dude, see all the laptops, desktops, and RTX Sparks (Mac Mini on steroids) that Jensen is releasing at Nvidia - Apple is far behind.

1hViews 16
Load more posts