Apple and Meta Team Up to Run Muse Spark Open on Macs

Apple and Meta are collaborating to make the Muse Spark Open model easier to run locally on clustered Mac Studio systems, the two companies announced. That single sentence carries real weight: two of the industry's most competitive rivals are now working on the same infrastructure problem — getting large AI models to run on Apple silicon rather than in the cloud.

By Sophie Lindqvist

What did Apple and Meta actually announce?

According to the companies, the partnership is about local execution. Muse Spark Open, Meta's model, will be made easier to run on Mac hardware — specifically on clustered Mac Studio machines, where several units work together as one system.

The companies did not announce a product, a price, or a release date. What they announced is a technical collaboration, and the target hardware tells us most of the story.

The takeaway: Meta's model is being adapted for Apple's machines, and the two firms say so themselves.

Why does clustered Mac Studio hardware matter?

A single Mac Studio is a compact desktop built around Apple's own chips. One unit has serious memory bandwidth for its size, but a large AI model usually needs more than one machine can offer.

Clustering solves that. You link multiple Mac Studio units so they behave like a single, bigger computer. The model's workload spreads across the combined memory and processing power of every unit in the cluster.

That is the mechanism behind this deal. Muse Spark Open is big enough that running it comfortably on one desktop is a stretch, so Apple and Meta are working to make it run well across several linked Macs instead.

The takeaway: clustering turns several small desktops into one machine big enough for a large model.

Why would Apple and Meta work together?

The two companies compete in operating systems, in headsets, and in AI assistants. A joint engineering effort between them is unusual, and the fact that they announced it jointly, according to the companies, signals that both see value in the outcome.

For Apple, the upside is straightforward. If a well-known open model like Muse Spark Open runs well on Mac hardware, that is a concrete argument for buying Macs to do AI work — locally, on the desk, without a data center.

For Meta, the appeal runs the other way. Muse Spark Open is an open model. The easier it runs on widely available consumer hardware, the more developers and researchers can pick it up, test it, and build on it.

The takeaway: each company gets something it could not get as easily alone.

What does 'running locally' change in practice?

Local execution means the model runs on your machines, not on rented cloud servers. That shifts who controls the setup.

It moves the cost from ongoing cloud fees to upfront hardware. It keeps data on the cluster itself rather than sending it to an external service. And it makes the whole arrangement independent of a cloud provider's uptime, pricing, and policies.

None of those benefits is automatic, of course. Making a large model run smoothly across a cluster of desktops is an engineering problem, which is exactly the problem Apple and Meta say they are now working on together.

The takeaway: local clusters trade cloud bills and cloud dependencies for hardware you own.

Who is this for?

The target audience is not the casual user. Clustered Mac Studio systems are a serious investment — multiple desktop-class machines linked together — which puts this setup in reach of developers, research groups, and companies that want to run AI models on their own infrastructure.

The companies have not detailed the tools, timelines, or support they will provide. The announcement, per the companies, establishes the collaboration and its goal: making local execution of Muse Spark Open on clustered Mac Studio systems easier than it is today.

The takeaway: this is infrastructure news for people who run models, not a consumer feature.

What comes next?

The announcement leaves the specifics open. Neither Apple nor Meta has published dates, supported configurations, or a technical roadmap — only the shared goal, stated by the companies, of easier local runs on clustered Mac Studio hardware.

The direction, though, is clear. If Apple and Meta deliver, Muse Spark Open becomes materially more practical to self-host on Apple silicon, and clustered Macs become a more credible alternative to cloud GPUs for open-model workloads.