Apple Partners with Meta to Run 816-Billion AI Model Locally

The collaboration signals a shift in Silicon Valley alliances as rivals seek to advance local artificial intelligence capabilities on desktop hardware. For investors, the integration highlights Apple's emerging strength in AI hardware while providing Meta with a dedicated desktop ecosystem for its open-weight models.

Open-Weight AI Model Tuned for High-End Apple Hardware

The joint effort centers on Muse Spark Open, an 816-billion-parameter model using a mixture-of-experts architecture, according to CNBC Markets. Meta described it as the most capable open-weight model released under a license permitting limited commercial use. As an open-weight model, developers can download its parameters from sites like Hugging Face without extra configuration for Apple's silicon.

To run the model locally rather than in the cloud, customers will need to pair between two and four Mac Studio PCs equipped with Apple's M5 Ultra chips and maximum memory. Apple's hardware makes enough memory available to make running such a large open-weight model practical, Meta AI chief Alexandr Wang told CNBC Markets. Running the model locally provides an added layer of security and saves users money when performing coding, video analysis, and agentic work.

However, the hardware setup requires a substantial financial commitment. A Mac Studio configured with the required M5 Ultra chip and 256GB of memory costs at least $10,799, with a 512GB RAM version scheduled for release later this month. Because running Muse Spark Open requires clustering multiple machines, developers must purchase between two and four setups.

On the stock market as of Thursday, Apple Inc ($AAPL) traded up 0.48% at $344.62, placing the stock above every close in its 261 sessions on record and on track for a fourth straight daily gain. MetaDAO ($META) stood at $4.98, down 7.07%.

Shift in Silicon Valley Leadership and Alliances

The announcement marks a notable realignment between two tech giants whose executive leadership has historically clashed. Relations deteriorated after Apple introduced privacy changes in 2021, which Meta estimated in early 2022 would reduce its revenue by $10 billion. Public partnerships between the two companies had been absent since 2017, when Apple removed system-level Facebook integration from the iPhone.

Leadership changes at Apple have helped open the door to new collaboration. John Ternus succeeded Tim Cook as Apple CEO last month, bringing a fresh approach to hardware engineering and product releases. Cook, who frequently criticized Meta over user privacy, traded public insults with Meta CEO Mark Zuckerberg for years. Meanwhile, Meta overhauled its AI strategy under Alexandr Wang, who joined after Meta invested more than $14 billion into his former startup Scale AI last year, alongside key hires such as former Apple director Daniel Gross and former GitHub CEO Nat Friedman.

The collaboration also highlights shifting dynamics in the broader AI landscape. Meta has sought to challenge proprietary leaders like OpenAI and Anthropic, recently launching its Muse Wonder foundation model earlier this week following the release of the smaller Muse Glimmer in August. Meanwhile, Apple has expanded its partner network after integrating OpenAI's ChatGPT into Siri in 2024—a partnership Apple later sued OpenAI over in July alleging trade secret theft—and striking a deal with Google earlier this year to use Gemini models in Apple Intelligence.

What to Watch

• Monday, Oct 19: General Electric Company ($GE) Q3 2026 earnings before the market open (consensus EPS $2.06).
• Tuesday, Oct 20: Netflix Inc ($NFLX) Q3 2026 earnings after the market close (consensus EPS $0.84).
• Wednesday, Oct 21: Meta Platforms Inc ($META) Q3 2026 earnings after the market close (consensus EPS $6.31).
• Wednesday, Oct 21: Tesla Inc ($TSLA) Q3 2026 earnings after the market close (consensus EPS $0.45).

Related on rawfin.tv

• Apple Inc (AAPL) live price and chart
• Market Open — every episode