Meta AI Chief Alexandr Wang has credited former Apple executive Daniel Gross with helping establish an unusual collaboration between Apple and Meta. The partnership allows Meta’s Muse Spark Open artificial intelligence model to run on Apple’s high-end Mac Studio computers.
Wang told Axios that Gross, who now works at Meta, played an important role in bringing the companies together. The collaboration has been under development for several months, according to Wang.
Daniel Gross Helped Bridge Apple and Meta
Wang acknowledged Gross’s contribution to building cooperation between the two technology companies. “I think he played a really big role in kind of getting the two companies to work nicely together,” Wang told the publication. He also responded to the renewed cooperation with a light-hearted remark. “We're excited for mommy and daddy to be playing nice again,” he said.
Apple and Meta have a long history of disagreements, particularly over user privacy and control of digital platforms. Their tensions became more visible in 2018, when Apple CEO Tim Cook criticised Facebook’s handling of the Cambridge Analytica scandal.
Meta CEO Mark Zuckerberg later described Cook’s remarks as ‘extremely glib.’ Relations deteriorated further in 2021 after Apple introduced privacy changes that affected how companies tracked users across apps.
Also Read: Xcode for Apple Developers: Tools for Building iOS Apps
Meta’s Muse Spark Open Runs on Mac Studio
The collaboration centres on Muse Spark Open, an AI model containing 816 billion parameters. Meta describes it as its most capable US-developed open-weight model to date. The model can operate across two or four connected Mac Studio computers equipped with Apple’s M5 Ultra chips. Apple’s MLX framework distributes the workload across the machines.
Muse Spark Open targets demanding tasks, including extended coding sessions, multi-step reasoning, and video analysis. Developers can download the model through platforms such as Hugging Face, subject to its licensing conditions. The local setup also offers a privacy advantage for users. Wang said Meta cannot access the prompts or files developers provide when running the model on their own computers.
‘Meta has no control or visibility to any of that,' Meta’s AI chief said. However, running the model requires substantial hardware investment. The configuration needs at least two Mac Studio systems, making the setup expensive for individual developers.