Apple is facing a growing challenge within its AI division, as key researchers from its foundation models team continue to exit the company. In the past month alone, four experts have departed to join Meta’s Superintelligence initiative. Among them is Bowen Zhang, a specialist in multimodal AI who most recently led internal efforts around large-scale models. He follows Ruoming Pang, previously a team lead at Apple, along with Tom Gunter and Mark Lee, both engineers closely involved in early Apple Intelligence development.
The wave of exits comes at a time when Apple is actively restructuring its internal AI leadership. Responsibilities for AI have shifted across departments, including Craig Federighi’s software group and Mike Rockwell’s technology development teams. Despite these changes, reports suggest morale within the foundation models team has been declining.
While Apple has reportedly offered retention bonuses, compensation still falls short of the aggressive offers made by Meta, which is using a mix of high salaries and long-term vision to attract top researchers. Meta’s approach is more open and experimental, in contrast to Apple’s historically closed and privacy-driven AI culture.
There’s also growing uncertainty inside Apple about the future direction of its AI efforts. The company is reportedly weighing a hybrid approach – continuing internal development of models while also licensing third-party systems from OpenAI and Anthropic. This mixed strategy has led to internal friction and confusion among researchers.
Losing multiple senior AI engineers in such a short span raises serious questions about Apple’s ability to lead in foundational model development. With next-gen Siri and Apple Intelligence features already delayed to 2026, the company now faces the additional challenge of rebuilding its AI leadership from within.
The departures could shift Apple’s position in the broader AI race, especially as rivals like Meta, Google, and Microsoft double down on in-house model development. Unless Apple can clearly define and commit to its AI roadmap, it risks falling behind – not just in innovation, but in attracting and retaining the people required to deliver it.
via Bloomberg
