The AI talent war, Apple is "bleeding" severely, the core team is being "emptied" by OpenAI and Meta

Apple is facing severe challenges in the AI "talent war."

According to the Financial Times, since the beginning of this year, about a dozen Apple AI researchers have left to join competitors such as Meta, OpenAI, xAI, and Cohere, including several top researchers and team leaders.

Among them, the news of Ruoming Pang, head of Apple's foundational model team, jumping to Meta last month is particularly noteworthy, as it is part of Zuckerberg's strategy to attract industry heavyweights with signing bonuses exceeding $100 million.

Aaron Sines, AI recruitment director at talent recruitment firm Razoroo, stated:

"The departure of Ruoming Pang has a huge impact, sending a warning signal about confidence in Apple's future development. Many of our client companies are saying, 'Look at Apple, now is the time to recruit.'"

In the face of external doubts, Cook held a rare all-hands meeting last week, promising to increase investment in AI research and development, and emphasized that Apple needs to "win" in the AI field.

Core team continues to lose members, foundational model research impacted

The list of Apple AI talent losses continues to grow.

Reports indicate that OpenAI has poached Apple foundational model research engineers Brandon McKinzie and Dian Ang Yap in recent months, while Canadian AI startup Cohere recruited machine learning scientist Liutong Zhou in June.

In addition to team leader Pang, several senior members of Apple's foundational model team have also joined Meta, including Mark Lee, Tom Gunter, Bowen Zhang, and Shuang Ma. British researcher Floris Weers also left in July to join a unnamed startup.

It is understood that most of these departing personnel were previously involved in Apple's AI model research papers released last year, when Apple began showcasing its research achievements in this field.

Reports cite insiders saying that Apple's core foundational model team is relatively small, with about 50 to 60 people, under a larger AI and machine learning department.

Siri upgrade hindered, major internal restructuring

Apple's challenges in the AI field are not limited to talent loss. The company is struggling to upgrade the iPhone voice assistant Siri using cutting-edge large language models to provide smarter voice interaction responses, but progress has not been smooth.

Due to delays in the release of the new Siri, there has been a significant internal restructuring at Apple. The responsibility for Siri's development has been transferred from AI expert John Giannandrea, who was poached from Google in 2018, to Mike Rockwell, head of the Vision Pro project.

The upgrade of Siri is an important part of the "Apple Smart" product lineup. This set of AI features was announced at last year's Worldwide Developers Conference, aimed at boosting hardware sales Bank of America analyst Wamsi Mohan pointed out that while Apple is known for not rushing to launch imperfect features, "the pace of external AI development is extremely fast," and Apple's slower approach may leave it "far behind" in terms of user expectations.

Industry Salaries Rising, Talent Competition Intensifying

The fierce competition in the AI talent market has driven up salary levels across the industry.

Sines noted that after news broke of Apple employee Pang receiving a high salary, he saw a significant surge in interest from AI researchers in new job opportunities.

Sines added:

"There are only about one to two thousand people in the world who truly have foundational model experience and know how to develop and deploy foundational models."

This scarcity has further intensified the talent competition among tech giants. According to previous media reports, Microsoft has poached over 20 AI employees from Google's DeepMind research division.

Randstad Digital CEO Graig Paglieri stated:

"Companies now view elite AI talent as a strategic asset—equally important as intellectual property or even entire business units—and are pursuing this talent with the intensity and valuation frameworks previously reserved for acquisitions."