Meta nabs Apple AI chief – a win for Zuckerberg or a sign of Apple’s weakness?
Meta has reportedly secured another coup in the escalating race for AI dominance, poaching Apple’s head of AI models, Rooming Pang, in a deal said to be worth tens of millions of dollar a year.
While Pang’s defection might be viewed as a clear win for Mark Zuckerberg’s much-hyped ‘super intelligence’ project, it also casts a spotlight on Apple’s patchy progress in artificial intelligence, and a broader AI talent war that could ultimately reshape the pecking order in Silicon Valley.
This latest hire highlights the scope of Meta’s ambitions, the latest in its expensive recruitment spree as the tech giant looks to catch up with OpenAI and Google in the AI arms race. But the move is not without risk.
Meta’s ‘superintelligence’ dream
Pang’s signing follows a string of headline-grabbing moves by Zuckerberg, who is reshaping Meta’s AI division into a focused ‘superintelligence’ unit.
Its goal is to build AI systems that can eventually outperform humans at nearly every task – an ambition that has attracted both admiration and scepticism within the sector.
The unit is reportedly already stacked with heavyweight names from OpenAI, Google Deepmind, and Anthropic.
In the last 48 hours alone, Meta has also onboarded Yuanzhi Li, formerly OpenAI, and Anton Bakhtin, from Anthropic.
Meta has declined to comment on hiring, but Pang’s move is seen internally as a strategic leap in expertise around on-device and smaller foundation models – areas Apple has explored extensively, if not always successfully.
Pang led Apple’s foundation models team, responsible for building the core AI powering Apple Intelligence and the upcoming Siri upgrade.
His departure marks the most significant talent loss for Apple’s AI ranks since it launched its push to integrate the technology throughout its suite of products.
Zuckerberg has been directly involved in luring top minds like Pang, often reportedly meeting candidates personally – an approach some have seen as signalling commitment, but others have seen as a sign of Meta’s growing desperation to overcome repeated hurdles and personnel churn within its teams.
Apple’s AI ambitions exposed
While Meta’s splashy hires may be suggestive of a winning strategy, Pang’s exit also paints a more troubling picture for Apple.
His former team – known internally as the Apple Foundation Models (AFM) group – has reportedly been under pressure, facing morale issues, internal scrutiny, and growing interest from leadership in using third-party models from OpenAI or Anthropic to power Siri.
Sources have suggested Pang’s departure could be the first of several, with engineers eyeing exits amid unease over Apple’s shifting AI strategy.
The AFM team will now be led by Zhifeng Chen, under a restructured management model – a move seen by some insiders as damage control in an attempt to plug a leadership vacuum.
In contrast to Apple’s secretive and often cautious approach to product rollouts, Meta has embraced bold declarations and big spending.
The social media giant has pledged tens of billions toward AI infrastructure in 2024 alone, with plans to scale even higher.
Bloomberg Intelligence estimated that Meta, Amazon, and Microsoft will collectively pour over $370 billion into data centres and AI infrastructure next year.
But spending alone doesn’t secure innovation. Sam Altman, OpenAI’s chief, recently took a dig at Meta, suggesting it struggles with originality. “I don’t think they’re a company that’s, like, great at innovation,” Altman said.
Culture clash?
Meta’s recruitment approach, reportedly offering packages upwards of $100 million, has raised eyebrows, but also questions about sustainability.
Multiple AI insiders have criticised Meta’s frequent internal reshuffles, the churn of top AI talent, including members of the Llama team, and what they see as a lack of clarity in Zuckerberg’s long term AI vision.
Critics also argue that Meta’s AI output has, so far, lacked substance. While rivals launch high-performing, multi-modal models and tackle complex real-world problems, Meta has drawn ridicule for ‘gimmicky’ outputs like celebrity chatbots.
The company is still seen as playing catch-up rather than setting the pace.
Still, Zuckerberg appears undeterred. The Meta boss is betting that sheer firepower in capital, compute, and an elite roster of engineers, can vault the company ahead.
By targeting specialised talent in less crowded AI subfields, such as on-device optimisation and multi-agent reasoning, Meta may carve out an advantage where others focus on broader language models.
From OpenAI’s blockbuster deal with Jony Ive’s startup, to Google’s $2.7bn Character.AI acquisition, companies are no longer just chasing products, they’re acquiring people.
And in Pang’s case, his leap from Apple to Meta offers a revealing snapshot of a sector where talent may be even more valuable than models.