Rolfe Winkler, reporting for The Wall Street Journal last Monday (Apple News+):

Mark Zuckerberg added another big name to Meta Platforms new “Superintelligence” AI division, hiring a top Apple AI researcher as part of a weekslong recruitment push, according to a person familiar with the hire.

Ruoming Pang is the first big name from Apple to jump over to Meta’s Superintelligence Lab, a blow to the iPhone maker, which is working to improve its own AI products. Pang, who led Apple’s foundation model team, is set to receive a pay package from Meta in the tens of millions of dollars, said the person.

Meta is offering huge pay packages—$100 million for some—to attract talent to the unit, which is led by former Scale Chief Executive Alexandr Wang after Meta made an investment in his company valuing it at $29 billion.

Zuckerberg, Meta’s chief executive, has never been one to inspire a sense of creativity at any of his companies. Aside from the core Facebook app, all of Meta’s most successful products in the 2020s have come through acquisition: Instagram, WhatsApp, and Meta Quest, née Oculus. Facebook is the app for racist boomers who don’t know how computers work, but Instagram and WhatsApp are core pillars of the modern internet. Instagram is the most important social network, if you ask me, with YouTube and TikTok very closely behind in second and third place, respectively. People care about celebrities, and all the famous people use Instagram all day long. WhatsApp, meanwhile, is how the entire world — except, notably, the United States, which Zuckerberg is infuriated by — communicates. Businesses are built on WhatsApp. The Oculus acquisition was the precursor to Meta’s most successful hardware product ever, the Meta Ray-Ban sunglasses.

None of these technologies can be attributed to Zuckerberg because, if they were his, they would be garbage. People overestimate Facebook’s importance, in my opinion. While yes, it did — and continues to — have a stronghold over social networking, it really lives in a silo of its own. The first “true” global social network was Twitter, quickly followed by Instagram, which now remains the preeminent way for notable people around the globe to share what they’re up to. Threads and X, previously Twitter, have a stronghold over the news, celebrity gossip, and “town square” section of the internet. Facebook is where people go to communicate with people they already know, whereas Instagram and Twitter were always the true pioneers of modern social networking. (I’m inclined to include YouTube in this, too, but I feel YouTube is more of a television streaming service than a social network, especially nowadays.) I wouldn’t say Facebook is a failure — because that’s a stupid take — but Zuckerberg is not the inventor of social networking. Jack Dorsey, the founder of Twitter, is, as much as I despise him.

Now, Meta’s latest uphill battle is artificial intelligence, and as usual, Zuckerberg’s efforts are genuinely terrible. They’re not as bad as Apple, but they’re close. The latest version of Meta’s most powerful model, Llama 4, was so bad that Meta had to put out a specially trained version to cheat on benchmarks with. Naturally, Zuckerberg’s instinct to remedy this is by doing some good-old-fashioned business, buying out talent for obscene prices and conquering the world that way. If the Biden administration were in power right now, Zuckerberg’s shenanigans would be shut down by Washington immediately, because they’re just blatantly anticompetitive. But because laws no longer exist under the current regime, Zuckerberg gets away scot-free with paying AI researchers $200 million to come work for Meta. Scale AI was once an independent company contracted by Google and OpenAI, but not anymore, because it’s effectively controlled by Meta for the low price of $29 billion.

As much as I want to, I can’t put the blame entirely on Zuckerberg, only thanks to this sliver of reporting from Mark Gurman at Bloomberg:

Pang’s departure could be the start of a string of exits from the AFM group, with several engineers telling colleagues they are planning to leave in the near future to Meta or elsewhere, the people said. Tom Gunter, a top deputy to Pang, left Apple last month, Bloomberg reported at the time.

The Apple Foundation Models team, or the AFM group, should be the last team to hemorrhage staff at Apple right now. It might be the only thing left to save Apple from impending doom, i.e., falling so far behind in AI that it can never recover. Not only is Apple unwilling to pay its top researchers competitively, but its senior leadership also has no interest in catering to their needs. I still can’t get over that reporting from a few months ago that said Luca Maestri, Apple’s then-chief financial officer, declined the AI group’s request for graphics processing units because it supposedly wasn’t worth the money. Who gave the finance nerd the discretion to make research and development decisions? Just thinking about it now, months later, makes me irrationally livid. Just pay the researchers as much money as they need before Apple no longer has a fighting chance. I really do think this is life and death for Apple — it either needs to hire some third-party AI company, or it has to start paying its researchers. They’re the bread and butter of the AI trade. How is this happening now?

Apple isn’t OpenAI, Google, or Anthropic — three successful AI companies with talented engineers and market-leading products. All three firms have lost researchers to Zuckerberg’s gambit in the last month, and while that’s bad for them, it’s even worse for Apple, which is playing on the same level as Meta. If Apple were an established AI company, then this wouldn’t really be that big of a deal. But if you’re a cutting-edge AI researcher with a Ph.D. in machine learning or whatever under your belt, I don’t see why you’d go work for Apple — which is losing engineers presumably for some reason — instead of Zuckerberg, OpenAI, or Google. Meta’s paying hundreds of millions, and Google and OpenAI are established AI companies — what competitive advantage does Apple have here?

I wouldn’t say Zuckerberg is playing chess while everyone else is playing checkers — he’s just cheating at checkers while nobody’s looking. It’s just that Apple, which remains in the same seeded bracket as Meta, isn’t playing at all. The takeaway here is that Apple has to start playing, not that Zuckerberg is doing something unusual. He isn’t — he’s playing out of the same playbook he’s had for decades.