Meta’s ‘highest-paid employee’ Alexandr Wang on the WhatsApp message from Mark Zuckerberg that changed his career: Have time for a...
In April 2025, Alexandr Wang was sitting in a staff meeting at Scale AI, the data-labelling company he co-founded, when his phone pinged.
In April 2025, Alexandr Wang was sitting in a staff meeting at Scale AI, the data-labelling company he co-founded, when his phone pinged. It was a WhatsApp message from Mark Zuckerberg: "Have time for a call?" Wang's first thought, as he tells it now, was simply, "huh, interesting."That call ran for 44 minutes. Within two months, Meta had put $14.3 billion into Scale AI for a 49% stake and brought Wang in to run its AI efforts. Wang and Zuckerberg retold the story together on Cheeky Pint, the podcast hosted by Stripe co-founder John Collison, in an episode recorded around the release of Meta's new Muse Wonder AI model.The timing of that message matters. Meta had released its Llama 4 models on April 4, 2025, to a lukewarm response. Around April 20, Zuckerberg said, he started calling people across the industry for advice on what to do next. Wang was one of them.Alexandr Wang's memo told Zuckerberg to hire fewer, better peopleTwo days after the call, Wang sent Zuckerberg a long memo. Zuckerberg described it on Cheeky Pint as Wang laying out what he would do "if I were in your position."The first point was about people, and how few of them Meta needed. "The very first thing is that you don't need a lot of people, and talent density is the thing that matters," Zuckerberg recalled. That idea still shapes Meta Superintelligence Labs. Its core research team numbers under 200, "closer to 100," according to Wang, inside a company of roughly 75,000 employees.The hire raised eyebrows. Wang, now 29, is an MIT dropout who built Scale AI into a key data supplier for the big AI labs, but he had never trained a frontier model. Collison pointed this out on the show. Zuckerberg said he spoke to the heads of research at all the major labs before deciding, and came away sure: "Alex is the guy."What followed was a recruiting sprint Zuckerberg called "supernova." He cleared his calendar for a month or two just to meet researchers. "I don't think either of us slept that much for the first couple of months," he said.Wang also recalled one early meeting that left him uneasy. When he said Meta's first priority should be a strong reasoning model, one of the existing leads asked whether consumers needed one at all. Wang said he called Nat Friedman, now head of Meta's AI products, straight after.Meta's CFO warned Wang the AI bet had to workBy the end of 2025, the numbers had become enormous. Zuckerberg said Meta was preparing to "write this check for like, I don't know, like $100 billion, $200 billion" before anyone had seen the new models. "I'm like, Alex, are you sure this is going to work?"At an executive gathering in Hawaii, Wang said, CFO Susan Li pulled him aside several times to drive the point home. "If it's good, it's all good. We're great. But if it's bad, it's really, really, really bad." Wang admitted that all he had to show at the time was a "vibe report" on research progress and a few plots.Muse and Muse Wonder have put Meta back in the AI raceThe bet produced Muse Spark in April 2026, followed by the Muse personal agent app on September 8. Muse climbed to the top of Apple's App Store within days. Wang's own Muse, which he named Oiler, buys increasingly obscure teas from farms in Taiwan and Japan every week. Zuckerberg uses his to critique his MMA technique in the one-minute breaks between rounds.On October 12, Meta released Muse Wonder, the first model it calls frontier-class. For now, it sits behind paid Muse subscriptions. Zuckerberg said the team was "very, very critical" of its own work along the way, and Wang describes the lab as a room full of "insecure overachievers."Asked what kind of boss Zuckerberg is, Wang called him "very Bayesian," someone who updates his beliefs and then moves the whole company to match. A five-word WhatsApp message, and the 44-minute call that followed, is about as clear an example of that as Meta has.You use AI every day. Now get your AI Quotient. Take the AIQ test.
Topics in this story
Gathered from external sources. Rights to this text belong to whoever originally published it.