Key Takeaways

  • Mark Zuckerberg called Alexandr Wang after Llama 4 stalled internally, opening the conversation with a simple question: “What do you think we should be doing?”
  • To bring Wang in as Chief AI Officer, Meta structured an unusual deal to purchase a 49% stake in Scale AI while allowing Scale to keep operating.
  • Wang walked away from nine years running Scale AI, which he considered his life's work, because Meta's battered reputation in frontier AI created a turnaround opportunity.
  • Zuckerberg maintained only semi-annual contact with Wang as a vendor before the crisis, but moved rapidly to overhaul leadership once Llama 4 missed expectations.

The Cold Call After Llama 4

Before Llama 4 fell short, Mark Zuckerberg and Alexandr Wang spoke roughly once every six months. Scale AI was an important vendor for Meta, but Wang was not in Zuckerberg's inner circle. That dynamic shifted when internal reviews showed Llama 4 was missing its performance targets.

Wang remembers the initial outreach: “Meta was an important customer of Scale... prior to this instance, maybe we spoke once every like six months or so... And then he had called and it was actually not obvious at first what the end point was going to be, because the first conversation was after Llama 4, and it was kind of just asking, 'What do you think we should be doing?'”

Most executives double down on struggling projects in public while quietly reassigning blame behind closed doors. Zuckerberg took the opposite approach. He acknowledged reality early. As Wang noted, “Llama 4 was not on the trajectory that Meta wanted as a company. It definitely was, I think internally speaking, a disappointment.” Instead of defending sunk costs, Zuckerberg cold-called an external founder and asked for a blunt assessment.

Buying 49% to Hire One Operator

Recruiting Wang was not a standard executive hire. Wang had spent nine years building Scale AI from scratch. Leaving meant walking away from his identity as a founder.

“I did really genuinely think of Scale as my life's work for the whole time that I was working on it,” Wang said. “That was a tough emotional process.”

Zuckerberg solved the problem by building a creative deal structure. Meta acquired a 49% minority stake in Scale AI, keeping the business operational, while Wang and several colleagues moved directly to Meta. Wang took the role of Chief AI Officer, stepping into Meta Super Intelligence Labs with broad operational freedom.

Many founders avoid big tech companies because they dread corporate bureaucracy. What convinced Wang was that Meta was struggling. “This is a point at which Meta on AI certainly looked like damaged goods in many ways and was maybe not the most appealing place to work on AI,” Wang explained. “That kind of triggered my entrepreneur side.”

When a division looks like damaged goods, the internal status quo loses its protection. The board stops defending legacy managers, and the CEO grants wartime authority. Wang saw an opening to reshape a major tech lab because the previous roadmap had visibly failed.

What to Do With This

Audit your roadmap for projects that missed their internal milestones over the last two quarters. Do not ask the internal project lead how to fix it; call an outside founder who uses your product, share the unvarnished numbers, and ask what they would do if they took over tomorrow. If their answer requires restructuring your team, make the change this week instead of waiting for another review cycle.