Key Point Summary – Zuckerberg To Build Super Intelligence
• Mark Zuckerberg is personally hiring for a new AI “superintelligence” team
• Meta aims to beat OpenAI and Google to artificial general intelligence
• Llama 4 disappointment triggered Zuckerberg to take control
• Meta to invest billions, including a deal with Scale AI
• Zuckerberg’s vision includes superhuman AI woven into everyday tools

Zuckerberg Goes Full Founder Mode To Build Super Intelligence

Mark Zuckerberg has a new obsession: superintelligence. Not content with playing catch-up in the generative AI race, the Meta CEO is assembling a secretive new team to make artificial general intelligence (AGI) a reality—and he’s doing it hands-on, one hire at a time.

Frustrated by Meta’s lagging progress and the underwhelming debut of Llama 4, Zuckerberg has entered what insiders call “founder mode.” In recent weeks, he’s personally met with dozens of AI experts, pitching them over dinner at his California homes or pulling them into group chats where recruiting happens in real time.

He’s not delegating this. He wants to own it.

The goal? To ensure Meta builds what he calls super intelligence before rivals like OpenAI and Google. That means AI capable of reasoning, learning, and performing a wide range of human tasks—maybe even better than we can.

Personal Pitches And Big Promises

Inside Meta, the effort is called the “superintelligence group,” and it’s anything but typical. Sources say Zuckerberg wants about 50 elite hires, including a new head of AI research, who will work directly under him and sit close to his desk in Menlo Park.

One tech entrepreneur who declined a formal job but attended one of Zuckerberg’s “recruiting parties” said the CEO is “treating this like Facebook’s founding.” Another AI engineer described a WhatsApp group called “Recruiting Party,” where senior Meta execs ping each other around the clock about targets.

This isn’t just hiring—it’s a full-blown war for brains.

And Zuckerberg has ammo. Unlike some of his rivals, he tells candidates Meta doesn’t need to fundraise for this push. Its advertising empire, he argues, generates enough cash to build a multi-gigawatt data center and fuel the largest AI training clusters on the planet.

Llama 4 Letdown Sparks A Pivot

Until recently, Meta was all-in on Llama, its series of open-source large language models. But the release of Llama 4 in April flopped. Internally, staff complained about inconsistent performance, and externally, developers called the models overhyped.

That, sources say, was the last straw for Zuckerberg.

While Meta had hyped a massive new version, dubbed “Behemoth,” the company quietly shelved its release. Leadership feared the model hadn’t made meaningful gains. “It felt like putting out a sequel that nobody asked for,” said one engineer.

That led Zuckerberg to take control.

The CEO, insiders say, believes Meta has all the pieces—money, data, scale—but lacks the aggressive focus needed to beat companies like OpenAI, Anthropic, and Google. By building a separate team and leading recruitment personally, he’s signaling that AI isn’t just another product. It’s the product.

The Scale AI Deal And Vision For Integration

Zuckerberg’s most significant move may be the pending multibillion-dollar investment in Scale AI, a data labeling firm that helps train complex models. The deal is expected to be Meta’s biggest external investment yet and would bring Scale CEO Alexandr Wang into Zuckerberg’s inner circle.

Sources say Wang is likely to join the superintelligence group once the deal closes.

The hope? That Scale AI will supercharge Meta’s training data, allowing its future models to leapfrog the current generation. Meta already spends tens of billions annually on AI infrastructure, and Zuckerberg has said those investments will soon top “hundreds of billions.”

What would that buy? A vision where Meta’s super intelligence powers everything—from chatbots in Messenger to smart AR glasses that read your surroundings, offer advice, and possibly even anticipate your emotions.

Zuckerberg wants AGI to do more than answer questions. He wants it to know you.

Corporate Drama, Internal Resistance

While Zuckerberg’s push has thrilled some, others inside Meta feel blindsided. Long-time AI researchers at the company worry that the new team will compete with or override their efforts. There’s no clear blueprint yet for how the superintelligence group will work with Meta’s existing AI units.

“There’s definitely confusion,” said one staffer. “It’s exciting, but also chaotic.”

Meta’s AI division has already been under intense pressure. Staff were reportedly working nights and weekends to meet deadlines for Llama 4, only to see it fall flat. Morale took a hit. Now, with a new team forming outside the existing hierarchy, some worry they’ll be sidelined.

But Zuckerberg isn’t backing down.

In one internal meeting, he told staff that Meta has “one shot” to lead this race. “The company that solves AGI first,” he said, “will shape the next 50 years of technology.”

Big Risks, Bigger Stakes

Of course, building AGI—or super intelligence—is still more science fiction than science fact. While narrow AI tools like ChatGPT have exploded in use, true AGI remains elusive. Critics say the field is full of hype and unrealistic promises.

Zuckerberg knows this. But he’s betting the risk is worth it.

Meta’s rivals are making similar plays. OpenAI is pushing ahead with GPT-5. Google DeepMind continues to make breakthroughs. Even Elon Musk is expanding xAI with a mix of public bluster and private recruiting.

But none of them are running their recruitment like Zuckerberg is—out of their own living rooms.

He’s also one of the few founders still in control of a trillion-dollar tech empire. If anyone can force a company to pivot hard, it’s him.

Why It Matters To Everyone

This isn’t just a story about corporate ambition. Zuckerberg’s plan to build super intelligence could reshape how billions interact with technology. Imagine personalized assistants that manage your day, glasses that explain the world around you, and AIs that know your health, habits, and history better than you do.

It’s thrilling—and terrifying.

Privacy advocates are already raising flags. Meta doesn’t have the cleanest record when it comes to handling user data. Giving it an AI brain with access to your life feels like a sci-fi plot waiting to go wrong.

But Zuckerberg sees no alternative. “If we don’t build it,” he reportedly told one recruit, “someone else will. And they might not care what happens next.”

The Road Ahead

No one knows if Meta will succeed. However, with Zuckerberg making super intelligence his personal mission, the stakes are higher than ever.

In tech, bets like these either change the world or explode spectacularly.

Either way, the countdown has started.