Meta Hires Brilliant Shengjia Zhao as Chief Scientist of AI Superlab

Key Points

• Shengjia Zhao becomes Chief Scientist at MSL
•  Meta poaches talent from OpenAI, Google, Apple
•  AI reasoning models set as top priority
• Meta’s $B investments include 1 GW AI cluster

Meta CEO Mark Zuckerberg announced that Zhao, a key figure behind OpenAI’s biggest innovations, will officially lead research at Meta Superintelligence Labs (MSL). This elite new unit is focused on building next-generation AI systems that move far beyond today’s capabilities.

Zhao, known for his work on ChatGPT, GPT-4, and OpenAI’s first AI reasoning model o1, has quietly been helping shape Meta’s new AI efforts. Now, his leadership role is formalized.

We are excited to announce that @shengjia_zhao will be the Chief Scientist of Meta Superintelligence Labs!

Shengjia is a brilliant scientist who most recently pioneered a new scaling paradigm in his research. He will lead our scientific direction for our team.

Let’s go 🚀 pic.twitter.com/D93KQWIvFl

— Alexandr Wang (@alexandr_wang) July 25, 2025

“Shengjia co-founded the lab and has been our lead scientist from day one,” said Zuckerberg in a post on Threads. “Now that our recruiting is going well and our team is coming together, we’ve formalized his leadership role.”

Zhao will work under Alexandr Wang, the ex-Scale AI CEO. While Wang isn’t a researcher, his operational experience gives MSL an edge in scaling infrastructure—something crucial as Meta prepares to rival OpenAI’s recent innovations like the GPT-5 upgrade rumored for August.

We’re excited to have @shengjia_zhao at the helm as Chief Scientist of Meta Superintelligence Labs. Big things are coming! 🚀

See Mark’s post: https://t.co/SL7h4sGfwx pic.twitter.com/DRnpqBe1wD

— AI at Meta (@AIatMeta) July 25, 2025

Shengjia Zhao to Lead Meta’s AI Reasoning Breakthroughs

Zhao’s appointment signals a major shift. Meta is placing big bets on AI reasoning, the next frontier in artificial intelligence.

At OpenAI, Zhao helped develop o1, a reasoning model seen as the next evolution in AI after large language models. While OpenAI continues to push boundaries, Meta had no clear rival in this domain—until now.

Shengjia Zhao — formerly of OpenAI — will be chief scientist at Meta’s new Superintelligence Lab https://t.co/uCS4R6YC8V

— Axios (@axios) July 25, 2025

Zhao’s job is clear: build a model that can think, reason, and solve problems—not just predict text. It’s a space heating up fast, especially as companies race to control how AI models are governed and aligned, seen in moves like the anti-woke AI executive order.

A Flood of Talent Joins Zhao

Meta’s not stopping with just Zhao. The company is on a hiring spree, recruiting:

• Jiahui Yu, Shuchao Bi, Hongyu Ren – key OpenAI researchers

• Trapit Bansal – Zhao’s collaborator on AI reasoning

• Multimodal AI researchers from OpenAI’s Zurich office

• Talent from Google DeepMind, Apple, Safe Superintelligence, and Anthropic

Zuckerberg himself has been emailing researchers directly, even inviting them to his Lake Tahoe estate. According to reports, some of Meta’s offers hit nine figures and include short windows to accept—known as “exploding offers.“

Infrastructure to Match the Ambition

Backed by billions, Meta is building serious infrastructure for MSL. By 2026, Zhao’s team will gain access to Prometheus, a 1-gigawatt cloud cluster located in Ohio. That’s enough power to supply over 750,000 homes and will support the massive model training runs needed for frontier research.

It’s an investment designed to help Meta catch up—and potentially outpace—initiatives like OpenAI’s Oracle deal, which gives OpenAI access to huge computing resources.

A New Era for Meta’s AI Research Strategy

With Shengjia Zhao at the helm of MSL, Meta now operates with two top scientists:

• Yann LeCun, head of Meta’s FAIR lab, focused on long-term AI

• Zhao, now driving Meta’s short- and mid-term breakthroughs in reasoning and general AI

MSL differs from FAIR in its focus on building models that are useful now, not just in a decade. This focus aligns with trends we’re seeing across the industry, including voice-powered and multimodal AI, like Le Chat’s new voice recognition feature.

Zhao’s experience in developing real-world-ready reasoning systems puts Meta in a strong position to compete with both OpenAI and Google. The real challenge now? Turning ambition into working models that can outperform today’s best.

Meta is betting big, and Zhao’s leadership could be the key to delivering results.