Mark Zuckerberg, CEO of Meta Platforms, is leading a new team focused on artificial general intelligence (AGI), internally called “superintelligence.”

The group will include around 50 experts, including a new head of AI research.

Zuckerberg has personally met with potential recruits at his homes in Lake Tahoe and Palo Alto. The team will be based near his office at Meta’s Menlo Park headquarters.

Meta also plans a major investment in Scale AI, a data service provider used to train models. The deal could exceed US$10 billion, becoming Meta’s largest external investment.

This move follows concerns over the performance of Meta’s Llama 4 model and delays in launching a larger model called “Behemoth.”

.source-ref{font-size:0.85em;color:#666;display:block;margin-top:1em;}a.ask-tia-citation-link:hover{color:#11628d !important;background:#e9f6f5 !important;border-color:#11628d !important;text-decoration:none !important;}@media only screen and (min-width:768px){a.ask-tia-citation-link{font-size:11px !important;}}

🔗 Source: Bloomberg

🧠 Food for thought

1️⃣ Meta’s AI restructuring mirrors industry shifts toward specialized AGI teams

Zuckerberg’s creation of a dedicated “superintelligence” team reflects a broader industry pattern of separating foundational AGI research from product-focused AI development.

This mirrors Meta’s previous AI reorganization that divided teams into “product-focused” and “AGI research” units to accelerate innovation while competing with OpenAI and Google 1.

The strategy appears to be a direct response to perceived shortcomings in Meta’s existing AI programs, particularly the underwhelming reception of Llama 4, which received mixed reviews despite high expectations 2.

These specialized teams represent a significant shift from Meta’s earlier AI research structure under FAIR (Facebook AI Research), which was organized around broader technical domains rather than AGI-specific goals 3.

By personally recruiting AI talent and physically positioning them near his office, Zuckerberg is demonstrating a management approach similar to what we’ve seen during previous strategic pivots, suggesting the importance he places on AI leadership.

2️⃣ Meta’s standalone funding approach contrasts with competitor financing models

Zuckerberg’s pitch to recruits about Meta’s financial self-sufficiency represents a strategic advantage in the capital-intensive AI race compared to companies dependent on external funding rounds.

Meta’s approach of leveraging its advertising business to finance AI development (including multi-gigawatt data centers) offers stability while other AI companies face challenging fundraising environments 4.

The financial commitment aligns with Zuckerberg’s previous statements about allocating “hundreds of billions” toward AI in the coming years, signaling Meta’s willingness to match or exceed competitors’ spending 5.

Meta’s self-funded model potentially provides greater autonomy in pursuing long-term AGI research compared to companies like OpenAI that must balance investor expectations and profitability concerns.

3️⃣ Meta’s superintelligence pursuit reflects evolving definitions of AGI success

Zuckerberg’s focus on “superintelligence” highlights how tech leaders are embracing increasingly ambitious AI goals, even as experts remain divided on near-term AGI feasibility.

While Meta pursues AGI, many AI researchers express skepticism about current methods leading to human-level intelligence, with a survey indicating significant doubts about timelines 6.

The push comes amid growing mainstream discussions about AGI’s implications, with the concept shifting from academic circles to broader public discourse.

Meta’s strategy appears to position AGI capabilities as competitive differentiators for consumer products – specifically mentioning integration with social media, chatbots, and Ray-Ban glasses – rather than pursuing AGI as an isolated research goal.

This approach to AGI differs from Meta’s earlier AI research under FAIR, which emphasized open research and community collaboration rather than proprietary superintelligence development 3.

Recent Meta developments

……