Meta Launches Meta Superintelligence Labs: New Era of AI in 2025

Meta is taking the next step in the AI ​​race by creating a lab dedicated to superintelligence. The goal? To surpass human capabilities and deliver "personal superintelligence for all."

Creation of Meta Superintelligence Labs

For several months, Meta has been nurturing ambitious ambitions in artificial intelligence. The company led by Mark Zuckerberg has just reached a new milestone by formalizing the creation of Meta Superintelligence Labs. This unique laboratory aims to develop "superintelligence"-level AI systems, surpassing human performance and capable of reasoning, learning, and adapting autonomously and multimodally.

This initiative marks a strategic step in the rivalry between Meta, OpenAI, Google DeepMind, and Anthropic. It is part of Meta's desire to reposition its technology portfolio and become a key player in the field of advanced artificial intelligence. By bringing together its R&D teams from FAIR and its product divisions, Meta is completely restructuring its AI strategy to accelerate the development of Llama 4 and 5-type models, while designing cross-functional applications capable of simultaneously processing text, images, video, voice, and sensory data.

An Unprecedented War for Talent

To support this ambitious project, Meta is deploying colossal human and financial resources. The company has recruited several strategic talent from competing companies. Among them, Alexandr Wang, former CEO of Scale AI, will head the new lab, alongside Nat Friedman, former CEO of GitHub, who will oversee product direction and open source model strategy. This leadership team, with extensive experience in AI, will structure all advanced research projects and lead the integration of superintelligence models into Meta products.

At the same time, Meta has been conducting an aggressive recruitment campaign for several weeks, targeting talent from Google DeepMind, OpenAI, and Anthropic. According to several sources, nearly a dozen engineers and researchers who contributed to the development of GPT-4 and Claude have already joined the lab. This direct competition between technology giants heralds a fierce war for talent, where Meta does not hesitate to offer exceptional compensation packages of up to $100 million for the best profiles.

Record investments to support this ambition

Beyond human resources, Meta is investing massively to support its ambitions in the field of artificial intelligence. In 2025, the company plans to spend between $65 and $70 billion expanding its computing infrastructure, developing new data centers, and purchasing specialized GPU clusters for training very large-scale models. Meta also announced that it has acquired a 49% stake in Scale AI, consolidating its capabilities in processing and annotating critical data for AI training.

These financial resources aim to position Meta at the forefront of AI research, capable of competing with OpenAI, Google, and Anthropic, which are also accelerating the deployment of increasingly powerful models. With his Meta Superintelligence Labs, Zuckerberg aims to create multimodal models that would work not only on his social applications, but also in industrial, robotic, and immersive environments.

A strategy based on multimodal AI and superintelligence

The core of Meta's strategy is based on the development of multimodal models capable of understanding and producing content in different forms: text, images, videos, voice, and environmental signals. The idea is to offer integrated, autonomous, and customizable artificial intelligence that could be deployed both in Meta applications (Facebook, Instagram, WhatsApp) and in professional or healthcare use cases.

The company also aims to reach the stage of superintelligence: general-purpose AI surpassing human cognitive abilities on a wide range of tasks. This level of artificial intelligence, which is still a matter of debate in the scientific community, represents the technological horizon that Meta hopes to achieve within the next five years. The Menlo Park-based firm is increasing its scientific publications and technical demonstrations around its Llama 4.1 model, while announcing the advanced development of Llama 4.2, planned for early 2026.

Ethical Issues and Industrial Challenges

However, this strategy poses numerous challenges, both technically and ethically and regulatoryly. Meta will need to ensure it manages the biases, reliability, and societal impact of its systems, as increasingly autonomous AI raises questions of liability, data protection, and algorithmic transparency. The company will also need to manage the organizational complexity of integrating its new teams and coordinating international projects spread across California, Austin, and Europe.

Faced with growing concerns about generative AI, Meta is also committed to publishing some of its work as open source and collaborating with academic researchers to ensure the security and robustness of its models. The company thus intends to reconcile technological innovation and social acceptability to remain a key player in a rapidly evolving sector.

With the launch of Meta Superintelligence Labs ...

.... Zuckerberg clearly asserts his ambition to gain a head start in the global race for artificial intelligence. By leveraging unprecedented human, financial, and technological resources, and focusing its efforts on superintelligent and multimodal AI, Meta intends to establish itself as the driving force behind the next generation of universal intelligent assistants and systems.

This initiative confirms that the competition for control of tomorrow's AI is now being played out at the scale of super-labs and specialized cloud platforms. Meta, with its infrastructure, ecosystem, and open-source models under development, could well upset the established balance in the coming months.

Sources:

• CNBC — Zuckerberg memo on Meta Superintelligence Labs : https://tinyurl.com/CNBCmemo

• Houston Chronicle — Meta AI Labs opening : https://tinyurl.com/houstonChro