Meta has made a significant financial commitment to the artificial intelligence (AI) landscape by investing $14.3 billion in Scale AI and enlisting its CEO, Alexandr Wang, to join Meta’s team working on advanced AI technologies, often referred to as “superintelligence.” This move underscores Meta CEO Mark Zuckerberg’s ongoing endeavor to rejuvenate the company’s AI capabilities amidst stiff competition from major entities such as Google and OpenAI.
The late Thursday announcement detailed a “strategic partnership and investment” that aims to enhance collaboration between the two companies. As part of the agreement, Scale AI will remain an independent entity, but Meta will acquire a 49% stake, a move that positions Scale’s market worth at over $29 billion.
While Wang transitions to Meta, he will still hold a position on Scale’s board of directors. Jason Droege, formerly Scale’s chief strategy officer with executive experience at Uber Eats and Axon, steps in as the interim CEO. This strategic shift towards superintelligence, which is similar to what others call artificial general intelligence (AGI), marks the latest direction for Zuckerberg, who had previously pivoted the company towards the metaverse in 2021, underscoring this by rebranding Facebook to Meta and investing heavily in virtual reality innovations.
In the wake of ChatGPT’s launch in 2022, igniting an AI race, major tech corporations have strategically aligned with innovative AI startups without outright acquisitions. Microsoft, for instance, incorporated key staff from startup Inflection AI to spearhead its AI division. Similarly, leadership from AI chatbot pioneer Character.AI transitioned to Google, while Amazon secured a deal with San Francisco’s Adept, acquiring their CEO, pivotal staff, and a license to their AI technology.
Scale AI, co-founded by Wang and Lucy Guo in 2016 when Wang was a student at MIT, gained support from the prestigious startup incubator Y Combinator. Wang, like Zuckerberg before him, left academia to pursue entrepreneurial ventures. Scale’s unique offering involved using human workforces to enhance AI predictions—specifically, they assisted in refining self-driving car technology by identifying various street objects. This service attracted collaborations with industry giants such as General Motors and Toyota.
The company aimed to elevate what Amazon’s Mechanical Turk had established by providing a customized service linking AI developers with freelance workers for AI enhancement projects. As AI large language models evolved, Scale adapted its offerings to meet the growing demands, applying its expertise to enhance systems from major players like Anthropic, OpenAI, Meta, and Microsoft. It’s unclear, however, how the new Meta partnership will influence Scale’s existing client relationships.
Wang has consistently forged ties with the U.S. government, securing contracts to deliver AI solutions to the Pentagon and maintaining associations with political figures and entities. Meta has also expanded its AI contributions to federal agencies, differentiating from competitors by open-sourcing its Llama system to encourage modification and utilization by a broader audience. Although Meta claims extensive usage of its AI systems, it has yet to unleash its most advanced system, the Llama 4 Behemoth.
Yann LeCun, Meta’s chief AI scientist and a respected figure in the AI field, has shown some skepticism about the strong reliance on large language models. He has highlighted the challenges in developing AI with a comprehensive understanding of the real world, memory retention, and advanced reasoning capabilities, which current models only superficially achieve. LeCun’s vision focuses on achieving and potentially surpassing human-level intelligence. During the recent VivaTech conference, he remained discreet on the Scale deal but affirmed Meta’s focus on extraordinary human-like AI capabilities.
Co-founder of Meta’s AI research arm, LeCun, alongside Rob Fergus, who briefly departed for Google but returned to oversee Meta’s AI research again, underscores the organization’s enduring commitment to long-term AI research. Fergus has articulated the goal as creating transformative human-level experiences that redefine human-technology interaction, reinforcing Meta’s dedication to groundbreaking AI development.