Meta's reported move toward Scale AI is not just another funding headline. It is a bet that the next phase of AI competition may depend as much on data strategy and leadership as on model releases.
The company is reportedly investing nearly $15 billion in Scale AI, taking a 49% stake in the data-labeling startup and bringing CEO Alexandr Wang into Meta to help lead a new "superintelligence" lab. The scale of the deal has revived an old question around Mark Zuckerberg's biggest moves: is Meta seeing something early, or trying to close a gap under pressure?
A familiar kind of risk for Meta
The reported Scale AI deal echoes some of Meta's earlier large and controversial bets. The company previously spent $19 billion on WhatsApp and $1 billion on Instagram, both deals that drew skepticism when they closed.
Those acquisitions later became central parts of CEO Mark Zuckerberg's business. That history matters because it frames the current debate. Meta has shown a willingness to pay heavily for assets it believes can become strategically important, even when investors and founders question the price at the time.
Scale AI is a different kind of target. It is not an emerging social platform with a visible consumer network. Its value sits closer to the engine room of artificial intelligence: the data used to train high-performing AI models.
Why Scale AI matters to model building
For several years, leading AI labs including OpenAI have used Scale AI to produce and label training data. The company and its data annotation competitors have also been hiring more specialized workers, including PhD scientists and senior software engineers, to create high-quality data for frontier AI labs.
That shift points to a larger reality in AI development. Data annotation is no longer only a low-level support function. As models become more advanced, the quality, design and specialization of training data can affect how useful those models become.
Meta may see a close relationship with Scale AI as a way to improve an area where it has faced internal concern. According to a person familiar with the matter, Meta's leaders have complained about a lack of innovation around data in the company's leading AI teams.
The timing also adds pressure. Earlier this year, Meta's generative AI unit launched Llama 4, a family of AI models that did not match the capabilities of models from Chinese AI lab DeepSeek and was widely viewed as a disappointment.
Alexandr Wang becomes part of the reset
Meta's reported plan is not only about Scale AI as a company. It is also about Alexandr Wang. The 28-year-old CEO is known in Silicon Valley as ambitious, well-connected and a strong salesman, and he has spent the past few months meeting with world leaders to discuss AI's impact on society.
That background could make him valuable inside a company trying to sharpen its AI direction. Wang has built a prominent startup serving frontier AI labs, and Meta is looking for ways to regain momentum in a market where OpenAI, Google and Anthropic remain major rivals.
There is also a clear caveat. Wang has not previously led an AI lab of this kind, and he does not have the same AI research background as leaders such as Safe Superintelligence's Ilya Sutskever or Mistral's Arthur Mensch.
That may explain why Meta is also said to be recruiting high-profile talent such as DeepMind's Jack Rae for the new AI research group. The company is trying to address not only a data challenge, but also a people challenge. According to data compiled by SignalFire, Meta lost 4.3% of its top talent to AI labs in 2024.
The neutrality problem for Scale AI
The future of Scale AI after the reported deal is less straightforward. The market for AI training data is changing. Some AI labs have brought data collection inside their own organizations, while others have increased their use of synthetic data, meaning data generated by AI.
Scale AI has also faced business questions. In April, The Information reported that Scale AI had missed some of its revenue targets.
Robert Nishihara, co-founder of Anyscale, told TechCrunch that frontier AI labs are exploring new and compute-intensive ways to use and optimize data. His summary was blunt: "Data is a moving target," Nishihara told TechCrunch in an interview. "It's not just a finite effort to catch up — you have to innovate."
A closer relationship between Meta and Wang could also make some Scale AI customers uncomfortable. AI labs that compete with Meta may prefer data partners that appear more neutral.
That possibility could help competitors such as Turing, Surge AI and newer data providers such as LM Arena. Turing CEO Jonathan Siddharth told TechCrunch by email that customer interest increased after rumors around Meta's deal with Scale AI. "I think there will be some clients who will prefer to work with a partner that's more neutral," he said.
What this means for Meta's AI race
The reported Scale AI investment gives Meta a possible path to improve one of the core inputs behind advanced AI systems. It could bring Meta closer to a major data supplier, add Wang to a new superintelligence effort and signal that the company is willing to spend heavily to compete.
But the risks are just as visible. Meta still has to prove that better data strategy can translate into stronger AI models. It also has to solve talent challenges, build a credible new lab and manage the consequences of tying Scale AI more closely to one of the industry's largest AI competitors.
The competitive backdrop will not wait. OpenAI is preparing the release of its next flagship model, GPT-5, as well as its first openly available model in years, which will compete with Meta's current and future Llama releases.
Only time will show whether this becomes another prescient Meta bet or an expensive attempt to catch up. For now, the reported deal makes one point clear: Meta believes the fight over AI will be shaped not only by models, but by the data and people behind them.