Hajra Naz
Meta is making a bold move in AI by investing nearly $15 billion in Scale AI, a leading data-labeling startup. The tech giant is also taking a 49% stake in the company and bringing in Scale AI CEO Alexandr Wang to co-lead a new “superintelligence lab” within Meta. This major partnership signals Meta’s effort to catch up in the rapidly evolving AI race.
The move reminds many of Meta’s past high-stakes investments—like its $19 billion acquisition of WhatsApp and $1 billion for Instagram. Both were seen as overpriced at the time but ultimately turned into core pieces of Meta’s empire. The big question now is whether the Scale AI deal will follow a similar path or end up as a costly mistake.
Instead of buying a social app this time, Meta is betting on something foundational: the data that trains AI models. Scale AI has been a key player in the AI world, helping labs like OpenAI and Anthropic label and structure the high-quality data used to build their models. Recently, these companies have been hiring top talent—PhDs, machine learning experts, and senior engineers—to create better data faster and more efficiently.
Meta’s internal teams have reportedly struggled with innovation around data, and the company sees this partnership as a way to solve that. Earlier this year, Meta launched its Llama 4 model, which failed to match up to competitors like China’s DeepSeek, drawing criticism and adding pressure to improve. Meanwhile, Meta has seen a talent drain: 4.3% of its top AI talent left in 2024 for other AI firms, according to SignalFire.
READ MORE: Is DeepSeek better than ChatGpt?
Now, with Wang joining Meta’s AI leadership, the company hopes to turn the tide. At just 28 years old, Wang has gained a strong reputation in Silicon Valley as a visionary founder and skilled communicator.
He’s been meeting with global leaders about the societal impact of AI—but it’s worth noting that Wang lacks a deep AI research background compared to leaders like Ilya Sutskever (Safe Superintelligence) or Arthur Mensch (Mistral). That’s why Meta is also reportedly hiring seasoned researchers, including Jack Rae from DeepMind, to support the new AI lab.
As for Scale AI’s future, it’s unclear how the company will operate after this deal. Some AI labs are now generating synthetic data in-house, while others are experimenting with new data collection methods. Scale AI reportedly missed revenue targets earlier this year, casting doubt on its current growth.
Other companies are watching closely. Some AI labs might now avoid working with Scale AI, seeing it as too closely tied to Meta. This could give an edge to competitors like Turing, Surge AI, and even LM Arena, a new player in the data space. Turing CEO Jonathan Siddharth said interest from new customers has already increased, likely due to the news of Meta’s involvement.
Experts like Robert Nishihara, co-founder of Anyscale, say the race to improve AI data is far from over. He said.
“Data is a moving target, It’s not just about catching up—you have to innovate constantly.”
Meta has a long way to go to match leaders like OpenAI, which is preparing to launch GPT-5 and a new public model to challenge Meta’s current and future Llama models. Whether this partnership with Scale AI and Wang gives Meta the edge it needs—or turns into another tech misstep—remains to be seen. One thing is certain: the AI race is only speeding up.
