Meta’s Rare Big Bet on Scale AI – A New Chapter in the AI Arms Race

Meta rarely writes billion-dollar checks to outside AI startups — but this one might be an exception.

According to Bloomberg, Meta is reportedly in talks to invest several billion dollars into Scale AI, a San Francisco-based startup that provides high-quality training data for AI models. If finalized, this could become Meta’s largest-ever external AI investment — a rare move for a company known more for in-house development than big outside bets.

So, why Scale AI?

Founded in 2016 by Alexandr Wang and Lucy Guo, Scale AI built its name in the trenches of data labeling — curating massive, high-quality datasets across images, text, and other formats that are essential for training AI systems. Over time, it has evolved into a key player in AI data infrastructure, serving clients like OpenAI, Microsoft, Nvidia, the U.S. Department of Defense — and yes, even Meta.

Scale's 2023 revenue hit $870 million, and it's projected to more than double that to $2 billion this year — a serious growth curve. The company closed a $1 billion Series F round just last month, bringing its valuation to $13.8 billion. But rumor has it they’re now eyeing a much higher valuation — potentially up to $25 billion in secondary deals.

It’s not just about business either. Meta and Scale are already working together on a project called Defense Llama, exploring large language models for military use. So this isn’t a cold-start investment — it’s an expansion of a growing partnership.

Meta joins the global AI gold rush

Meta’s potential investment follows a broader trend: tech giants aren’t just building AI — they’re stockpiling it. Microsoft’s $13 billion into OpenAI set the tone, and Amazon and Google have backed Anthropic with multi-billion dollar deals. Everyone’s scrambling to secure their place in the AI supply chain, from compute to data.

What sets this apart is Meta’s usual playbook: a strong preference for internal R&D and open-source efforts like the Llama models. But as competition heats up, data is emerging as a bottleneck — and that’s where Scale AI fits in. In AI, better data often beats bigger models.

Mark Zuckerberg already signaled Meta’s massive AI push earlier this year, pledging up to $65 billion in AI investments for 2025. With Llama 4 on the way and a goal of reaching over a billion users with its AI assistant, Meta clearly wants to go head-to-head with the likes of OpenAI and Google DeepMind. Getting deeper into Scale AI could give it a competitive edge.

Meta stock surges on AI optimism

Investors seem to like what they’re seeing. Meta's stock has jumped 18% over the past month, fueled by growing excitement around its AI roadmap. Just last Friday, the stock climbed another 1.91%, closing at $697.71 — a fresh show of confidence from the market. If this Scale AI deal goes through, it could further reinforce the bullish AI narrative surrounding Meta.

AI investment is booming

Stanford’s 2025 AI Index Report, released recently, confirmed what many in the space are seeing: AI adoption and investment have hit new highs. Private AI investment rebounded sharply in 2024, reaching a record $252.3 billion globally, with the U.S. alone attracting $109.1 billion — 12 times that of China.

In this context, Meta’s interest in Scale AI feels less like a surprise, and more like a necessary move in a race where falling behind isn’t an option.

My take

This potential deal signals that Meta is finally opening its wallet in a way we haven’t seen before — and it's doing so strategically. With OpenAI and Anthropic already tied to Microsoft, Amazon, and Google, Meta’s options for scaling its AI infrastructure through partnerships are limited. Scale AI offers a rare combination of operational scale, strategic alignment, and political access (thanks to its ties to the Pentagon).

If this deal goes through, it would mark a watershed moment for Meta — a rare billion-dollar bet on an external partner. It also underscores how critical data infrastructure has become in the AI gold rush. For investors, the question is whether Meta’s bet will pay off in AI dominance or serve as a cautionary tale of overpaying for hype.

It also underscores an important trend: AI’s future won’t just be decided by model size or compute power — but by who controls the data pipelines.

$META