Big tech companies investing in AI startups is not new. The likes of Microsoft, Google, and Amazon have been engaging in the practice for a while now. Meta has also shown some activity in that space, but not as much as its competitors.
Earlier this year, Reuters reported that the company plans to invest up to $65 billion in AI-related projects.
It appears that $10 billion of this amount would be invested in Scale AI, an AI startup that helps enterprises develop their own models or apply foundation models to their business. They do that through data labelling and reinforcement learning with human feedback (RLHF).
To Catch Up With Microsoft and Google
Until now, Meta’s strategy revolved around open source leadership with Llama, Meta AI, and other research efforts. But the tectonic plates are shifting.
Microsoft’s $13 billion bet on OpenAI, Amazon’s $8 billion investment in Anthropic, and Google’s investments in Anthropic have effectively locked down much of the top-tier model infrastructure within tightly integrated ecosystems. Big tech companies are partnering with the best out there to make sure they leverage the technologies to their advantage.
Scale AI is one of the major players in the data collection and labelling market. It already works with tech giants like Microsoft and OpenAI. As per reports, the company was recently valued at approximately $14 billion in a 2024 funding round supported by Meta and Microsoft and was reportedly in discussions earlier this year for a tender offer that would value it at $25 billion.
As per the reports, Meta’s significant external AI investment in Scale AI is noted as the company’s largest to date. This move is unusual for the Mark Zuckerberg-led company, which typically relies on internal research and a more open development approach to advance its AI capabilities.
Unlike its competitors, Meta does not offer cloud computing. Hence, the investment in Scale AI could help the case. However, the company has yet to clarify this.
Better Rapport With The US Government
Beyond the numbers, there can also be optics at play. For a company constantly under regulatory scrutiny, Meta’s partnership with Scale AI, a company that regularly testifies before the US Congress and holds DoD contracts, adds a layer of institutional credibility.
The Defense Llama project, a version of Meta’s model tailored for US national security use, has already laid the groundwork for this purpose. It is a customised large language model based on Meta’s Llama 3.
Scale AI mentioned that it was built within Scale’s Donovan platform and deployed exclusively in secure government environments. It supports military and intelligence use cases such as operational planning and adversary analysis.
The model was trained using Scale’s data engine and fine-tuned on military doctrine, international law, and DoD ethical guidelines to provide responses tailored to national security professionals, including alignment with ODNI’s writing style.
Growing Market for Data Collection and Data Labelling
According to an SNS Insider report, the global data collection and labelling market was valued at $3.0 billion in 2023 and is projected to reach $29.2 billion by 2032, growing at a CAGR of 28.54%. This rapid growth is being driven by the rising adoption of AI and machine learning across industries, which require vast volumes of high-quality labelled data.
With applications ranging from speech recognition and image classification to autonomous decision-making, companies are increasingly investing in automated and human-in-the-loop annotation processes. Ethical AI development, reduced algorithmic bias, and demand for diverse datasets are also shaping innovation in this field.
The report states that the interest in labelled text data is gaining traction for NLP-driven tools like chatbots and sentiment analysis, making it the fastest-growing segment. “Some big-money companies, like Google, Microsoft, and Amazon, are spending a lot on labelled datasets for improving AI-backed searches, voice-assistants, or other modern recommendation systems,” read the report.
Moving on From Llama 4 and More
Llama 4 launched with a couple of complaints and controversies surrounding the benchmarks. While the company denied any wrongdoing, this did not give it a good public image.
The company’s interest in scaling things up with an investment in Scale AI should allow them to do better, faster, and improve the state of their AI offerings, whether for enterprises or consumers.
The post Why Meta is Investing in Scale AI appeared first on Analytics India Magazine.
