OpenAI Confirms Ongoing Collaboration with Scale AI Following Meta’s Stake Acquisition
OpenAI will continue its collaboration with Scale AI despite Meta’s $14.8 billion acquisition of a 49% stake in the AI data labelling company, according to OpenAI CFO Sarah Friar at the VivaTech 2025 conference in Paris.
“We don’t want to ice the ecosystem because acquisitions are going to happen,” Friar said. “And if we ice each other out, I think we’re actually going to slow the pace of innovation.”
What Is Scale AI, and Why It Matters to OpenAI
Scale AI specialises in providing curated, labelled training data, which is critical for training large language models (LLMs) like OpenAI’s ChatGPT and Meta’s Llama models. The firm plays a vital role in developing next-gen generative AI technologies by ensuring model accuracy through expertly labelled data.
Despite Scale AI’s leadership now aligning with Meta—CEO Alexandr Wang will head Meta’s superintelligence unit—OpenAI remains committed to a multi-vendor strategy for training data procurement.
“We don’t just buy from Scale,” Friar added. “We work with many vendors on the data front.”
Specialized Human Expertise Now Essential for AI Model Training
As large language models become more advanced, their development demands expert-level training data. OpenAI is increasingly working with historians, scientists, and PhDs to train AI models in niche, high-impact domains.
“As models have gotten smarter, you’re going into a place where you need real expertise… Academics and experts are finding novel things in their space through collaboration,” said Friar.
Key Implications for the AI Industry
- OpenAI won’t restrict collaborations based on competitive stakes.
- Meta’s Scale AI deal won’t block OpenAI’s access to premium training data.
- Collaborative, cross-vendor ecosystems are critical to AI’s growth.
- Human expertise is becoming central to next-gen model development.
