Data Marketplaces & Intermediaries

Scale AI loses ground as the training-data oligopoly reshuffles

Source: Pebblous · Sep 12, 2026

Scale AI's $14.3 billion deal selling 49% of itself to Meta was one of the largest transactions in the training-data space to date, and it was widely read as a validation of Scale's incumbent position. The aftermath tells a more complicated story: Scale has reportedly lost major customers to competitors since the deal closed, as the market reshuffles around Surge AI, Mercor, and Handshake AI as the names now picking up share.

The mechanism is straightforward and worth internalizing: once a data vendor takes a strategic investment from one large AI lab, competing labs have an obvious reason to worry about neutrality and data handling, and to diversify away from that vendor rather than concentrate more spend with it. A deal structured to look like a strength — deep alignment with one of the largest buyers in the category — can simultaneously become a reason for every other buyer to look elsewhere.

That's directly relevant to how we think about our own customer concentration and positioning. The market isn't converging on a single winner; it's reshuffling between several well-capitalized players, and vendor-neutrality (or the perception of it) appears to be a real factor in who wins share, not just price or data quality.

Key Points

• Scale AI sold 49% of itself to Meta for $14.3B
• Since that deal, Scale has reportedly lost major customers to competitors
• Business is reshuffling toward Surge AI, Mercor, and Handshake AI as the new leading names
• A useful case study in incumbent vulnerability even after a large strategic investment