Shift Toward Neutral Human-Data Vendors

The Meta and Scale tie up turned neutrality into a product feature. Once one of the biggest human data vendors became aligned with a frontier lab, other labs had a stronger reason to buy from suppliers that could act as an outside referee, source participant groups they did not already control, and provide second opinion evaluation on safety, red teaming, and model quality. Prolific fits that need because it combines self serve access, API workflows, and a large pre profiled participant base.1 2 3 4 5

• Prolific says customers stepped back from Scale after the Meta transaction over independence concerns, which pushed work back into the market. It also says the beneficiaries were not just Prolific, but other providers like Surge and Handshake, showing this was a category shift toward neutral vendors, not a one company event.1 2
• The practical reason neutrality matters is that labs already have internal annotator pools. They still buy from outside vendors when they need validation, broader geography, more languages, or participant types that are missing internally. That makes an independent supplier useful even when a lab can build some capacity itself.1 4
• This reshuffle happened inside a much larger boom. By April 2026, estimated annualized revenue reached about $350M at Prolific, $1.1B at Handshake, $1B at Mercor, $1.2B at Surge, and $1.5B at Scale. In a market this large, a loss of trust at one major vendor can redirect substantial spend very quickly.2 3 5

Going forward, neutrality should become more valuable, not less. As AI labs buy more human input for evaluation, trust and safety, and regulatory validation, the winning suppliers will look less like raw labor brokers and more like independent measurement and workflow layers sitting between model builders and the humans judging their systems.1 2 4