EU AI Act Boosts Micro1's Expert Labeling
micro1
This turns local knowledge into billable infrastructure. Under the EU AI Act and related EU implementation guidance, AI developers face growing pressure to show that systems work across Europe’s cultural and linguistic diversity, not just in English or a generic Western data set. That pushes labeling work away from anonymous crowd labor and toward curated pools of native speakers and local experts, which fits micro1’s model of matching labs with pre vetted specialists across countries and charging more for screened, compliance ready workflows.
-
The practical shift is from cheap tagging to expert judgment. A model team may need a Polish lawyer, a German clinician, or a French native speaker to judge whether an answer is safe, biased, or culturally off. That is harder to source, slower to manage, and easier to price at a premium than generic RLHF work.
-
The closest comps are Mercor and Handshake, which also sell pre vetted expert networks into AI post training. But Scale grew up around broad labeling workflows and software, while micro1 is positioned more like a specialist marketplace for hard to reach experts, which matters more as buyers care about credentials and geography.
-
The revenue signal is that buyers are already paying up for expert supply. micro1 is estimated at $100M revenue by November 2025, Mercor at $500M by August 2025, and Handshake reached $280M by August 2025 after adding AI labeling. The common pattern is that scarce expert pools are becoming a core input to model quality.
The next leg of the market is likely to split cleanly in two. Commodity labeling will keep getting automated, while regulated and high consequence AI work will flow to networks that can prove who the annotators are, where they are, and why their judgment matches the market being served. That is the lane where micro1 can keep expanding margins as Europe and other regions tighten AI rules.