its ability to subsidize RLHF pricing

This is the core cost structure advantage that large, mixed workload vendors have over specialist RLHF shops. Scale sells cheap, repeatable annotation at huge volume, then layers higher touch RLHF and eval work on top, which lets it accept lower prices on premium projects without losing the overall account. Micro1, by contrast, is built around scarce experts, fast matching, and managed workflows, so it has less room to win on price and more need to win on quality and speed.

• Scale is not just an RLHF vendor. It bundles data labeling with self serve software, data management, model validation, deployment tooling, and usage based plans with volume discounts. That makes RLHF one line item inside a broader platform sale, not the whole economics of the customer relationship.
  2 sacra 3 sacra
• Micro1 monetizes mainly by marking up expert labor plus workflow, QA, compliance, and payroll. Its product is strongest when a lab needs a doctor, lawyer, physicist, or senior engineer in under 48 hours, not when a customer is optimizing pennies on large generic labeling volumes.
  1 sacra 4 sacra
• The market is also splitting. Commodity crowdwork is getting cheaper, while demand is moving toward credentialed experts, cultural nuance, red teaming, and trust and safety review. That shift helps specialists, but it does not remove pricing pressure because buyers still compare every premium task against lower cost adjacent options.
  4 sacra 5 sacra 6 sacra

Going forward, the winners in human data will look less like labor brokers and more like control layers over different kinds of human input. Scale is pushing outward from labeling into full stack AI infrastructure, while Micro1 is pushing upward into harder expert workflows and adjacent products like recruiting and robotics data. That should widen the gap between broad platforms that win bundles and specialists that win the hardest tasks.