All insights

Platform economics beat labor arbitrage — margins fund flywheels that body shops cannot

Scale AI's 50%+ gross margins fund ML pre-labeling and workflow optimization, creating a flywheel; Indian BPOs at 10-15% margins cannot invest in R&D and remain trapped competing on price

@GowriShankarNag — #MarketMapMondays: The Label Paradox of AI, Antler India · · 5 connections

Scale AI operates at 50%+ gross margins. Those margins fund ML pre-labeling, workflow optimization, and quality infrastructure — creating a flywheel where better tech justifies premium pricing which funds more tech. Indian annotation BPOs operate at 10-15% margins, leaving no R&D budget, no platform investment, no differentiation — a cycle the article calls the “BPO gravity well.” Scale AI raised 67x more capital than iMerit despite being founded four years later with 6x fewer people.

This is Technology helps moat businesses but kills commodity businesses playing out in real time: in commodity annotation, productivity improvements flow to customers as lower prices; in platform businesses, the same improvements compound internally. The pattern reinforces Domain-specific skill libraries are the real agent moat, not core infrastructure — the defensibility ladder climbs from commodity labor to platform infrastructure to accumulated domain expertise. It also validates Sell the work, not the tool — model improvements compound for services, against software: Scale AI increasingly sells outcomes (structured training data), not hours.