When Meta dropped $14 billion for a 49% stake in Scale AI and poached its founder Alexandr Wang, the conventional wisdom was that Scale would be relegated to a mere Meta supplier. But nearly a year later, CEO Jason Droege says the company is on track to surpass $1 billion in revenue this year — and it's doing so by shifting away from the data labeling that made it famous.
Scale's core strategy is pivoting hard toward building internal AI applications for government and enterprise clients. The new business unit is already pulling in $200 million in annualized revenue, and Droege predicts it will overtake data labeling as Scale's biggest moneymaker within 18 months.
The Meta deal locked in a five-year contract worth at least $450 million annually, but it came at a cost: OpenAI, Scale's original marquee customer, walked away. Competitors say Scale has been sidelined in recent data labeling bids due to Meta's involvement. Droege insists the accelerated pivot isn't about client rejection — it's that the data labeling market is slowing down, while the demand for enterprise AI services is just taking off.
Whatever the motivation, the strategy is yielding results. On the commercial side, Scale signed Ernst & Young and the Mayo Clinic. On the government side, it landed a $500 million contract with the Department of Defense last week — Project Thunderforge — to integrate AI agents into military mission planning. And last month, Scale joined Palantir and others in the Golden Dome missile defense program, a project with a total budget of $185 billion.
Investor Coatue argues that Scale's unique combination of large-model training experience and government-enterprise delivery capability is what will sustain its lofty valuation and eventual IPO.