Today’s biggest thread isn’t a new model release—it’s the bill coming due. Apollo Global Management says that outside the biggest technology firms, most companies still aren’t seeing measurable financial benefits from AI. In Apollo’s analysis of the S&P 493, revenue per worker has hovered at about $550,000 since 2019, a sign that productivity gains are either absent, delayed, or captured elsewhere. The implication is brutally practical: mainstream corporate AI spending is getting reframed from “growth” to “cost center,” and more CFOs are demanding clear returns before scaling experiments into full deployments.
This is a familiar pattern in enterprise tech, but the details matter. Apollo’s framing doesn’t claim AI doesn’t work; it claims the value chain is lopsided. The largest AI-linked firms can show better AI-linked outcomes, while the broader corporate base can’t yet connect AI usage to revenue per worker in a way that survives boardroom scrutiny. That mismatch is now shaping budgets, forcing tighter questions around which use cases actually move metrics—and which ones merely generate demos. In other words, AI’s most visible progress today may be the accounting.
