Who Prices Cognitive Labor in the Age of Agents? A Position on Compute-Anchored Wages

A new economic framework argues that AI agents function as production technology converting compute capital into cognitive labor, rather than being labor themselves. This shift means wage equilibrium for human cognitive workers will be anchored to compute capital rental rates rather than traditional labor markets, with human wages bounded by a formula incorporating compute intensity and relative productivity.

This research presents a fundamental reframing of how AI agents impact labor economics, moving beyond the conventional wisdom that cheap-to-replicate agents simply collapse cognitive wages to zero. By reconceptualizing agents as capital rather than labor, the authors identify a crucial mechanism: the elastic supply constraint that determines wage equilibrium shifts from the labor market to the compute capital market. This distinction has profound implications for understanding AI's economic trajectory.

The Compute-Anchored Wage (CAW) framework provides a quantifiable relationship between human wages and underlying compute economics. Rather than agents directly competing with humans in labor supply, they compete indirectly through the compute infrastructure that enables them. This means wage trajectories depend heavily on compute capital pricing, rental rates, and efficiency improvements—metrics tied to hardware manufacturing, data center economics, and semiconductor supply chains rather than pure labor supply dynamics.

For investors and developers, this reframes where economic value concentrates. Instead of human cognitive workers facing unlimited downward wage pressure, competitive pressure flows through compute infrastructure costs. Organizations adopting AI agents need cost-effective compute access to achieve competitive wages relative to human labor. This creates secondary investment opportunities in compute provision and optimization, while also suggesting that compute capital pricing becomes the critical variable shaping labor market equilibrium.

The framework's treatment of complementary versus substitutable tasks introduces additional complexity. Not all cognitive work faces identical wage pressure; tasks where agents and humans complement each other may see different dynamics than pure substitution scenarios. This suggests sectoral differentiation in AI's labor market impact rather than uniform cognitive wage compression.

• →AI agents function as production technology converting compute capital into cognitive labor, not as labor supply themselves.
• →Human cognitive wages are theoretically bounded by compute rental rates, compute intensity per agent unit, and relative human-to-agent productivity.
• →Wage equilibrium for cognitive labor shifts from traditional labor markets to compute capital markets as the elastic supply margin.
• →Complementary and substitutable tasks experience different wage dynamics, inverting assumptions about skill-biased technical change.
• →Compute infrastructure pricing becomes the primary price-setter for cognitive labor rather than labor market competition.