Will AI Kill the Firm? (www.project-syndicate.org)

The article argues that agentic AI — systems that can coordinate, decide and execute — is eroding the oldest rationale for the modern firm: reducing transaction costs and supervising specialized investments (Coase, Williamson). Historically firms solved information scarcity and coordination through hierarchy; now cloud platforms, smart contracts and generative models can design, enforce and act on agreements autonomously. AI is collapsing the exploration/exploitation split (e.g., drug discovery algorithms that both invent molecules and model scalable production; code-generation systems that write, test and deploy in a loop; real-time marketing optimizers), producing an “AI surplus” of ideas and initiatives that outpaces firms’ absorptive capacity (Cohen & Levinthal). The result is internal overload and coordination entropy rather than smooth scaling. For AI/ML practitioners and organizations this matters technically and strategically: value is shifting from centralized command to continuous sensing–learning–acting loops, so governance must move decision rights to where insights emerge. Practically this implies investing in shared data and incentive infrastructures, programmatic enforcement (smart contracts), transparent flows, and decentralized authority models (partner-like autonomy, networked federations). Failure to rewire governance risks Kodak/Nokia-style paralysis; firms that orchestrate-and-empower rather than command-and-control can convert the AI surplus into sustained advantage.