AI outsourcing transaction costs are the search, verification, monitoring, and trust costs that determine whether AI makes firms smaller or larger. Noah Smith's framing starts from Coase/Williamson transaction cost theory: companies exist because some work is cheaper to coordinate inside a firm than through arm's-length market contracts.source: noah-smith-ai-outsourcing-transaction-costs-2026.md
The pro-outsourcing case is straightforward. The internet already lowered transaction costs by making contractors easier to find, evaluate, communicate with, and monitor; AI agents could push that further by helping one person size markets, code, price, market, sell, integrate tools, and run more functions without hiring. That would reinforce solopreneurship and lower headcount at AI-native firms.source: noah-smith-ai-outsourcing-transaction-costs-2026.md
Smith's counterpoint is that AI also manufactures information. If agents can cheaply create fake companies, fake websites, fake reviews, and shifting personas, then procurement and partner selection may require expensive verification. Even without outright fraud, today's AI agents may be too ephemeral for long-term business relationships unless their reliability, goals, and behavior can be reestablished over time.source: noah-smith-ai-outsourcing-transaction-costs-2026.md
This creates a bifurcation thesis for the ai-labor-market: some low-trust or short-horizon businesses may become one-person companies with AI agent "crews," while high-trust, long-horizon processes may move inside large firms that can verify agents and counterparties through internal procedures. In that world, organizational-moats include trusted internal coordination systems, not just talent density or speed.source: noah-smith-ai-outsourcing-transaction-costs-2026.md
The argument also rhymes with japanese-corporate-diversification: if external contracting becomes less reliable, broad internal task bundles and salaryman-like roles can look less like historical weirdness and more like a response to high coordination and trust costs.source: noah-smith-ai-outsourcing-transaction-costs-2026.md
Open question: will AI verification tools, identity systems, audits, reputation markets, or cryptographic provenance lower these costs fast enough to keep outsourcing cheap? If not, AI may simultaneously produce more solopreneurs and more monster firms, hollowing out the middle rather than simply shrinking every company.source: noah-smith-ai-outsourcing-transaction-costs-2026.md
Related pages: ai-labor-market, organizational-moats, japanese-corporate-diversification, decentralized-ai-compute, production-llm-reliability, noah-smith.