Op-Ed: The Fatal Conceit Gets a GPU Cluster
- Jeffrey Depp
- 19 hours ago
- 3 min read
Committee for Justice Senior Fellow Jeffrey Depp has published a new article in Truth on the Market examining Senator Bernie Sanders’ proposed American A.I. Sovereign Wealth Fund Act and its implications for innovation, private property, and the future of artificial intelligence.
The article, “The Fatal Conceit Gets a GPU Cluster: Bernie Sanders’ Plan to Socialize AI,” is the latest in a series of essays exploring AI policy through the complementary lenses of Austrian economics and Public Choice theory. While each article addresses a different policy proposal, they share a common theme: the belief that markets—not centralized political institutions—are best equipped to guide the development of emerging technologies.
AI and the Limits of Central Planning
The Sanders proposal begins from a simple premise: because modern AI systems are trained on humanity’s collective knowledge, the public should receive an ownership stake in the largest AI companies through a federally managed sovereign wealth fund. Under the bill, qualifying AI companies would surrender 50 percent of their equity to a government-controlled investment fund managed by a politically appointed commission.
The article argues that this proposal rests on a fundamental misunderstanding of how wealth is created.
Drawing on the work of Ludwig von Mises, F.A. Hayek, and Israel Kirzner, the article explains that economic value does not arise simply because information exists. Rather, value is created through entrepreneurial discovery—the process by which innovators recognize opportunities, combine resources in new ways, and develop products that consumers voluntarily choose to purchase.
Artificial intelligence is no exception. AI models do not become valuable merely because vast quantities of data are available. Their value arises from the creative and highly uncertain process of transforming that information into useful technologies. Attempting to socialize those entrepreneurial returns mistakes the raw material of innovation for the innovation itself.
The Public Choice Problem
The article also examines the proposal through the lens of Public Choice economics.
Supporters of the legislation envision an independent commission that would manage the government’s ownership stake in AI companies and exercise corporate voting rights “in the public interest.” The article argues that this vision overlooks a central insight developed by James Buchanan and Gordon Tullock: government officials respond to incentives just as market participants do.
Rather than representing some objective conception of the public good, a commission charged with directing AI development would inevitably face political pressure from organized interest groups, competing constituencies, and incumbent firms seeking to shape regulation in their favor. The result is likely to be greater rent-seeking, regulatory capture, and bureaucratic expansion—not more democratic innovation.
A Continuing Conversation
This article builds on several recent publications examining AI governance from an Austrian and Public Choice perspective.
Earlier essays argued that markets—not government mandates—are better suited to coordinate the enormous investments required to power AI infrastructure, including electricity generation, transmission, and workforce development. Others explored Hayek’s “knowledge problem” and questioned whether policymakers can realistically predict the future trajectory of rapidly evolving AI technologies.
Taken together, these pieces advance a common argument. Markets are not simply efficient mechanisms for allocating known resources. They are dynamic processes for discovering knowledge that does not yet exist. Consumers, entrepreneurs, and investors continuously generate information through competition, experimentation, and voluntary exchange. Government planning, by contrast, necessarily relies on assumptions about technologies, risks, and opportunities that remain unknowable until the discovery process unfolds.
This distinction carries profound implications for AI policy. The central question is not whether artificial intelligence should benefit society—it undoubtedly should. The question is whether that objective is more likely to be achieved through decentralized market discovery or centralized political direction.
