The surge of investment in artificial intelligence infrastructure has triggered a backlash in the United States, resulting in the cancellation of at least $85 billion worth of data center projects over the past three years. In July, New York became the first state to impose a temporary moratorium on new data centers, with other states and the federal government considering similar measures. While concerns about the economic and environmental costs of these facilities are valid, the bigger question remains: Is the U.S. missing out on a vital opportunity by halting the development of more data centers?
According to Josh Zoffer, a former official with the National Economic Council under President Joe Biden and now an AI investor, data centers represent a critical test of American industrial resolve. This situation underscores a troubling reality: developing strategic capabilities often requires sacrifices in the present. The failure to recognize this truth has historically led policymakers to disrupt similar infrastructure projects. Two noteworthy examples illustrate this point.
In the 1970s, the global oil crisis forced the U.S. to acknowledge its strategic vulnerability due to energy dependency. In a 1977 speech, President Jimmy Carter warned that establishing a secure and diversified domestic energy base would require Americans to endure a period of “higher costs” and “greater discomfort.” However, Carter’s honesty did not yield the desired outcomes.
Economist Jeffrey Currie argues that Carter’s successors took away a different lesson. They shied away from recognizing resource scarcity and instead responded to supply shocks by easing price tensions and dipping into strategic reserves. While this approach bought time, it delayed necessary investments to achieve true energy security.
Current tensions in the Strait of Hormuz serve as a reminder of the energy transition Carter advocated. Although the massive shale gas reserves discovered since then have reduced U.S. dependency significantly, vulnerability to oil shocks persists, as highlighted by rising gasoline prices.
In contrast, China has spent decades pursuing electrification, which Jeffrey Currie describes as “buying the ability to choose.” The resulting flexibility—where electricity can be generated from coal, gas, solar, wind, or uranium—has shielded China from the current energy crisis.
Furthermore, the U.S. has also squandered the chance to avoid reliance on China for rare earth minerals. In the latter half of the 20th century, the Mountain Pass mine in California was a cornerstone of local rare earth mineral production, ensuring a stable and self-sufficient supply chain. However, by the late 1990s, operations at the mine were drastically reduced due to environmental concerns, regulatory changes, and competition from cheaper Chinese producers, leaving U.S. industry dependent on imports for over 90% of its rare earth needs, according to the U.S. Geological Survey.
These examples highlight a significant weakness in the U.S. decision-making process. Markets efficiently determine the cost of taking action but are inefficient at assessing the cost of inaction. As a result, when social returns are involved—such as in broad investments to secure energy and mineral supplies—markets often fail.
To successfully guide, coordinate, and finance investments across multiple markets, Mariana Mazzucato and Dani Rodrik advocate for public-private partnerships in their research, “Conditional Industrial Policy: Classification and Exemplars.” They argue that conditionality is crucial for ensuring success: governments must create incentives to steer private corporate decision-making toward outcomes they would not pursue otherwise.
Such arrangements can be structured so that the government provides private companies with one or more advantages—such as loans, tax incentives, or expedited licensing—in exchange for conditions related to guidance and risk/reward sharing. For example, tech giants in the AI sector could be required to address local concerns by absorbing environmental and social costs, like sharing technical data or committing to joint investments in grid expansion. The risk/reward aspect would entail the public sector sharing in some of the returns if it also assumes part of the risk, potentially through profit-sharing mechanisms or gaining access to computing power.
Compared to recent history, policymakers today are better positioned to recognize the strategic importance of data centers. The global economic system has shifted, highlighting the value of this infrastructure. Following World War II, the focus was on rebuilding collaboratively, with integration as the goal and globalization as the means. Now, economic geography dominates a more fragmented landscape, with power stemming from control over strategic points.
This shift suggests that policymakers are increasingly inclined to view strategic investments through a multi-market lens. In this context, the strategic importance of data centers becomes less ambiguous when compared to earlier energy or rare earth investment considerations. Jensen Huang, CEO of Nvidia, has described AI as a five-layer cake, with data centers forming the foundational layer, further emphasizing their role in the overall economy.
However, the challenge of directing investment toward strategic industries is not unique to the United States. In a 2024 report on European competitiveness, former European Central Bank President Mario Draghi acknowledged that revolutionary innovation, defense industrial capabilities, and cross-border energy networks constitute public goods that “would not be sufficiently available in the absence of joint measures.”
The issue of data center construction should evolve into a bipartisan priority. The success of public-private partnerships will hinge on their design; overly stringent conditions could stifle innovation, while too lenient terms might allow private players to monopolize benefits without contributing fairly.
