In early 2022, China launched the Eastern Data and Western Compute initiative, a cornerstone of President Xi Jinping’s long-standing ambition to establish Chinese dominance in artificial intelligence. The logic behind the EDWC is simple: AI relies on data centers, and data centers require massive amounts of power. While China’s population — and therefore most of its AI users — is concentrated in the east, the western provinces boast abundant energy resources and cooler climates that reduce cooling demands. Relocating computing facilities westward is a highly rational strategy for building power-reliant data centers in regions where electricity is easiest to generate.
To claim global AI leadership, however, Beijing must first unseat the reigning champion: the United States. Unfortunately, the American public remains largely oblivious to what military historian Niall Ferguson has dubbed “the most dangerous arms race in history.” The profound national security implications are too often ignored. Indeed, the war in Ukraine provides a real-world case study of AI’s centrality to modern defense. Kyiv’s ability to hold its own against a vastly larger Russian military is, fundamentally, an AI-driven achievement.
Rather than focusing on national security, the U.S. debate over data center infrastructure often devolves into debates about potential spikes in consumer electricity bills. While our primary adversary mobilizes its energy grid to secure an AI advantage, we risk conceding defeat by treating victory as too expensive. The irony is that this anti-data center backlash, curiously enough, is unique, among democracies, to the U.S. and is not actually grounded in real data.
Because electricity is a finite commodity, the expectation is that increased demand will drive up prices. While data centers will contribute to increased demand for electrical power in the future, estimates are that other drivers — everything from electric vehicles to heat pumps — will account for most of the overall increase in electricity demand. A recent study by the Electric Power Research Institute shows that electricity rates are decreasing in places where data centers are opening. Why is this?
First, it is important to understand that electricity is a highly regulated commodity, not a simple free market. Also, consumer costs are affected by both the cost of producing electricity — generation — and the cost of delivering it — transmission and distribution. The costs of maintaining the grid that moves power from a generating facility to an end user are largely fixed. By increasing the total number of users on the grid, the financial burden of these fixed costs is distributed more broadly, lowering rates for everyday consumers. In fact, the EPRI study found that this economies-of-scale effect caused average residential retail electricity prices to fall by around 4% when local data center capacity doubled.
The unprecedented scale of the new data centers coming online has sparked debates over whether these facilities are paying their “fair share.” While spreading existing fixed costs is potentially beneficial to everyday consumers, hyperscale operations often require dedicated infrastructure additions, including high-voltage transmission lines, upgraded substations, and new power-generation facilities.
The real worry is “cost shifting,” in which residential customers subsidize the multibillion-dollar grid upgrades needed by hyperscale developers. If a planned data center’s projected energy demand fails to materialize, consumers worry they could be left holding the bag for stranded infrastructure investments.
In many cases, however, companies must sign long-term agreements obligating them to pay most of their expected costs even if they end up using less energy than originally anticipated. The result is that energy companies will be able to invest confidently in critical infrastructure, knowing that returns are virtually guaranteed. Rather than viewing data centers as a drain on supply, it is more accurate to view them as a vital revenue source that makes necessary upgrades to the power system possible.
Even with grid upgrades, bringing new generating capacity online to meet this demand is essential. Whether relying on renewables or fossil fuels, there is no reason an energy-rich nation such as the U.S. cannot expand supply to meet demand. While there are constraints, ranging from transformer shortages to regulatory guidelines that slow construction, Adam Smith’s proverbial “invisible hand,” guided by data center demand, should incentivize overcoming these obstacles. There is some evidence that this is already happening with gas-powered plants. Again, the result in the long run should be lower prices. There is even emerging evidence, albeit from China, that AI itself can be deployed to optimize energy efficiency.
PAUL SRACIC: STOP CALLING IT THE CLOUD. IT’S A FACTORY
Beijing obviously doesn’t have to concern itself with local pushback when building its AI infrastructure. In the U.S., our right to question new development is at the heart of democratic self-government. But it’s also our responsibility to base arguments on data, not an instinctive distrust of tech giants. Arguably, the widespread fear of rising electricity bills is being deployed backward: rather than using it as a weapon to block data centers, we should harness that anxiety to support the expansion of power generation. Flooding the grid with new energy supply is the ultimate way to drive economic prosperity.
We are in a race for the defining technology of our lifetime. If the U.S. is serious about protecting its economic and national security, aggressively expanding energy production and modernizing the grid isn’t merely an option; it’s the essential first step.
Paul Sracic is a senior fellow at the Hudson Institute.