The Bitcoin Policy Institute, a Washington, D.C.-based think tank, has proposed a model that would distribute "data center dividends" to households in communities hosting artificial intelligence infrastructure. The report aims to address growing opposition to data centers by returning a portion of existing property tax revenue directly to residents, potentially providing annual payments between $4,500 and $8,900 per household without imposing new taxes on developers.

Sam Lyman, head of research at the nonprofit, stated that the proposal is designed to ensure that Americans who build the county infrastructure for these facilities benefit from the AI boom, rather than just developers in Silicon Valley. The institute argues that this approach could reverse political momentum by giving rural residents a material stake in the AI economy.

The proposal emerges as opposition to data centers intensifies nationwide. The report notes that local moratoriums on data centers have increased from six in 2024 to 59 last year and 294 last month. A Gallup poll cited in the report found that 71% of Americans oppose the construction of an AI data center in their area, a higher figure than the 53% who oppose a nearby nuclear plant. The report attributes part of this opposition to influence campaigns by a network of groups funded by Neville Roy Singham, a tech tycoon living in Shanghai. These groups, including the Party for Socialism and Liberation and CodePink, have used aggressive media campaigns to describe tech executives as "oligarchs" and promote the message "People Over Profits." Rob Joyce, former director of cybersecurity at the National Security Agency, described these actions as "cognitive warfare" intended to push Americans toward outcomes favored by Beijing.

The financial model relies on redirecting property tax revenue that counties already collect from data centers. After funding essential services such as schools and police, local governments would return remaining revenue to residents through annual checks, direct deposits, or tax credits. The report estimates that a single one-gigawatt AI data center could generate enough tax revenue to provide the cited dividend range. This estimate is based on fiscal year 2024 data from Loudoun County, Virginia, which collected approximately $685 million in personal property taxes on data center computer equipment. Researchers calculated this to be roughly $165 million in annual property tax revenue per gigawatt of AI infrastructure.

The report highlights West Feliciana Parish, Louisiana, as a case study where policymakers are attempting to distribute new wealth. The parish expects an AI campus from Hut 8 to generate about $90 million annually. While Louisiana lawmakers passed Act 434 to allow property tax credits funded by this revenue, they removed a provision that would have authorized direct cash payments. The report estimates that if cash payments had been permitted, households could have received between $5,600 and $11,200 annually, depending on how much revenue was distributed after funding government services.

Lyman argued that direct payments create stronger public support than tax cuts or expanded government spending because residents see the benefit immediately. He noted that many Americans feel left out of the AI economic revolution, and the dividend model aims to give them a share of the material prosperity generated by the technology.