Americans oppose huge AI data centers in their towns. Tiny ones in their homes may be a different story

Americans oppose huge AI data centers in their towns. Tiny ones in their homes may be a different story

News ClipCNBC·Liberty County, TX·5/9/2026

Public opposition is growing against large-scale data centers across the U.S., leading to legislation for bans or pauses in several states, including Maine where a ban was vetoed. Concurrently, a new model of small, home-based data center nodes is being piloted by companies like PulteGroup, Span, and Nvidia as a potential way to alleviate strain on existing infrastructure and distribute compute closer to users. This new approach faces questions regarding scalability, regulatory approval, and cybersecurity.

oppositionelectricitygovernmentmoratoriumzoningenvironmental
Nvidia
Gov: Maine's governor, Maine's legislature, National Conference of State Legislatures
Across the U.S., public opposition to large-scale AI data center construction is intensifying, driven by concerns over land usage, rising electric bills, and the power of big tech. This discontent has spurred legislative action, with 14 states, including Oklahoma and New York, currently considering laws to ban or pause new data center developments. Maine's legislature recently passed a statewide ban, though Governor Janet Mills subsequently vetoed it, and an override attempt failed. Amid this opposition, the real estate industry is exploring an alternative: integrating small, fractional data center nodes directly into individual homes. Homebuilder PulteGroup is piloting a program with Nvidia and California-based startup Span to install liquid-cooled Nvidia RTX PRO 6000 Blackwell GPUs on newly built homes. These nodes process cloud computing workloads, with homeowners receiving benefits like smart panels, battery backup, and discounted electricity and internet, while Span sells the compute capacity to hyperscalers. Proponents, like Balaji Tammabattula, COO of BaRupOn, suggest this model could alleviate pressure on existing infrastructure, improve energy efficiency by repurposing waste heat, and reduce the need for new large-scale construction. However, experts like Gerald Ramdeen of Luxcore and Sviat Dulianinov of Bright Machines caution that home environments lack the power density, security, and environmental controls required for enterprise-grade AI training. Cybersecurity vulnerabilities and regulatory complexities, particularly with Homeowners Associations (HOAs), also pose significant challenges, as noted by Aimee Simpson of Huntress and real estate professional Jeff Lichtenstein. Despite limitations, the concept is being explored as a niche layer of future infrastructure for specific workloads like AI inference or batch processing.