An NVIDIA-backed startup called Span is deploying compact GPU servers directly onto residential homes, paying homeowners roughly $150 per month while covering their full electricity and internet bills. The company argues this approach solves a critical bottleneck in AI infrastructure: traditional data centers take four to seven years to build due to grid interconnection delays, while distributed home-based units can scale six times faster at one-fifth the cost.

What Exactly Is Being Installed on Homes?

Span's XFRA node is a liquid-cooled server unit roughly the size of an air conditioning unit that mounts on the exterior of your home. Each unit packs serious computing hardware designed to run AI inference workloads, which is the process of using a trained artificial intelligence model to generate outputs from new inputs.

The hardware inside each XFRA node includes:

• GPU Count: 16 NVIDIA RTX Pro 6000 Blackwell Server Edition graphics processing units, which are specialized chips designed for heavy computational tasks
• Processing Power: Four AMD EPYC CPUs paired with 3 terabytes of memory, enough to run even modest-sized large language models according to Carnegie Mellon computer scientist Mahadev Satyanarayanan
• Cooling System: Liquid cooling technology that eliminates the need for loud fans, addressing one of the biggest concerns about residential data center noise

Span installs the entire unit at no upfront cost to homeowners. In exchange, you pay a flat monthly fee of approximately $150, and Span covers your complete electricity and internet bills. The company then sells the computing power your node generates to AI companies, cloud gaming providers, and other high-performance computing customers.

Why Is This Solving a Real Problem for AI Companies?

The AI industry faces an acute infrastructure crunch. According to Lawrence Berkeley National Laboratory data cited in the source material, substation upgrades required for a standard 100-megawatt data center take between four and seven years in most parts of the United States. More than 2,060 gigawatts of energy capacity sat stuck in grid interconnection queues as of late 2025, creating a severe bottleneck for companies needing more computing power.

Span's distributed approach bypasses this problem entirely. The company can deploy 8,000 XFRA units approximately six times faster and at roughly one-fifth the cost of a comparable centralized data center. This matters because the average American home only uses about 40 percent of its electrical capacity at any given time, according to Span's analysis. Most homes with a standard 200-amp utility connection have around 80 amps sitting idle, and Span's system captures that unused headroom.

"Fundamentally, it's an infrastructure play. We're uniquely positioned to build infrastructure that can simultaneously help us meet what is clearly an insatiable demand for more compute, much more cost-effectively," said Arch Rao, founder and CEO of Span.

Arch Rao, Founder and CEO at Span

How to Evaluate Whether Hosting an XFRA Node Makes Sense for Your Home

• Financial Trade-off: You receive approximately $150 monthly in utility bill coverage, which works out to $1,800 annually. Assess whether this payment justifies hosting enterprise-grade GPU hardware on your property and the potential data access implications
• Noise and Disruption: While Span claims XFRA units are fanless and run quietly, verify this during the 100-home trial launching later in 2026. Existing data centers have triggered class-action lawsuits over noise levels reaching 84 decibels, affecting over 1,300 homes within a one-mile radius in some cases
• Data Privacy Concerns: Since Span manages both your electrical connection and internet service, your internet traffic routes through Span's network. This raises legitimate questions about whether Span could observe your browsing patterns. The company has not yet published detailed data-handling policies for this arrangement
• Physical Security: Data centers require significant physical security due to the hardware's value. Some users have flagged potential theft and security risks with distributed residential units, though Span has not yet addressed these concerns publicly

Span launched in 2018 with a track record in home energy management. For the XFRA rollout, the company has partnered with NVIDIA, which supplies the GPU hardware, and PulteGroup, one of the largest homebuilders in the US, which will test the system in new residential communities. Prototype deployments are already live in Northern California, with a broader 100-home trial rolling out later in 2026.

Span's chief revenue officer, Ryan Harris, stated that the company targets one to two megawatts of compute by year-end, scaling to over 1 gigawatt annually from 2027. This aggressive timeline reflects the urgency AI companies feel about securing additional computing capacity.

What Are the Real-World Obstacles to Scaling This Model?

The biggest challenge isn't technical; it's social. Communities near existing data centers have already begun fighting back. In Dowagiac, Michigan, residents near the Hyperscale data center filed a class-action lawsuit in federal court, reporting noise readings as high as 84 decibels affecting more than 1,300 homes within a one-mile radius. One resident described the sound as "like having a vacuum cleaner running all the time." In New Jersey, two neighbors sued DataOne USA over an industrial hum that never stops. Virginia residents near Google's "Mango Farm" complex describe the sound as "an internal organ vibration" that triggers anxiety attacks.

The Environmental and Energy Study Institute points out that most local noise ordinances cover noisy parties, not 24/7 industrial facilities running indefinitely. Span's fanless design directly addresses noise concerns, but skeptics note the 100-home trial is small. Scaling to thousands of units across diverse neighborhoods will surface new challenges that prototype testing may not reveal.

Whether homeowners and their communities embrace this model depends on whether the financial incentive outweighs concerns about data privacy, physical security, and long-term neighborhood impacts. The next few months of testing will provide crucial data on whether distributed residential compute can become a mainstream solution to AI's infrastructure crisis.