A device roughly the size of a home air conditioner, bolted to the side of a newly built house, quietly processes AI queries from cloud providers across the country while the family inside sleeps. No fans, no noise complaints, and a monthly utility bill that may drop to zero. That’s the actual pitch behind XFRA, and the company behind it isn’t a fringe operation – it’s an Nvidia-backed startup that already has one of America’s largest homebuilders running tests.
The setup sounds almost too convenient to be real. But before anyone signs over their garage wall, the headline figure floating around social media – that homeowners stand to pocket over $220,000 a year – deserves a much closer look. Verified sources tell a different story, and the actual financial arrangement, while genuinely interesting, is more modest than the viral number suggests.
Span started in San Francisco in 2018, specializing in hardware that helps homes manage electrical loads. The company’s new product, XFRA, applies that same electrical control technology to a completely different problem: the AI industry’s near-total inability to connect new computing power to the grid fast enough. The partnership with Nvidia and homebuilder PulteGroup, announced in April 2026, turns residential homes into distributed AI data center nodes – an approach that is simultaneously straightforward and genuinely unprecedented at scale.
What an AI Data Center Home Actually Looks Like
Each XFRA node contains 16 Nvidia graphics processing units (GPUs), four central processing units (CPUs), and three terabytes of RAM. To put that in perspective, Mahadev Satyanarayanan, a computer scientist at Carnegie Mellon University known as the “Father of Edge Computing” for his foundational work in distributed systems, described the hardware as “pretty beefy,” noting that “even a modest-size large language model could run on a 16-GPU cluster.”
Each node draws about 12.5 kilowatts at full power, according to Chris Lander, XFRA’s vice president – meaning roughly 8,000 XFRA nodes match the power demand of a medium-sized 100-megawatt data center. For raw context on what that means for household energy use: an XFRA node running at full power would consume as much energy in three days as the average U.S. household uses in an entire month.
The hardware sits outside the home, mounted on the exterior wall alongside existing HVAC and electrical equipment. Span uses Nvidia’s liquid-cooled RTX PRO 6000 Blackwell Server Edition GPUs, which require no fans, making the units quiet. That fan-free design directly addresses one of the most common community complaints about conventional data centers: constant mechanical noise. Each XFRA node sits beside a house, runs almost silently thanks to liquid cooling, and taps into the spare electrical capacity that most modern American homes already have.
The question of how the unit draws power without disrupting normal household use is central to Span’s pitch. Most newer single-family homes are wired for 200 amps of service but typically use closer to 80 at peak. Lander put it simply: “Virtually all homes with 200-amp utility services have 80 amps available at all times, so we set that as the maximum power consumption for a single XFRA node.” Appliances run normally, and if household consumption spikes, the home battery kicks in first to keep the GPUs fed.
What Homeowners Actually Get Paid
The $220,000-a-year figure in the headline circulating online has no foundation in anything Span has published or confirmed. No verified source – not the company’s white paper, not CNBC, not Scientific American, not pv magazine – contains that number. It appears to be a wildly inflated claim spreading without any citation.
Here is what the sources actually confirm. According to pv magazine, Span says it will install the Span smart panel and battery backup at no cost to the host, and also provide them with discounted rates for electricity and internet. The financial model most frequently cited across multiple outlets: Span takes over both the electricity and internet bills, then either charges a flat fee – the example mentioned was $150 per month – or nothing at all.
Ryan Harris, Span’s chief revenue officer, stated that some homeowners in places where XFRA nodes provide the most value could receive “discounted electricity up to and including free electricity and free internet access.” That’s meaningful for households in high-rate states, but it’s a long way from a six-figure annual income. Contrary to social media claims suggesting homeowners earn significant passive income or “get paid” to participate, the model involves homeowners paying Span a flat monthly fee typically cited as around $150.
The real benefit comes in the form of offsetting bills you’d already be paying, plus free infrastructure worth thousands of dollars. The Span systems include a smart electrical panel, the XFRA unit, a home backup battery, and in some cases, solar panels – all installed at no cost to the homeowner. That bundle has genuine value, particularly the 16 kWh backup battery, which homeowners also receive along with covered utility bills and free internet.
Why the AI Industry Needs Your Living Room
The hunger for distributed AI infrastructure isn’t manufactured urgency. AI’s rapid growth has strained power grids nationwide and, in some cases, resulted in higher electricity bills for homeowners. The constraint isn’t computing hardware – it’s getting power to where that hardware needs to go.
Access to electricity has become one of the AI industry’s biggest constraints, with utilities unable to connect power-hungry data centers to the grid fast enough. Building new substations to support the demand doesn’t happen quickly. According to Scientific American, substation upgrades to support a 100-megawatt data center now take four to seven years in most parts of the U.S.
The numbers on demand growth make that timeline feel genuinely alarming for the industry. More than 2,060 gigawatts of generation and storage capacity sat in interconnection queues as of late 2025 – meaning projects that want to connect to the grid but simply can’t get through the line. Meanwhile, Goldman Sachs projects that US data center power demand is expected to climb from 31 gigawatts in 2025 to 66 gigawatts by 2027 – more than doubling in two years. The International Energy Agency goes further, projecting that global electricity consumption for data centers will double to reach around 945 TWh by 2030.
Distributed residential nodes don’t solve AI model training, which requires thousands of chips exchanging massive data simultaneously in centralized facilities. But inference – when trained models answer queries or generate content – requires far less coordination between processors, with many requests processed independently and routed to whichever node is closest to the user. Deloitte estimates that inference workloads will account for roughly two-thirds of all AI compute by 2026 – and that’s exactly the category of workload that XFRA nodes are designed to handle.
The speed and cost advantage of distributed deployment is where Span’s pitch gets most compelling. According to CNBC, Span says it can install 8,000 XFRA units about six times faster and at five times lower cost than the construction of a typical centralized 100-megawatt data center.
The Rollout Plan and the Real Unknowns
Span has already conducted in-house testing of prototypes with paying customers, and the company is now planning a proof of concept for Q3 2026, deploying 100 nodes in residential new-construction homes. Span CRO Ryan Harris revealed the program will take place in a southwestern state, likely Nevada or Arizona. The company plans to scale from that 100-home pilot to 80,000 nodes across the US by 2027, providing over 1 gigawatt of distributed compute.
The partnership with PulteGroup is key to that rollout. PulteGroup, one of the nation’s largest homebuilders, is in the early testing phase as it assesses the capabilities and economics of the XFRA nodes. The model targets newly built homes specifically, since they come with the newer 200-amp electrical infrastructure the system requires.
Not every voice in this conversation is enthusiastic. Ari Peskoe, director of the Electricity Law Initiative at Harvard Law School, called the homeowner subsidy “fascinating,” but raised a practical question about clustering: “If there’s a block that has several homes with these devices, maxing out compute and energy would force a lot of power to that local area.” Rich Brown, who spent more than 30 years researching energy technologies at Berkeley Lab, noted that “utility operators and planners depend on diversity of loads to average out peaks and valleys,” and that distributed data centers “would eliminate some of the benefits of that diversity by filling in all the valleys and perhaps creating new peaks.”
Security is a legitimate concern that hasn’t been fully answered. Unlike a data center behind layers of physical security, an XFRA node sits on a residential wall next to a fence. Experts have flagged concerns about grid strain, physical theft, and side-channel attacks alongside the financial benefits. And for homeowners accustomed to controlling their internet traffic, others are concerned about how easy it might make it for companies to harvest online data. Those concerns aren’t hypothetical – similar questions arise with any always-on internet-connected device routing traffic through a third-party system.
Robert Davies, an associate professor of physics at Utah State University who researches the environmental impacts of data center expansion, warned that efforts to reduce the ecological harms of data center power usage could actually exacerbate the problem, and in a preliminary analysis, he calculated that only 30 to 40 percent of homes may be suitable for mini data centers due to integration constraints, the need for stable internet, and participants being willing to have the technology installed.
Read More: To Bee or Not to Bee: Meta’s Plans for a Nuclear-Powered AI Data Center Come to an End
What This Means
The XFRA program is currently limited to newly built homes partnering with PulteGroup, and the first commercial pilot doesn’t launch until Q3 2026. If you own an existing home, you cannot sign up yet. This initiative remains in the early stages, focused initially on new construction, and is not yet available for retrofitting existing homes nationwide.
If you’re in the market for a newly built home in the Southwest, ask your builder directly whether they’re partnering with Span, and read any hosting agreement carefully before agreeing. The economics on offer – free installation, a smart panel, battery backup, and subsidized or free electricity and internet – are genuinely valuable, particularly in states where electricity rates are rising. But the fine print on long-term contracts, liability for equipment on your property, and what happens to your utility rate if a block of homes starts drawing at near-capacity simultaneously hasn’t been published yet.
The broader picture is real regardless of whether XFRA itself succeeds. Fourteen US state legislatures are currently considering data center bans or pauses, and 47% of Americans oppose hyperscale data centers in their neighborhoods. Distributed residential infrastructure sidesteps that opposition almost entirely, which is precisely why the model is attracting serious capital and serious hardware partners. The $220,000 figure that circulates online is not real – no verified source confirms it. What is real is a free smart panel, a 16 kWh battery, and potentially a zero electricity bill. Whether that’s worth hosting AI infrastructure on your home’s exterior wall depends on the terms, and those terms haven’t been finalized.
Disclaimer: This information is not intended to be a substitute for professional financial advice, investment advice, tax advice, or legal advice, and is provided for informational purposes only. Always seek the guidance of a qualified financial advisor, accountant, or other licensed professional regarding your personal financial situation or investment decisions. Do not make financial, investment, or tax decisions based solely on information presented here. Past performance is not indicative of future results, and all investments carry risk, including the potential loss of principal.
AI Disclaimer: This article was created with the assistance of AI tools and reviewed by a human editor.
Read More: Amazon Data Center Under Scrutiny After Reports of Rare Cancer Cluster