A British company, Conflow Power Group, proposes to cut the environmental footprint of artificial intelligence by turning streetlights into off-grid, solar-powered mini data centers using Nvidia hardware. It frames this as a response to the rapidly growing electricity, carbon, and water demands of conventional AI data centers, worldwide.
Rising AI energy and water use
Globally, AI data centers already consume an estimated 415 terawatt-hours of electricity annually, with projections suggesting this could rise to about 945 terawatt-hours by 2030, prompting some governments to expand nuclear generation to keep up. A Cornell University study warns that current AI growth in the United States alone could add around 44 million tons of carbon dioxide per year by 2030, roughly equivalent to putting 10 million gasoline cars on the road. Researchers at the University of California, Riverside further estimate that AI-related cooling could demand about 1.7 trillion gallons of fresh water per year globally by 2027.
The iLamp concept
Conflow Power Group’s answer is the iLamp, a solar-powered streetlight that incorporates Nvidia Jetson AI processors and other electronics while operating entirely off-grid. The iLamp requires no external electricity connection and instead relies on integrated, low-maintenance, self-cleaning solar panels that generate between 200 and 600 watts, depending on configuration, while the fixture itself consumes only about 80 watts. The surplus power can run Nvidia Jetson modules, each drawing roughly 15 watts, along with add‑on technologies such as AI-enabled cameras, sensors, Wi‑Fi, and GPUs.
Economics and business model
The base iLamp unit is reported to cost £7,500 (about $10,000 USD) and comes with a 20‑year warranty, but customers can offset or even reverse this cost because AI providers like OpenAI pay to use the on-board compute resources. Instead of bearing ongoing energy and maintenance expenses typical of traditional lighting and IT infrastructure, owners may earn revenue from the processing capacity embedded in the streetlights. This model aims to turn what is usually a pure cost center—public or private lighting—into a long-term income-generating asset while maintaining near-zero operational carbon emissions.
Distributed, low‑carbon AI infrastructure
By deploying iLamps widely in cities and towns, Conflow Power Group envisions creating a large, distributed AI data center built into existing urban lighting networks. Because the units run on local solar generation, the associated AI computation is largely decoupled from fossil-fuel-based grids and does not require the water-intensive cooling systems of conventional data centers, thereby reducing both carbon and water footprints. Locating processing closer to end users also promises lower latency for AI services, combining environmental benefits with potential performance gains.
More information is available here.
Image above: http://www.ilamp.com/index.html
You must be logged in to post a comment.