The rapid expansion of artificial intelligence is pushing power demand to new highs, and the pressure is being felt far beyond server racks. 

As hyperscale data centres multiply, utility companies are warning of strained grids and higher electricity bills for communities. 

Against this backdrop, a US wind energy startup, backed by Bill Gates, Airloom, is advancing an alternative turbine design it believes can deliver reliable power faster and at lower cost. 

With the company set to showcase its progress at CES 2026, its technology is drawing attention as a possible energy solution for the AI-driven data centre boom. 

Data centres intensify the energy debate

AI workloads require enormous and continuous power, turning data centres into some of the largest electricity consumers in modern economies. This surge has raised concerns about resource use, grid stability, and long-term energy security. 

Utilities in several regions have already signalled that growing demand from these facilities could translate into higher costs for residential and industrial customers. 

A prototype of a roller-coaster-like wind power system. Image: Airloom Energy.

As policymakers and companies search for cleaner and more scalable power sources, wind energy remains a key option. 

However, conventional wind turbines face challenges related to size, cost, land use, and long deployment timelines. These limits have slowed progress in reducing the levelised cost of energy for wind, especially as supply chains tighten and suitable sites become harder to secure.

A low-profile rethink of wind turbines

Airloom is attempting to bypass those constraints with a radically different structure. Instead of towering horizontal-axis turbines, the company uses a low-profile system standing about 20m to 30m high. The design features a loop of adjustable wings that travel along a track, visually closer to a roller coaster than a classic windmill. 

As the wings move, they generate power in the same way traditional blades do. According to the company, this approach uses 40% less mass while delivering the same energy output. 

The structure also relies on 42% fewer parts and 96% fewer unique components. These reductions are intended to simplify manufacturing and maintenance while cutting overall costs.

The firm says the streamlined design allows projects to be deployed in less than a year, compared to timelines of up to five years for conventional turbines. Overall, the company estimates its systems can be rolled out 85% faster and at 47% lower cost than standard horizontal-axis wind turbines.

From pilot site to CES spotlight

In June, Airloom broke ground on a pilot site to test and validate its technology in real-world conditions. 

Construction is now under way, with the facility designed to confirm performance data such as the power curve while refining operations and maintenance processes. Full build-out is expected ahead of commercial demonstrations planned for 2027. 

While it is not possible to bring a working wind installation to CES, the startup will present its engineering concepts and results at the event. It plans to use the global tech stage to highlight how alternative wind designs could support energy-hungry industries, including AI infrastructure, without placing additional strain on local communities.

Expanding where traditional wind cannot

Beyond cost and speed, the company’s compact and modular turbines open doors to locations that are often off-limits to large wind farms. The system can be installed in low-wind regions, remote islands, mountainous areas, and restricted zones such as military bases or near airports, where large spinning blades are impractical.

The turbines are built with mass-manufacturable, US-developed components, reducing logistical hurdles tied to transport and installation. This flexibility could help unlock new domestic wind capacity at a time when energy independence is becoming a strategic priority.

Meanwhile, Airloom is also exploring offshore wind applications, defense-related uses, and disaster relief scenarios.