Startup plans to install AI data centers inside offshore wind turbines to tackle energy crisis

Aikido Technologies Merges Wind Power Generation with AI Computing Facilities

San Francisco-based Aikido Technologies is proposing a bold approach to address the mounting energy demands of artificial intelligence by integrating AI data centers directly into floating offshore wind turbines. The startup's AO60DC platform combines renewable energy generation, computing infrastructure, and natural cooling into a single semi-submersible structure, aiming to reduce strain on land-based data centers and electrical grids 1

1

.

In 2025 alone, AI data centers consumed 448 TWh of electricity—roughly equivalent to Germany's annual consumption—with projections suggesting this figure could double within five years 1

1

. This staggering energy consumption of AI infrastructure has intensified concerns about AI sustainability as communities push back against both server farms and the energy infrastructure needed to power them. "Over the past year, as we watched the growing challenges around powering and cooling new data centers, we realized our platform already had ample power and effectively free cooling built in," said Aikido Technologies CEO Sam Kanner 1

1

.

Underwater Data Centers Housed in Ballast Tanks

The innovative design places computing hardware inside the three structural legs of a floating wind foundation, with each leg extending approximately 20 meters deep into the ocean. Inside these ballast tanks, which primarily contain fresh water to maintain platform stability, Aikido plans to install data halls rated between 3 to 4 megawatts each 1

1

. A single platform could support roughly 10-12 megawatts of computing capacity, powered by a 15-18 megawatt wind turbine mounted above 1

1

.

Seawater cooling provides a critical advantage for this offshore approach. The system uses passive thermal transfer, moving heat from servers through steel walls directly into surrounding ocean water. Aikido claims the thermal impact remains limited to just a few meters around the structure 1

1

. This cooling method could achieve a Power Usage Effectiveness (PUE) rating of 1.08, significantly better than the global average of 1.5 for conventional facilities 1

1

.

Proof of Concept A1DC System Targets North Sea Deployment

Aikido Technologies plans to test its concept with a Proof of Concept A1DC system in the North Sea off Norway's coast, generating roughly 100 kilowatts while operating AI servers inside the turbine structure. The company aims to launch this prototype before the end of 2026 2

2

. "We have this power from the wind. We have free cooling. We think we can be quite cost competitive compared to conventional data-center solutions," Kanner told IEEE Spectrum 2

2

.

Battery storage integrated into the platform will store excess energy and smooth fluctuations in wind output, enabling near off-grid operation. While a grid connection will serve as backup during extended low-wind periods, the goal is to operate primarily on locally generated renewable energy 1

1

2

2

.

Challenges Ahead for Offshore AI Infrastructure

Despite the promise, significant hurdles remain. Saltwater exposure, severe weather, and constant wave motion create demanding conditions for computing hardware typically housed in controlled environments 2

2

. Long-term maintenance demands could prove challenging for these semi-submersible platforms. However, Aikido notes that all required technologies already exist, with offshore wind and oil suppliers capable of manufacturing modular steel structures at existing sites. Data halls will be prefabricated onshore before port-side integration 1

1

.

If successful, this approach could reshape how AI computing facilities are deployed, moving them away from contested land sites while directly addressing energy and cooling challenges. The next five years will determine whether offshore installations can scale beyond experimental deployments to meet surging AI demand.🟡 electricians. The image conveys the typical setup of a computing facility.