Aikido Technologies to deploy offshore data centers inside floating wind turbines by year-end

Aikido Technologies Tackles AI Power Consumption with Ocean-Based Solution

As AI power consumption threatens to overwhelm electricity grids, San Francisco-based Aikido Technologies has unveiled plans to build data centers inside offshore wind turbines, combining renewable energy generation with computing infrastructure in a single platform 1

1

. The startup aims to deploy a 100-kilowatt submerged data center prototype off the coast of Norway in the North Sea by the end of 2026, marking a significant step toward addressing the AI data centers power crunch that has pushed some companies to consider space-based alternatives 2

2

.

Data Centers Inside Offshore Wind Turbines: Engineering Innovation

Aikido's approach uses a semi-submersible design similar to offshore oil and gas platforms, featuring three ballast-filled legs that maintain buoyancy and stability. The company plans to place AI-grade compute facilities in the upper sections of each leg, with the capacity to house 3 to 4 megawatts per leg, creating a potential 9 to 12 megawatt data center per turbine 2

2

.

If the prototype succeeds, the company hopes to deploy a larger version featuring a 15 to 18 megawatt turbine feeding a 10 to 12 megawatt data center off the UK coast in 2028 1

1

.

The floating offshore wind platforms address multiple power, cooling, and space demands simultaneously. Fresh water stored in the ballast tanks serves dual purposes: maintaining platform stability and cooling AI chips through a closed-loop system. Warm water is pumped back into the ballast, where the chilly waters of the North Sea dissipate the heat through the steel walls 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," Aikido CEO Sam Kanner told IEEE Spectrum 2

2

.

Offshore Data Centers Solve Multiple Infrastructure Challenges

The ocean-based data infrastructure concept addresses several pressing issues facing hyperscalers. Offshore winds prove more consistent than onshore alternatives, with battery storage bridging any production lulls 1

1

. The approach also eliminates NIMBY concerns from communities opposing data centers near their properties due to noise and pollution 1

1

. Using seawater cooling, the platforms tap into what Aikido describes as an "infinite heat sink," though the company claims thermal impact will be limited to a few meters around the structure 5

5

.

This matters because U.S. data centers consumed 183 terawatt hours of electricity in 2024, representing 4% of the country's total electricity consumption. If current expansion continues, that figure could more than double by 2030 5

5

. By co-locating renewable energy for AI infrastructure with compute facilities, Aikido aims to reduce both power grid strain and the environmental impact of AI's rapid growth.

Learning from Microsoft and Facing Technical Hurdles

Aikido Technologies isn't pioneering the concept from scratch. Microsoft tested underwater data centers over a decade ago, deploying 12 racks of servers in a cylinder off Scotland's Orkney Islands in 2018. During the 25-month trial, only six of more than 850 servers failed, with the data hall filled with inert nitrogen gas 1

1

. Microsoft open-sourced its patents in 2021 before discontinuing the project in 2024 1

1

. China also completed phase one of an underwater data center in Shanghai last year 3

3

.

However, offshore data centers face significant challenges. Corrosion from saltwater threatens equipment longevity, requiring hardened containers and connections . The ocean environment brings debris and marine growth concerns, while regulatory hurdles around heat discharge and marine life protection add complexity 5

5

. The floating wind sector itself faces developmental delays, rising costs, and higher interest rates as government subsidies evaporate, though Kanner hopes to revitalize the industry by reframing the business model 5

5

.

"Before we go off-world, we should go offshore," Kanner stated, positioning the concept as a more practical alternative to space-based data centers proposed by companies like Google and Amazon 4

4

. Aikido's vision extends to building offshore wind farms capable of supporting 30 megawatts to more than 1 gigawatt of AI compute, potentially meeting the rapidly growing demand for high-density AI infrastructure while addressing sustainability concerns 5

5

.