Portuguese 
English 
Spanish 
5 min of reading 
0 comments 
A proposal that combines floating wind energy, artificial intelligence, and maritime infrastructure repositions the debate on where to install data centers and how to handle the growing demand for electricity in the technology sector.
The expansion of artificial intelligence infrastructure has opened a debate that goes beyond chips and billion-dollar investments: the increasing demand for energy.
In this scenario, Aikido Technologies, a company based in San Francisco focused on offshore projects, has begun to advocate an unusual proposal: to install data centers for AI within floating wind platforms at sea.
According to the company, the idea combines electricity generation, storage, cooling, and processing in the same set, with the proposal to reduce pressure on land networks and decrease competition for land areas designated for new data centers.
-
From ex-PagBank to “talk to me, Jota”: the Brazilian who wants to retire the card reader, create voice payment, turn the cell phone into a card terminal, and target billions in transactions without the customer even opening the bank app.
-
Beneath Bermuda, scientists have found a colossal rock structure 20 km thick that looks unlike anything ever seen on Earth and may explain one of the strangest geological mysteries of the Atlantic.
-
For the first time in history, electrical discharges have been recorded in the atmosphere of another planet, and the discovery by Perseverance on Mars now forces NASA to completely rethink the safety of all equipment and habitats sent to the Martian surface.
-
Artemis II astronauts return from space and request Uncrustables after ten days for their first meal, and the Navy supplies the ship with an “abundant amount,” the manufacturer promises a lifetime supply.
Energy consumption of data centers enters the center of the debate
The initiative arose at a time of accelerated growth in electricity consumption by data centers in various regions of the world.
Data from the International Energy Agency indicate that these structures consumed about 415 terawatt-hours in 2024, equivalent to approximately 1.5% of global electricity demand.
The agency’s projection suggests that this volume could reach 945 terawatt-hours by 2030, driven primarily by the expansion of AI.
In the market, other estimates also reinforce the growth of energy demand in the sector.
The consultancy Gartner calculated 448 terawatt-hours in 2025 and indicated the possibility of nearly doubling by 2030.
The numbers help explain why network operators, governments, and technology companies have begun to treat the supply of energy for data centers as one of the main challenges of the new phase of artificial intelligence.
How Aikido Technologies’ proposal works at sea
At the center of the proposal is a concept called AO60DC.
Instead of building a campus on land and connecting it to distant wind farms, Aikido wants to bring processing closer to the renewable source.
The project combines a large-scale wind turbine, batteries, and data center modules installed on a floating foundation.
The platform has three structural legs, and each one can accommodate a computing module on top of the ballast tanks.
In practice, the company is trying to adapt the base of an offshore turbine to house computational infrastructure at sea.
The announced configuration anticipates a capacity of approximately 10 to 12 megawatts of computational load per unit, powered by a turbine of 15 to 18 megawatts.
The system also includes battery storage to handle variations in wind.
The conventional power grid does not disappear from the project but appears as a backup.
According to the company, daily operations would primarily depend on the energy produced within the structure itself.
In a statement reproduced by the specialized press, Sam Kanner, CEO of Aikido, stated that the company identified two central elements for the sector on the platform itself: availability of energy and natural cooling.
Natural cooling and energy efficiency of data centers
The cooling system is presented as one of the pillars of the proposal.
According to Aikido, the heat generated by the servers would be passively transferred through the steel walls of the tanks and dissipated into the surrounding seawater.
The company projects a PUE of less than 1.08, an index used to measure the energy efficiency of data centers.
In conventional installations, this indicator usually remains above this level.
Therefore, the company argues that the structure could reduce spending on climate control and support infrastructure compared to land-based data centers.
This performance, however, is still treated as a projection, as the technology has not been tested on a commercial scale.
Industrial chain and reuse of offshore areas
In addition to the technological appeal, the project also relies on an industrial logic.
The company claims that a good portion of the necessary components already exists in the offshore market, especially in the chains related to wind energy, oil, and gas.
The metal structures could be manufactured in facilities already used by these industries, while the data center modules would be assembled on land and integrated into the platform before operation at sea.
With this argument, Aikido tries to present the proposal not as a system created from scratch, but as a combination of technologies already available in established production chains.
The company’s assessment is that this industrial base can reduce some of the complexity of deployment, although operating in a maritime environment still imposes its own challenges.
Another point explored by the company is the availability of areas already mapped by the floating wind industry.
In its institutional material, Aikido states that there are more than 50 gigawatts in offshore locations considered ready for construction or potentially reusable for this type of deployment.
According to the company, some of these assets have lost traction as traditional wind projects but could gain a new function when associated with the demand for high-density computing.
The company’s reading is that the project can fit precisely at the intersection of two bottlenecks.
On one hand, the difficulty of licensing and energizing data centers on land is increasing.
On the other hand, some offshore wind projects are facing delays in advancing.
In this context, the proposal tries to present itself as an alternative to make use of areas already studied while simultaneously responding to the growing demand for computational capacity.
Prototype in Norway and commercial plan in the United Kingdom
For now, the timeline remains on a reduced scale.
The first announced demonstration is a unit of 100 kilowatts, associated with a refurbished Vestas V-17 turbine, with a launch planned in the North Sea, off the coast of Norway, by the end of 2026.
The next commercial stage targets the United Kingdom, where the company claims to have already identified a location and conducted engineering and business discussions for an operation planned for 2028.
In the long term, the plan is broader.
Aikido states that it intends to form offshore farms capable of supporting 30 megawatts to over 1 gigawatt in computational capacity.
So far, however, the proposal remains in the realm of announced projects and preliminary development stages.
AI infrastructure seeks alternatives outside urban centers
The discussion around the idea occurs at a time when the AI industry is seeking alternatives to expand its infrastructure without solely relying on the expansion of congested land networks or the construction of new campuses in urban and peri-urban areas.
In various markets, local resistance to the advancement of large data centers is compounded by delays in reinforcing transmission lines, substations, and licensing processes.
By transferring part of this infrastructure to the sea, the company attempts to advocate a solution that combines nearby renewable generation, natural cooling, and less land occupation onshore.
In practice, the advancement of the project will depend on the ability to transform this design into a continuous operation in an offshore environment, where factors such as maintenance, corrosion, connectivity, and equipment protection also weigh on the model’s viability.
Seja o primeiro a reagir!