The rapid expansion of Artificial Intelligence (AI) is driving unprecedented demand for data processing capacity — and with it, sharply higher consumption of energy, water, and physical space. Faced with mounting pressure on conventional infrastructure, technology companies are beginning to explore alternatives beyond dry land, including the open ocean and outer space.
One such proposal is being developed by Aikido, a US-based offshore wind energy company that aims to combine a floating turbine with a data centre. The first prototype is expected to be installed in the North Sea in 2027, featuring a 12-megawatt unit integrated into the ballast tanks of the floating platform.
The solution leverages two readily available resources: electricity generated by the turbine and seawater used in the cooling system. According to Aikido CEO Sam Kanner, the integrated design could enable cost structures competitive with those of conventional land-based data centres.
The ballast tanks extend approximately 18 metres below the ocean surface, while the upper section of the structure houses the computing equipment. Seawater would be used to absorb and dissipate the heat generated by the systems.
The project will allow Aikido to assess the technology's performance and its potential environmental impacts, including any effects on water temperature in the surrounding area. If testing proves successful, a commercial-scale version could move forward by 2028.
Pressure on data centre infrastructure is expected to continue growing. In the United States, electricity consumption associated with data centres is estimated at approximately 75 gigawatt-hours, and could reach 135 gigawatt-hours by the end of the decade, according to estimates cited in the project documentation.
A separate initiative comes from Orbital, a US space startup that aims to take computing capacity into Earth's orbit. The company's first satellite is expected to launch in 2027 aboard a SpaceX Falcon 9 rocket, with the objective of testing space-based computing operations and supporting AI workloads.
Orbital's proposal draws on the solar energy available in space and the conditions of the orbital environment to overcome some of the key constraints facing terrestrial data centres — particularly in the area of thermal management and cooling.
The strain on these facilities is set to intensify further. US data centre electricity consumption, currently estimated at around 75 gigawatt-hours, could climb to 135 gigawatt-hours before the decade is out, according to projections cited by the project.
As AI continues to demand ever-greater computational capacity, the challenge extends well beyond the technological. Companies and investors will also need to identify viable solutions for securing sufficient energy, cooling, and physical space to keep pace with the expanding digital economy.
Source: Diário Económico
Original article: https://www.diarioeconomico.co.mz/?p=546913
