Oceanic data farms: Can wave-powered tech solve AI's energy crisis without replicating extractive growth models?
Original framing: “Can floating data centres meet AI's huge energy demand?” — New Scientist
Future modelling suggests that even if wave energy scales, AI’s energy demand will outstrip supply without degrowth in computational excess (e.g., blockchain, crypto, and ad-tech). Scenario planning must account for ‘energy sprawl’—the land and ocean area required to power AI, displacing fisheries and marine ecosystems. Alternative models like federated learning or edge computing could reduce demand by 30-50%, but are sidelined in favor of capital-intensive solutions.
The floating data centre narrative exemplifies the ‘extractive imaginary’ of Silicon Valley, where the ocean is framed as a frontier for endless growth rather than a living system with limits.
This techno-utopianism obscures the structural drivers of AI’s energy hunger—surveillance capitalism’s commodification of attention and the growth imperative of surveillance capitalism—while sidelining Indigenous epistemologies that treat energy as relational, not extractive. Historically, such megaprojects (e.g., ocean thermal energy conversion) have failed due to ecological and logistical hubris, yet the myth persists, aided by publications like New Scientist that center Western innovation paradigms. The solution lies not in engineering fixes but in systemic shifts: degrowth in digital infrastructure, Indigenous-led stewardship, and circular economy models that prioritise collective well-being over corporate profit. Without these, the floating data centre will join the graveyard of failed ‘green’ megaprojects, leaving behind a trail of ecological debt and marginalised communities.