Underwater AI Data Centres Proposed Off Maine and Alaska Raise Whale Concerns

What happened

An AI company has proposed building two underwater data centres, with the reported sites located off Maine and Alaska. Critics warn that the facilities could disrupt endangered whale populations in those waters, according to TechRadar's coverage.

The proposal puts subsea computing infrastructure at the centre of a debate about where AI capacity should be built. It also raises the question of how an operator should weigh environmental concerns raised by critics against the infrastructure demands of AI workloads. Public detail in the available reporting is limited, so the specifics of the design, permitting timeline and mitigation plans remain unclear.

Why it matters

AI growth is pushing data centre capacity into new and unconventional locations, and the siting decision is becoming as consequential as the technical design. Underwater facilities are often pitched as a way to reduce cooling demands, but any gain in efficiency has to be set against the ecological and community costs of building in marine environments. For leaders in digital transformation, this is an early signal that infrastructure choices will increasingly be judged by the people and species they affect, not only by performance and cost.

For service designers, the story also highlights a stakeholder group that is easy to overlook. Coastal communities, fisheries, conservation groups and regulators are all users of the proposal in practical terms. They will experience its consultation, permitting and monitoring processes directly, and their confidence will shape whether the project can proceed.

By the numbers

• 2 underwater data centres proposed by the AI company, according to TechRadar's reporting
• 2 US states named in the reporting as proposed sites: Maine and Alaska

The Renascence take

Most coverage will frame this as a clash between AI ambition and whale conservation. The more useful reading is that it is a stakeholder-experience problem, and the operators who treat it that way will move faster than those who treat it as a communications issue.

Treat the whale question as a design brief, not a risk to be managed after the fact. Map every group that will live with this infrastructure, from marine researchers to harbour workers, and design the consultation journey around their evidence needs rather than around the approval timeline. Trust is cheapest to build before the first cable is laid, and it is almost impossible to rebuild once it is lost. Any operator with a data centre pipeline should be asking the same question now: who gets a credible say, and how would we know if they felt unheard?