Eastport Temporarily Bans Underwater Data Centers

Eastport has temporarily stopped underwater data centers, turning one unusual google news headline into a direct conflict over AI infrastructure and local control.

The Maine city’s action targets an early-stage proposal from DeepGreen Western Passage SPV LLC. The company wants to combine tidal generation with submerged computing equipment in the Western Passage, near Eastport and the Canadian border.

This is not simply another dispute about a warehouse full of servers. DeepGreen proposed placing major power and computing infrastructure in working coastal waters. Those waters support fishing, navigation, tourism, tribal interests, and a marine environment that residents consider economically inseparable from the city.

The moratorium creates breathing room while Eastport considers citizen-initiated ordinances that would impose stronger restrictions. It does not settle whether submerged computing works, whether DeepGreen can finance the project, or which government controls every part of the site.

What it does settle, at least temporarily, is the order of operations. Eastport wants rules and public consent before an experimental infrastructure plan gains momentum.

That reversal matters beyond Maine. Data center developers often present local permitting as a later implementation question. Eastport is treating it as an early test that a project must pass before its technical vision carries much weight.

Eastport Put the Project on Pause Before Construction Began

Eastport’s moratorium stops a speculative proposal from gaining local momentum while the city decides what developments its rules should permit.

DeepGreen’s project began with a federal preliminary-permit application filed in February 2026. A preliminary permit allows a developer to study a hydroelectric site while preserving priority against competing applicants. It does not authorize construction.

The proposal described a 51-megawatt tidal power and subsea computing project in the Western Passage. DeepGreen sought a 48-month study period covering engineering and environmental work.

According to an early project description, the concept included 170 tidal turbines and 34 submerged computing pods. The active equipment would occupy about 27 acres within a broader 400-acre study area.

The pods would contain high-density computer servers for AI workloads. Docking structures on the seafloor would connect those pods with turbines, electrical systems, and monitoring equipment.

A cable would come ashore around Kendall’s Head. DeepGreen also said tidal generation could serve the computing equipment and provide electricity to nearby communities.

Those details made the proposal more concrete than a generic request to study tidal energy. They also connected the project to places residents already use and regulate.

The city’s response developed in stages. Washington County commissioners expressed opposition in April, citing the region’s dependence on maritime industries and the risks posed to local waters.

Eastport residents then pressed the City Council for a moratorium. Councilors initially declined, reasoning that the project remained preliminary. They instead agreed to communicate their objections to the developer.

The political situation changed after residents organized a formal initiative. The Eastport Coalition for Healthy Oceans gathered signatures for two proposed ordinances.

One measure would prohibit large-scale data centers within Eastport, including qualifying structures in its coastal waters. Another would require voter approval for industrial or commercial developments exceeding specified size thresholds.

The city scheduled hearings for August 5, according to its official hearing notice. Residents filled the proceedings and argued over environmental protection, economic development, navigation, fishing, and municipal authority.

The temporary ban prevents applications from moving through local channels while that larger decision remains unresolved. The permanent measures are separate from the moratorium and require their own legal and electoral process.

That distinction is important. Eastport has not completed a scientific finding that all underwater data centers are inherently unsafe. It has decided that existing rules are insufficient for evaluating one.

The measure also does not cancel DeepGreen’s federal process. Federal energy regulators can examine the preliminary-permit application under their own authority.

However, a federal study permit cannot manufacture local support. It also cannot eliminate every shoreland, land-use, cable-landing, port, environmental, or construction approval that a full project would require.

That gives Eastport leverage at the moment when DeepGreen has the least invested in physical infrastructure. The city is intervening before studies, partnerships, and sunk costs create pressure to treat construction as inevitable.

For readers arriving through google news, that is the first essential distinction. Eastport did not close an operating data center. It paused a proposal whose commercial, technical, and environmental assumptions remain unsettled.

Why This Google News Story Puts DeepGreen Under Pressure

The moratorium forces DeepGreen to prove that its plan is locally workable, not merely imaginable on an engineering diagram.

The proposal combines several difficult projects into one. DeepGreen needs tidal turbines, underwater server pods, marine foundations, power electronics, fiber connections, shore infrastructure, environmental monitoring, financing, and regulatory approval.

Each component can affect the others. A turbine layout that improves generation might complicate fishing or navigation. A location suited to cooling might make equipment recovery harder. A cable route might encounter protected shoreland.

DeepGreen also entered the debate without an established operating record for a comparable installation. Its managing member, Louis Wolfson, told local reporters that the company had not selected a turbine supplier when the proposal became public.

Wolfson described the plan as an effort to assemble proven technologies. Yet combining known components does not automatically produce a proven system.

The project depends on whether its interfaces work. Turbines must generate usable power under changing tidal conditions. Submerged pods must remain sealed, connected, serviceable, and thermally stable. Operators must recover or repair equipment without disrupting other marine users.

Those questions would challenge an experienced infrastructure consortium. They become more prominent when the lead developer has not demonstrated a completed tidal generation or subsea computing project.

DeepGreen’s public outreach created another pressure point. Its application said the company was pursuing a community benefits agreement with Eastport and an understanding with the Passamaquoddy Tribe at Sipayik.

However, city and tribal representatives told reporters that formal agreements had not been established. Eastport City Manager Brian Schuth said he had spoken with Wolfson but had not discussed a formal agreement.

That gap does not prove bad faith. Developers regularly identify agreements they intend to pursue before negotiations are complete.

Still, wording matters when a project depends on community consent. Residents may interpret an aspirational agreement as evidence that decisions are advancing before public institutions have accepted the premise.

DeepGreen says the project could deliver electricity, tax revenue, computing capacity, and economic activity. Those benefits remain claims rather than contracted outcomes.

The city’s temporary ban changes what the developer must show next. General promises about green computing will carry less weight than site-specific answers.

DeepGreen must explain who would manufacture the turbines, who would integrate the pods, and how the system would meet reliability requirements. It must identify realistic customers and a credible financing path.

The company also needs to clarify how much power would reach the community after the servers take their share. Tidal energy is predictable because tide schedules can be forecast. It is not constant because generation rises and falls with water movement.

A 51-megawatt nameplate does not mean the project would continuously deliver 51 megawatts. Actual output depends on turbine performance, deployment density, maintenance, environmental restrictions, and the site’s capacity factor.

The moratorium therefore shifts the contest from narrative to evidence. DeepGreen can no longer rely mainly on the appeal of pairing renewable energy with AI infrastructure.

Its strongest response would be a narrower and more testable plan. That might include independent environmental work, a defined pilot, a selected technology partner, and enforceable limits on marine disturbance.

Its weakest response would be treating local opposition as an information problem that better marketing can solve. Residents are questioning who controls the water, who assumes failure risks, and who receives measurable benefits.

Those are governance questions. Engineering presentations alone cannot answer them.

The Real Fight Is Local Consent Versus Technical Promise

Eastport’s dispute pits a developer’s integrated technology vision against a community that wants the right to reject the experiment.

Underwater computing offers an attractive premise. Servers produce heat, while cold surrounding water can help carry that heat away. Sealed pods can also reduce exposure to humidity, oxygen, dust, and human handling.

Microsoft supplied the best-known American precedent through Project Natick. The company retrieved a test vessel from waters near Scotland in 2020 after operating it for two years.

Microsoft reported a lower server failure rate inside the sealed, nitrogen-filled vessel than in conventional facilities. The experiment showed that submerged servers can operate for an extended period under controlled conditions.

It did not establish that every coastal site is suitable. Nor did it prove that a commercial installation paired with tidal turbines would be economical or acceptable to a working waterfront.

Project Natick involved a carefully bounded research deployment backed by one of the world’s largest technology companies. DeepGreen’s Eastport concept is broader.

The Maine proposal couples power generation and computing while promising community electricity and economic benefits. It would place numerous turbines and server pods in waters already supporting other activities.

That combination increases potential efficiency, but it also expands the number of ways the project can fail. A server-pod problem becomes a marine maintenance operation. A generation shortfall becomes a computing and grid problem.

The community’s concerns begin with physical space. Fishermen told local reporters that the Western Passage supports lobster and scallop activity, including areas associated with egg-bearing lobsters.

Equipment on the bottom could displace gear even if it occupies only part of the proposed study area. Cables, safety zones, vessel movements, and maintenance corridors can affect a larger footprint than the hardware itself.

Maritime navigation presents another uncertainty. Pilots and port users need to know whether foundations, turbines, or service operations would interfere with established routes.

Environmental effects require more than analogies to other data centers. Tidal turbines can create collision risks, underwater noise, electromagnetic fields, sediment disturbance, and changes in how people access an area.

The magnitude of those effects depends on equipment design and placement. That is precisely why residents object to assurances offered before a final design exists.

Cooling raises similar questions. Cold seawater can absorb heat effectively, but the environmental outcome depends on how a system transfers and disperses that heat.

A sealed pod using an internal cooling loop differs from a facility that pumps seawater through heat exchangers. Readers should not assume DeepGreen’s design will behave like every other underwater installation.

Maintenance creates a practical tradeoff. Sealing equipment away from people can reduce accidental failures, but technicians cannot simply walk into an underwater pod.

Operators must retrieve a unit or use specialized marine equipment. Weather, currents, vessel availability, and port capacity then become part of the data center’s repair model.

AI hardware also changes quickly. Accelerators may become commercially outdated before marine foundations reach the end of their useful lives.

A modular cradle could help by allowing pods to be swapped. Yet modularity only creates value if recovery is safe, timely, affordable, and repeatable.

DeepGreen’s proposal therefore places two different clocks against each other. Marine infrastructure is designed for long operating periods, while high-value computing hardware often turns over much faster.

Eastport residents are not required to solve that design problem for the developer. Their question is whether the city should expose shared waters to an experiment before those mechanics are clear.

The moratorium answers that question with a temporary no. The citizen proposal could turn it into a longer prohibition.

That sequence is the core reversal in this google news event. DeepGreen sought time for technical studies, but Eastport first claimed time for political and regulatory study.

What the Underwater Data Center Pitch Does Not Prove

The concept has credible technical precedents, but DeepGreen has not yet established that its specific Eastport plan can deliver the promised balance of energy and computing.

One risk is treating a preliminary permit as validation. Such a permit can secure priority for studying a site, but it does not certify engineering feasibility or environmental safety.

The permit process is designed to develop information. It should not be presented as confirmation that a project will receive a construction license.

A second risk concerns the claimed environmental advantage. Tidal energy has no fuel combustion during generation, and seawater cooling can reduce dependence on conventional cooling systems.

However, a lower-carbon design is not automatically a low-impact design. Construction vessels, foundations, cables, equipment retrieval, habitat disturbance, and end-of-life removal still matter.

The project also needs a credible energy architecture. AI computing requires dependable electricity, while tidal generation follows a cyclic production curve.

DeepGreen would need storage, grid connections, backup supply, workload scheduling, excess generation, or some combination of those options. The public proposal has not resolved that architecture in enough detail for outsiders to assess it.

If servers draw electricity from the grid when tidal output falls, the project’s marketing must distinguish locally generated energy from total consumption. If workloads pause with the tides, DeepGreen must identify customers that accept intermittent computing.

Flexible AI tasks can sometimes move across time and locations. Model training and batch processing may tolerate scheduling better than services requiring immediate responses.

Still, commercial customers usually expect measurable service levels. A data center cannot promise ordinary availability while leaving the backup power strategy undefined.

Financial feasibility is equally uncertain. Underwater construction requires marine surveys, specialized equipment, corrosion protection, retrieval planning, and long-term monitoring.

Tidal generation has faced difficult economics in many markets. DeepGreen would be combining that sector’s development risks with an unconventional data center design.

The proposal’s scale also works both ways. A large project can spread certain fixed costs across more equipment. Yet it demands more capital before the design has been demonstrated locally.

A smaller pilot would answer some questions without committing the city to the full concept. It could measure power output, marine interactions, equipment reliability, noise, heat transfer, and maintenance requirements.

Even a pilot would require defined safeguards. Eastport would need baseline environmental data, clear removal obligations, liability coverage, and transparent reporting.

The current debate has not yet produced agreement on who should design those conditions. A permanent ban would make that exercise unnecessary within the city’s jurisdiction.

Jurisdiction itself remains a significant uncertainty. The proposal crosses federal energy regulation, state environmental oversight, municipal land-use authority, coastal rules, and potentially tribal interests.

Eastport can regulate activities within its lawful reach. It cannot independently decide every federal question involving navigable waters or hydropower licensing.

DeepGreen cannot rely on federal jurisdiction to erase local dependencies either. A project still needs workable cable landings, service access, construction logistics, and relationships with the surrounding community.

Maine’s wider policy debate reinforces that tension. State lawmakers approved a temporary restriction on large data centers, but Gov. Janet Mills vetoed it after objecting to its treatment of a supported project in Jay.

Mills nevertheless acknowledged concerns about environmental effects and electricity rates. The dispute concerned how a moratorium should distinguish between projects, not whether scrutiny was appropriate.

The statewide veto left municipalities with more responsibility for immediate decisions. Bangor imposed its own six-month freeze, while other communities took different approaches.

Eastport’s case is more complicated because the proposed facility is partly marine infrastructure. Standard zoning language written for warehouses may not address submerged pods, tidal turbines, or seabed docking systems.

This is why the pause should not be described as proof that the technology is dangerous. It is evidence that the regulatory vocabulary has not caught up with the proposal.

The burden now rests with DeepGreen. The company must demonstrate that technical novelty does not become a shortcut around ordinary public consent.

Eastport Is Part of a Wider Data Center Backlash

The Eastport vote reflects a broader shift from welcoming data centers as abstract investment to examining their site-specific costs before approval.

For years, many local governments evaluated data centers mainly through land use, tax revenue, construction activity, and grid access. AI demand has changed the scale and urgency of those conversations.

Modern projects can request power loads comparable to industrial facilities. Communities increasingly want to know who funds grid upgrades, how rates are protected, and whether energy demand competes with other development.

Water use is another concern for land-based facilities that rely on evaporative cooling. Noise from cooling systems and backup generators can affect nearby residents.

An underwater data center changes some of those pressures without removing the public-interest test. Eastport is less focused on cooling towers than on fishing grounds, tidal equipment, cable routes, and seabed occupation.

The conflict therefore demonstrates why “data center” is becoming an inadequate regulatory category. A cloud facility inside an existing industrial building is different from a hyperscale campus.

Both differ from server pods attached to tidal power systems in coastal waters. Local rules need definitions that capture physical and operational effects, not merely the presence of computers.

Bangor adopted a temporary freeze because officials said they lacked enough information about infrastructure and resource impacts. Its six-month moratorium showed that Maine’s concern was not limited to DeepGreen.

Nationally, other cities and states have considered pauses, special electricity rates, environmental studies, and new zoning categories. The trend does not amount to a universal rejection of data centers.

Instead, it marks the end of automatic assumptions. A data center cannot be treated as a quiet commercial building when its energy and infrastructure requirements resemble heavy industry.

Developers also face a credibility challenge created by AI’s rapid expansion. Communities hear enormous forecasts about future computing demand, but individual projects may remain speculative.

A proposal can occupy regulatory attention long before it has customers, financing, equipment suppliers, or a final engineering design. Local governments must decide how much public work should support that speculation.

Moratoriums are imperfect tools. They can delay serious projects alongside weak ones. They can also create uncertainty for businesses and shift development toward neighboring jurisdictions.

A permanent prohibition can prevent a community from considering a smaller or safer design later. Broad definitions might accidentally capture research facilities or ordinary server rooms.

Eastport’s citizen measures therefore deserve the same scrutiny as DeepGreen’s proposal. Voters need to understand which projects the language covers, where municipal authority ends, and whether exceptions are possible.

The city must also consider what happens if a court finds part of an ordinance preempted or beyond local authority. Passing a rule is not the same as ensuring that it survives a legal challenge.

Still, those limitations do not invalidate the temporary pause. A moratorium can preserve the city’s choices while officials and voters examine unfamiliar infrastructure.

The larger lesson is procedural. Developers gain trust when they bring communities into decisions before selecting a site narrative and describing anticipated agreements.

DeepGreen approached Eastport with an integrated concept already attached to specific waters. Residents responded by building an alternative process around petitions, hearings, and a possible vote.

That is why the story traveled beyond local government coverage and into google news feeds. It captures a national argument in a particularly visible setting.

AI infrastructure is physical infrastructure. Its pipes, cables, substations, turbines, vessels, and landings occupy places that already have owners, users, histories, and political institutions.

Eastport’s action makes that reality difficult to ignore.

Three Signals Will Decide What Happens Next

The next phase will turn on the local vote, the federal permit record, and whether DeepGreen replaces broad promises with verifiable commitments.

The first signal is Eastport’s decision on the citizen-initiated ordinances. A permanent prohibition would present a much larger obstacle than the temporary moratorium.

The exact text matters as much as the result. Size thresholds, geographic coverage, definitions, enforcement provisions, and voter-approval requirements will determine the rule’s practical reach.

A decisive vote for prohibition would show that opposition extends beyond the most active speakers at public hearings. A rejection would not automatically equal support for DeepGreen, but it would preserve a route for further review.

The second signal is the federal preliminary-permit process. Regulators will determine whether DeepGreen receives priority to study the Western Passage site and under which conditions.

Observers should watch for agency requests, competing applications, interventions, environmental concerns, and changes to the proposed boundary. Each can reveal whether the concept is becoming more specific or encountering structural problems.

Receiving a preliminary permit would strengthen DeepGreen’s ability to conduct studies. It would not override Eastport’s moratorium or amount to construction approval.

A denial, withdrawal, or prolonged procedural dispute would weaken the company’s timetable. It could also redirect attention to DeepGreen’s other proposed location in Alaska.

The third signal is evidence of execution. DeepGreen needs named technical partners, a defined turbine system, a financing plan, and a realistic operating model.

A credible pilot proposal would be more informative than another broad description of an energy hub. It should include independent monitoring, published performance measures, retrieval plans, and financial responsibility for decommissioning.

Community agreements will also require specificity. Residents should be able to identify promised electricity, jobs, tax treatment, environmental safeguards, and remedies if targets are missed.

The Passamaquoddy Tribe’s role cannot be reduced to a line in an application. Meaningful engagement requires direct communication with recognized leadership and respect for tribal interests in the surrounding waters.

These signals should guide anyone following the story through a google news result. The headline describes a temporary ban, but the deeper contest remains active.

Eastport must decide whether any version of the project fits its future. DeepGreen must decide whether it can meet a standard shaped by local users of the water, not only by computing demand.

Developers elsewhere should watch closely. The fastest route to an AI infrastructure project no longer starts with securing power and announcing a site.

It starts with demonstrating why a community should host the physical system, who accepts its risks, and what enforceable value remains locally.

Will DeepGreen return with verifiable partners and a narrower plan, or will Eastport’s pause become a lasting boundary? The next filings and votes will answer more than another google news headline can.