16MAY

Aikido confirms Norway floating-wind compute pilot

4 min read

13:06UTC

Aikido Technologies confirmed an end-2026 METCentre Norway pilot of its AO60DC platform, combining 15-18 MW of floating wind with 10-12 MW of compute and a power usage effectiveness below 1.08.

← Back to OpenAI cuts $800bn; rivals double down Jump to analysis ↓

IndustryDeveloping

Key takeaway

Aikido's METCentre pilot tests whether floating-wind compute can sidestep the Nordic grid bottleneck.

Aikido Technologies, the San Francisco-based offshore-platform developer, confirmed on 3 March that it will deploy its AO60DC floating-wind-plus-compute platform at METCentre Norway by end-2026 ¹. Each unit pairs 15-18 MW of wind generation with 10-12 MW of compute load, with passive seawater cooling through the steel hull driving power usage effectiveness (a measure of total facility power against IT load) below 1.08. Onshore hyperscale campuses typically run above 1.20, with the best designs in cold climates around 1.15.

The concept is the architectural inverse of the behind-the-meter gas pattern that has built GE Vernova's 80 GW gas-turbine backlog . Where the gas pattern colocates generation on the campus fence because the grid cannot deliver, the offshore-wind pattern colocates compute on the generation platform because the grid cannot accept the wind output. Both designs route around the same constraint: substations that take years longer to commission than the load or the source they would connect. METCentre's pilot status keeps the technical claims testable; commercial-scale economics await the deployment cycle.

Nordic siting carries more weight in the pilot than the engineering claims. Equinix and CPP Investments are absorbing the existing onshore platform via the atNorth transaction context, and OpenAI's Narvik plans were among the sites it has now paused. A floating compute platform able to operate without a grid connection at all is the option that the most squeezed corner of the supply map will examine first. Aikido's 2028 first-commercial-UK target will then probe whether the design can carry capex and operating costs against a conventional Nordic colocation build.

Deep Analysis

In plain English

Aikido Technologies is building a floating platform that combines a wind turbine with a small data centre on the same structure. The wind turbine provides the electricity, and cold seawater flowing through the hull cools the computer equipment. The combination achieves very efficient energy use, because no power is wasted on cooling systems that typical data centres on land need. The company confirmed in March 2026 that it will test a pilot version at an offshore test facility in Norway by the end of 2026. If it works, it aims to deploy a commercial version off the UK coast by 2028. The appeal is that it can operate without connecting to the electricity grid at all, which sidesteps the long wait times for grid connections that are currently blocking most new data-centre construction.

Deep Analysis

Root Causes

The AO60DC concept addresses a structural inefficiency in the renewable energy value chain: offshore wind in Norway and Scotland produces power whose coastal transmission capacity is insufficient to deliver it to demand centres.

The grid bottleneck that forces data-centre operators to wait years for onshore connections is the same bottleneck that prevents offshore wind from reaching full utilisation. Co-locating compute on the generation platform bypasses both constraints simultaneously, because the compute load is the offtake agreement for power that would otherwise need transmission.

The broader driver is the disconnect between where electricity is cheapest to generate (offshore, at scale, in windy northern latitudes) and where electricity demand is concentrated (urban and suburban onshore). Aikido's design is an attempt to move the demand to the supply rather than extend the transmission.

What could happen next?

Opportunity
Aikido's 2028 commercial UK target, if the METCentre pilot validates the PUE claims and marine O&M costs, offers operators a grid-bypass route for Norway, Scotland and Iceland deployments where offshore wind capacity exceeds coastal transmission.
Long term · 0.55
Risk
The regulatory intersection of maritime law, offshore wind permitting and data-centre planning in Norway and the UK could add 2-4 years to commercial deployment timelines, making the 2028 commercial target ambitious.
Medium term · 0.7
Precedent
A successful METCentre pilot would be the first verified operational data point for seawater-passively-cooled compute at northern latitudes, establishing an independent PUE benchmark for future offshore-compute proposals.
Short term · 0.65

Source Landscape

This story draws on neutral-leaning sources

Primary parallel: The 1990s floating production storage and offloading (FPSO) vessel industry in deepwater oil and gas followed an identical development arc: technology demonstration on a single converted vessel, pilot commercial deployment at a proven field, then a wave of purpose-built units as the cost-and-reliability case was established.

The FPSO industry reached commercial maturity in roughly 15 years from first pilot to widespread deployment. Aikido's 2028 first-commercial-UK target implies a compressed 2-year arc from pilot to commercial, which is aggressive by FPSO standards but plausible if the compute architecture does not require novel subsea infrastructure.

Counter-parallel: Offshore wind-integrated aquaculture (fish farms paired with wind platforms) attempted the same co-location logic in Norway and Scotland between 2015 and 2022 and has not yet reached commercial scale, primarily because the regulatory framework for offshore hybrid facilities is slower than either the aquaculture or wind regulatory process alone.

Aikido faces the same regulatory fragmentation: the AO60DC sits at the intersection of maritime law, offshore wind permitting and data-centre planning, each governed by a separate authority.

At 10-12 MW of compute per AO60DC unit and a PUE of 1.08, the facility delivers roughly 9-11 MW of IT load from 15-18 MW of wind generation, with the difference covering cooling and ancillary loads. At a high-performance computing rack density of 25-30 kW per rack, a single unit hosts 300-400 racks, or roughly the scale of a medium-sized colocation suite.

The commercial economics depend critically on day-rate operating costs: offshore wind operations and maintenance runs at £80,000-120,000 per MW per year for floating platforms, against £20,000-30,000 per MW for fixed-bottom and under £5,000 per MW for onshore.

At that premium, the compute revenue must cover the power and cooling costs that onshore operators face and the marine operations overhead in addition. Revenue per rack of £100,000-150,000 per year (competitive GPU cluster rates) would produce a positive operating margin at offshore operating costs in a GPU-scarce market, but the same figure falls short in a commoditised colocation market.

Consensus view: Uptime Institute's PUE (Power Usage Effectiveness) research sets the 2025 global average PUE at 1.58 for all data centres and 1.20 for new hyperscale builds in cold climates.

Aikido's claimed sub-1.08 PUE via passive seawater cooling through the steel hull would place the AO60DC below the best recorded conventional data-centre PUE (Facebook's Luleå, Sweden facility at 1.06 in winter), which is technically credible for passive seawater at Norwegian latitudes but has not yet been independently verified at operational scale.

Counter-view: Lawrence Berkeley National Laboratory's DOE-funded US data-centre energy research cautions that small-scale demonstration PUE figures do not transfer directly to commercial-scale deployments: thermal management in a floating steel hull that doubles as a wind turbine base introduces structural vibration and saltwater-ingress risks that onshore data centres do not face.

The METCentre pilot uses a refurbished Vestas V-17, a small and technically non-current turbine, which limits the direct extrapolation to the 15-18 MW wind units the company projects for commercial deployments.

Key tension: Whether the seawater-cooling PUE advantage is sufficient to offset the higher operating and maintenance costs of an offshore floating platform, a cost structure that has historically been 2-3 times higher than equivalent onshore infrastructure in the floating-wind sector.

Sources:GlobeNewswire / Norwegian Offshore Wind·GlobeNewswire / Norwegian Offshore Wind

Mentions:Quaze Technologies →Norway →Uptime Institute →GE Vernova →Equinix →San Francisco →Lawrence →OpenAI →Sweden →Canada Pension Plan Investment Board →

First Reported In

Update #3 · OpenAI cuts $800bn; rivals double down

GlobeNewswire / Norwegian Offshore Wind· 16 May 2026

Read original →

Different Perspectives

Global hyperscale operators

Operators are still filing gigawatt-scale campuses and Meta is proceeding with its $10bn Lebanon, Indiana site despite the county-level bans nearby, betting Q2 capex outruns the patchwork of restrictions. Industry framing casts New York's freeze, Oregon's surcharge and Indiana's bans as taxes and levies that push build-out toward faster-permitting jurisdictions such as India and the Gulf.

EirGrid

EirGrid set a 900 MW instantaneous demand-loss ceiling because a single voltage dip can trip many data centres onto backup power at once, risking imbalance above 1,150 MW. It wrote the limit into a standing procedure rather than waiting for an emergency to force one.

US host communities and ratepayers

Prince William residents backed the 8-0 denial of Dulles South over the Occoquan watershed, drinking water for eight million people, while Oregon's approved tariff cuts residential bills 1.3% by charging large loads 29% more. Their position: consent and cost-attribution belong in law, not left to a developer's or a utility's discretion.

Hassan Allam Digital Infrastructure

Hassan Allam Digital Infrastructure, an Egyptian conglomerate rather than a foreign hyperscaler, reportedly secured a domestic hyperscale licence with a $400m first phase, per single-source reporting still to be verified. It reads as home-grown sovereign compute ambition, building national capacity rather than importing a US or Gulf operator's campus.

Damac Digital

Damac Digital keeps building toward roughly 6,000 megawatts of hyperscale capacity across 13 countries while Virginia taxes power and New York weighs a freeze. Every dollar or month of delay a US state adds is capacity a Gulf developer can site somewhere with faster permitting and no equivalent levy.

Acequia communities, Santa Fe County

Santa Fe County commissioners voted unanimously on 2 July to freeze any data centre over one megawatt, citing the acequia irrigation commons that has shared scarce water since Spanish colonial rule. They expect the low threshold to draw the same Fifth Amendment challenge RCM Hill brought against Hill County, Texas.