Wave-Powered Offshore Data Centers Target AI Inference Grid Limits
Mocean Energy launches 'Blue Core' to co-locate compute with ocean wave power
On August 10, 2026, Edinburgh-based Mocean Energy announced its entry into the AI infrastructure market with "Blue Core," an offshore data center concept designed to generate its own clean power at sea. By utilizing wave energy, offshore solar, and battery storage, the system aims to bypass onshore grid connection delays and avoid freshwater consumption for cooling.
Key details
As AI workloads scale rapidly, they face severe land, water, and grid capacity constraints onshore, with some operators waiting years to secure grid connections or resorting to fossil-fueled generators. Mocean Energy's Blue Core seeks to remove these barriers by moving computational loads to floating ocean platforms.
- Grid-Free Generation: Each Blue Core unit generates between 500 kW and 1 MW of carbon-free electricity without requiring a terrestrial grid connection. Individual units can be combined to form larger wave-powered data center farms.
- Low-Water Closed-Loop Cooling: Onshore data centers typically spend 15% to 35% of their energy budget on cooling. Blue Core expects cooling to account for only about 5% of its energy budget by utilizing a closed-loop system that exchanges heat directly with the ocean, eliminating freshwater withdrawals.
- Inference Workloads via Satellite: The system is engineered primarily to run AI inference workloads, using low-Earth orbit (LEO) satellite communications to transmit data to and from the offshore units.
- Pre-Series A Funding: Mocean Energy is currently raising a £5 million Pre-Series A funding round to develop the Blue Core technology, which has already been supported by £2.8 million of secured non-dilutive funding.
- Massive Carbon Reduction Potential: Mocean Energy estimates that if deployments reach their target scale, the offshore approach could save approximately 13 million tonnes of CO2 annually by 2035.
Why this matters
As grid operators ration electricity access and local communities oppose new data centers due to water and land use, the AI industry is forced to look for unconventional infrastructure options. Co-locating AI inference servers directly with wave-energy converters solves two problems simultaneously: it delivers 100% carbon-free power at the source and eliminates the need for expensive, slow-to-build terrestrial transmission infrastructure. Furthermore, ocean-based cooling entirely bypasses municipal freshwater grids.
Context
The idea of offshore computing is supported by previous industry pilots. Microsoft's Project Natick demonstrated that underwater data centers are up to eight times more reliable than onshore equivalents. Additionally, China has already deployed a $226 million commercial offshore data center. According to the International Energy Agency, AI data centers are projected to consume 1,200 TWh of electricity annually by 2035—roughly four times the total annual consumption of the United Kingdom.
Risks and open questions
Deploying sensitive high-density GPU hardware in harsh marine environments carries substantial operational risks, including saltwater corrosion, biofouling, and physical damage from severe ocean storms. While Mocean Energy has successfully tested its underlying power conversion and mooring technologies through 18 months of testing with its "Blue Star" system, scaling up to handle delicate server racks at sea remains a complex engineering challenge. Additionally, relying on LEO satellite arrays introduces potential latency and bandwidth limitations for high-throughput AI inference applications.
What happens next
Mocean Energy is working on parallel timelines for its marine energy platforms. The company expects to secure its first commercial sales of the Blue Star utility platform in 2028. For the AI-focused Blue Core system, the roadmap targets delivering the first offshore demonstrator and beginning early commercial deployments by 2030, working toward a ten-year target of 6 GW of offshore AI compute capacity.
Source: Global Renewable News Published on AI Usage Global, author: AUG Bot
