Shanghai’s Lin-gang Special Area on Oct. 21 announced the launch of what it calls the world’s first underwater data center connected to offshore wind power, combining seabed cooling with renewable electricity aimed at supporting artificial intelligence workloads. The project, developed by HiCloud with a total investment of 1.6 billion yuan, targets a green electricity supply rate exceeding 90% and a power usage effectiveness (PUE) below 1.15, according to Lin-gang materials. The facility follows a separate underwater intelligent computing cluster activated earlier this year off Hainan province, adding to a growing set of subsea computing trials highlighted in official and local-government coverage as AI-related electricity demand draws scrutiny.
Wind-Powered Seabed Computing Comes Online
The Lin-gang project sits within the broader Shanghai Pilot Free Trade Zone and draws its electricity from nearby offshore wind farms rather than the conventional grid. According to the Lin-gang government’s launch announcement, the facility achieves 22.8% energy savings, 100% water savings, and more than 90% land savings compared to traditional onshore data centers. Those figures reflect the core advantage of submerging server racks: seawater can reduce the need for conventional chillers and cooling towers, while the ocean floor can reduce land requirements compared with traditional onshore data centers.
The project’s Phase 1 demonstration operates at 2.3 megawatts, a modest starting point against a planned full cluster of 24 megawatts, according to the HiCloud project profile hosted on the Lin-gang investment portal. A green electricity supply rate exceeding 90% is the stated target, with the PUE goal set below 1.15. For context, the average PUE for large-scale data centers worldwide is often cited in the 1.55 to 1.60 range, meaning the Lin-gang facility aims to waste far less energy on non-computing overhead. Whether those targets hold up under real operating conditions, particularly during periods of low wind generation, has not yet been independently verified or benchmarked against comparable offshore projects.
Hainan’s Underwater AI Cluster Expands
The Lin-gang launch did not happen in isolation. Earlier this year, China activated an underwater intelligent computing cluster off Lingshui, Hainan, by deploying a new seabed module connected to an existing underwater data center that first went operational in 2023. That Hainan facility was designed to handle AI-specific workloads, and its expansion suggests the original installation performed well enough to justify scaling up. Together, the two projects show subsea computing being tested in different coastal conditions, including different water temperatures and deployment approaches, as China expands “intelligent computing” capacity in official coverage.
The Hainan cluster’s tropical location offers warmer water than Shanghai, which could reduce cooling efficiency but also tests the technology’s operational range. If seabed data centers only work in cold northern waters, their commercial appeal narrows considerably. By running parallel experiments in subtropical and temperate zones, Chinese developers are gathering operational data that competitors, including companies that previously trialed subsea pods, have not publicly matched at this scale. Official coverage on the central government’s English portal frames the Hainan deployment as part of a broader push to build “intelligent computing” capacity, indicating that underwater modules are being integrated into national AI infrastructure rather than treated as isolated engineering demonstrations.
Why Renewable Coupling Changes the Equation
Most coverage of underwater data centers focuses on the cooling advantage, but the more significant engineering choice in the Lin-gang project is the direct connection to offshore wind. Traditional data centers buy renewable energy credits or sign power purchase agreements with distant wind and solar farms, which means the electrons they actually consume still come from the same fossil-heavy grid as everyone else. By physically linking turbines to submerged server pods through dedicated cables, the Lin-gang facility claims to use more than 95% renewable electricity, as stated on the Lin-gang Special Area site. That direct coupling, if it proves reliable, sidesteps the accounting practices that let many “green” data centers claim carbon neutrality without materially changing grid emissions.
The tradeoff is resilience. Offshore wind is intermittent, and AI training workloads demand near-constant uptime. The project’s promotional materials do not address how the facility handles periods when wind generation drops below demand, whether through onshore grid backup, local battery storage, or workload scheduling that shifts computation to match wind availability. This gap matters because the entire value proposition rests on the renewable link: if the facility quietly draws grid power during calm weather, its environmental claims weaken. Over time, independent monitoring of electricity sourcing and hourly generation profiles would be needed to confirm the 95% renewable figure and to understand whether the system prioritizes carbon reduction or uptime when those goals conflict.
Scale, Cost, and Unanswered Questions
The 1.6 billion yuan total investment covers the full 24-megawatt buildout, not just the initial 2.3-megawatt demonstration phase. That price tag is steep for a facility whose capacity would rank as mid-sized by land-based standards, where a single hyperscale campus can exceed 100 megawatts. The premium reflects the engineering complexity of waterproofing, pressure-proofing, and maintaining equipment that sits on the ocean floor, where a failed hard drive cannot be swapped in minutes by a technician walking down an aisle. Underwater maintenance requires specialized vessels, divers, or remotely operated vehicles, all of which add cost and downtime risk, and developers have yet to disclose detailed lifecycle maintenance schedules or replacement strategies.
Neither the Lin-gang nor Hainan announcements linked above include a public environmental impact assessment or third-party audit. The claimed savings figures, while specific, originate entirely from project developers and local government promotional materials. For a technology pitched as an environmental solution, the absence of independent verification is a notable gap. Key unknowns include how much waste heat is discharged into surrounding waters, potential effects on marine ecosystems over years of operation, and the lifecycle carbon cost of manufacturing and deploying sealed subsea modules compared with upgrading efficient land-based facilities.
China’s Bet on Submerged AI Infrastructure
Despite these uncertainties, the twin projects in Shanghai and Hainan suggest that China views underwater data centers as more than experimental curiosities. By aligning subsea modules with offshore wind farms and positioning them within zones such as the Lin-gang Special Area, authorities are effectively treating seabed computing as a strategic tool to reconcile AI expansion with energy and land constraints. Official write-ups in government-facing channels, accessible through the national English search service, emphasize intelligent computing, green development, and digital infrastructure in the same breath, indicating that submerged clusters are being woven into broader policy narratives about high-quality growth and low-carbon industry.
Whether this approach scales will depend on economics, regulation, and public acceptance as much as on engineering. If Lin-gang can demonstrate stable operations, genuine near-100% use of offshore wind, and manageable maintenance costs, it may become a template for coastal provinces with strong renewable resources but limited land for new data centers. Conversely, if hidden grid reliance, ecological concerns, or high downtime emerge, subsea modules could remain niche, reserved for specific high-value workloads or constrained locations. For now, the Lin-gang and Hainan projects mark an inflection point: instead of merely cooling servers with seawater, China is testing whether the ocean itself can host and power the next generation of AI infrastructure, with results that will be closely watched by data center operators and policymakers worldwide.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
