newatlas.com via Reddit

China Comms Opens World-First Undersea Data Center

ai infrastructure china ai climate ai-infrastructure data-center renewable-energy

Key insights

  • Phase one runs at 2.3 MW across 192 server racks, with planned capacity expansion to 24 MW.
  • The circulating copper-pipe heat exchange system cuts cooling electricity by 22.8% and eliminates freshwater dependency entirely.
  • Tsinghua University Professor Li Zhen projects the approach could save about 50 billion kWh of electricity each year nationally.

Why this matters

Undersea data center siting simultaneously addresses land scarcity, freshwater dependency, and cooling energy consumption: three compounding constraints on AI infrastructure expansion. The 22.8% cooling electricity reduction and complete elimination of freshwater use mean the efficiency case does not rest on a single variable, making the model more credible as it scales toward the planned 24 MW. Professor Li Zhen's Tsinghua University estimate of about 50 billion kWh saved annually nationwide, if validated at scale, would represent one of the largest single reductions in AI infrastructure's energy footprint globally.

Summary

A China Communications Construction subsidiary switched on the world's first underwater data center off Shanghai in late May 2026. Using a circulating copper-pipe heat exchange system, it draws cooling from seawater, cutting electricity by 22.8% and eliminating freshwater dependency. Phase one: 2.3 MW across 192 server racks, scaling to 24 MW. Offshore wind covers approximately 95% of power. Essentially: (China Communications Construction) bets ocean-floor siting solves land, water, and heat constraints choking data center growth. - Cooling electricity cut 22.8%; land use down over 90% vs. surface facilities - Tsinghua Professor Li Zhen: nationwide rollout could save about 50 billion kWh of electricity each year At 24 MW scale, ocean proximity could become a structural input to AI infrastructure planning.

Potential risks and opportunities

Risks

  • Continuously releasing heat into local marine environments could trigger regulatory review or remediation orders from Chinese environmental agencies as the facility scales from 2.3 MW toward its planned 24 MW
  • Scale-up from a 2.3 MW proof-of-concept to 24 MW may surface structural or sealing failures in the copper-pipe system not detectable at phase one's limited deployment density
  • Nations lacking equivalent coastal geography and offshore wind build-out face a structural cost disadvantage if China replicates this model at national scale

Opportunities

  • Offshore wind developers and marine construction contractors gain a new primary customer segment as data center operators globally evaluate undersea siting feasibility
  • Subsea infrastructure specialists and copper-pipe heat exchange system suppliers are positioned for first-mover contracts as the China Communications Construction model expands toward 24 MW
  • Hyperscalers operating in water-stressed or landlocked regions have strategic incentive to fund or license undersea siting technology before Chinese-developed approaches establish global IP precedent

What we don't know yet

  • Long-term seabed durability of the copper-pipe heat exchange system: no maintenance schedule or inspection timeline disclosed in reporting
  • Ecological impact from continuously releasing heat into local marine environments: no environmental assessment or mitigation plan timeline cited
  • Whether the approximately 95% offshore wind figure reflects currently delivered grid supply or contracted future capacity: sourcing mechanism not specified