Answers are grounded in this paper and cite their sources.
Randomized trial evaluates energy sustainability and water use in AI infrastructure, suggesting innovative solutions.
America's AI build-out is running into three walls: electricity, cooling, and water. Data centers already use about 4 percent of U.S. electricity and could use up to 12 percent by 2028, and most new campuses are being built in water-stressed regions where evaporative cooling consumes millions of cubic meters of freshwater a year. Project POLARIS proposes building the next generation of AI campuses on cold northern coastlines, designed as integrated energy systems: small modular reactors and hydropower supply carbon-free electricity; the ocean's deep, cold water absorbs waste heat through sealed closed-loop heat exchangers, consuming no freshwater; and before reaching the ocean, that heat warms homes, greenhouses, and fish farms. Every component operates commercially today - deep-water district cooling in Toronto since 2004, fjord-cooled data centers in Norway, Arctic renewable data centers in Iceland, small modular reactors under construction in North America. The paper develops the reference architecture and thermal-electrical balances for a 500 MW campus, proposes a U.S.-Denmark-Greenland flagship demonstration aligned with the three governments' active 2026 economic discussions, and a domestic companion track under the U.S. Department of Energy's AI-infrastructure-on-federal-lands program. The recommended first step is a joint feasibility study and twelve months of oceanographic data at candidate sites.
No takes yet. Share an insight, caveat, or question.
Ernest Newton Sumrall (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: