Why Tech Giants Are Betting Billions on Liquid Cooling for AI Data Centers
Liquid cooling is becoming a critical edge in AI data centers, as firms like Trane acquire specialists and Texas races to host 6.5 gigawatts of new.
175 articles
Liquid cooling is becoming a critical edge in AI data centers, as firms like Trane acquire specialists and Texas races to host 6.5 gigawatts of new.
Data center emissions surged 32 times in one year, driven by the AI infrastructure boom as companies race to build massive new facilities worldwide.
Choosing the wrong AI model size wastes energy; one large model can emit 284 metric tons of CO2, making AI energy efficiency a critical business decision.
Apple's on-device AI strategy keeps data center power demands lower, with capital spending at $2.46B versus rivals committing over $100B each.
Oklo moved from groundbreaking to nuclear startup authorization in just 10 months, signaling that regulatory speed may no longer be the bottleneck for AI.
Aalo Atomics and Crusoe Energy will pair a small nuclear reactor with an AI data center at Idaho National Laboratory in 2027, a first-of-its-kind.
AI data centers are driving U.S. electricity demand to record highs, and Goldman Sachs says power, not capital, is now the key bottleneck for AI growth.
Ontario is investing $500,000 to convert a former coal plant into a nuclear industrial hub that could supply clean steam to heavy manufacturers.
Tech giants are building their own power plants to fuel AI data centers, with Meta investing C$13 billion in a 1 GW Canadian campus paired with a 932 MW.
A $60 million federal initiative is using AI to cut nuclear reactor development timelines, with 32 partners racing to make advanced reactors commercially.
GlobalFoundries secured a $300M federal award to replace copper wiring with silicon photonics in data centers, promising a 5x energy efficiency gain for.
Alberta wants to power AI data centers with nuclear energy, but has almost no nuclear workforce to build or operate the reactors needed.
Smart cooling fans and AI energy tools are helping data centers meet EU ErP 2026 rules, with enterprise AI sustainability programs unlocking up to $10.
Japan's distributed AI grid experiment will route data center workloads to regions with surplus power, potentially easing strain on grids powering.
A $20 million NSF initiative will deploy AI-powered self-driving labs nationwide, promising to accelerate chemical discovery by 100 times using robotics.
Nvidia is in talks to back $250 billion in financing for a 10-gigawatt Ohio data center, raising alarms about circular AI financing risks.
Antares Nuclear closed a $470 million round to deploy small modular reactors at U.S. military bases, targeting the Pentagon as AI drives surging energy.
AI data center power demand will hit 132 GW by 2026, spurring startups to build on-site generators, battery storage, and model compression tools to close.
AI data centers are adopting a gas-to-nuclear bridge strategy, pairing immediate natural gas power with future advanced reactors to meet surging AI energy.
AI data center construction is pushing industrial stocks to 30x earnings, matching tech valuations, as companies like Caterpillar surge over 50% this year.
Canada's CANDU reactors and SMR plans make it a threshold nuclear weapons power, with experts saying only political will, not technical capability, stands.
China's AI-powered nuclear system ADANES aims to solve the "black box" problem, but can opaque AI really be trusted to run reactors safely?
ABB and NVIDIA are partnering to build 800V data center power systems capable of handling 1-megawatt AI server racks as capacity races toward 220.
Big Tech's nuclear power deals are mostly options to buy electricity, not real commitments; nuclear costs three times more per unit than solar or wind.
The U.S. DOE's $5 billion Genesis Mission uses AI to accelerate scientific discovery in energy, nuclear waste, and water systems across national labs.
AI data centers caused a 75.5% power cost spike in the U.S.'s largest grid region, exposing cracks in the White House ratepayer protection pledge.
AI data center chips are projected to surge from $16.51 billion to $50.91 billion by 2035, as GPU demand and custom silicon reshape the semiconductor.
MIT researchers are developing nuclear plant automation without AI, using transparent logic systems that make every decision fully traceable for safer SMR.
India plans to invest $200 billion to expand nuclear capacity from 8.78 GW to 100 GW by 2047, with five small modular reactors targeted by 2033 and.
Massachusetts's MGHPCC became the first university research data center to earn LEED Platinum, proving AI computing can be powerful and green.
UC Berkeley's AI framework pairs language models with 500 million quantum calculations to predict better catalysts, cutting energy use in chemical.
Westinghouse just secured an $80 billion nuclear deal backed by DOE loans, using AI to fix the construction chaos that bankrupted it in 2017.
Space data centers could ease AI's power crisis, but orbital facilities cost $51B per gigawatt versus $16B on Earth, and key barriers remain unsolved.
A Rice University study finds underground saline aquifers could store up to 90% of U.S. AI data center carbon emissions as power demand quadruples by 2030.
AI is being used to manage nuclear waste and cut corporate carbon emissions, potentially saving over $150 billion in cleanup costs.
Fuel cells beat nuclear for AI data centers because they deliver 100 megawatts in 90 days; SMRs won't be ready until 2030.
GPU pulse loads from AI clusters are overwhelming power grids; DG Matrix and Skeleton Technologies partnered to fix data center electrical infrastructure.
Nuclear stocks surge as the Trump administration backs a $200M initiative pairing AI data centers with advanced reactors, plus a potential Saudi nuclear.
Nvidia's new Vera CPU claims 50% better AI performance than x86 chips, but its 450-watt power draw poses a major challenge for data centers.
Power companies can seize private land for AI data center transmission lines using eminent domain, but 75 projects were blocked in early 2026.
Japan's giraffe-shaped nuclear robot has safely inspected radioactive reactor welds since 1995, absorbing doses lethal to humans across 50 completed.
AI's energy hunger is driving a reckoning with 20 breakthrough technologies, from green hydrogen to fusion, that could power sustainable AI infrastructure.
Storage efficiency is AI's overlooked energy lever; Western Digital just won three sustainability awards for proving smarter storage cuts power use at.
Russia's Shelf-M small modular reactor has cleared a key regulatory hurdle, with construction set for 2027 and a 2030 startup to power remote sites and AI.
AI data centers can raise neighborhood air temperatures by up to 4 degrees Fahrenheit, and over 140 local groups have blocked $60 billion in new projects.
AI data center power bills are surging as GPU racks hit 80 kilowatts and a single training cluster can consume as much electricity as 50,000 homes.
The UN is urging a shift to "frugal AI," right-sizing models to cut energy costs and keep AI from derailing global climate and development goals.
New Jersey signed a law to procure 1,100 MW of nuclear power, including small modular reactors, as AI data centers strain the grid.
Tennessee's $11 million investment in TRISO nuclear fuel production could power 55 small modular reactors, delivering 4.5 gigawatts of clean energy for AI.
Small engineering tweaks, not breakthroughs, will solve AI's energy crisis, say award-winning Colorado State researchers studying data center efficiency.