Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Fix the driver behind crashes, sound loss and screen glitches3Repair Windows errors before they cause bigger problemsLiquid cooling can reduce a data center’s modeled lifecycle greenhouse-gas emissions by about 15% to 21% compared with air cooling. But Microsoft’s modeled comparison found a much larger reduction—about 85% to 90%—when the electricity supply was changed to 100% renewable energy. The figures describe different interventions: cooling changes how heat is removed; cleaner electricity changes the emissions tied to powering the facility.
What Microsoft’s “one-fifth” finding actually measures
A peer-reviewed 2025 study in Nature compared four cooling approaches: conventional air cooling, direct-to-chip cold plates, one-phase immersion, and two-phase immersion. Across its modeled lifecycle comparison, the liquid-cooled options reduced greenhouse-gas emissions by approximately 15% to 21%, energy demand by 15% to 20%, and blue-water consumption by 31% to 52% relative to air cooling. The upper end—not a guaranteed result for every system—is the basis for “about one-fifth.” The study and its methods cover equipment and materials as well as operation.
Microsoft’s summary gives a simpler figure for cold plates: roughly 15% lower lifecycle emissions and energy demand, and 30% to 50% lower water consumption. These are modeled averages, not measurements showing that one Microsoft facility cut its emissions by a fixed amount after a retrofit. The paper’s results depend on assumptions about equipment, capacity, electricity, and lifecycle inputs.
Emissions, energy demand, and water are separate measures. A 15% energy reduction does not automatically mean a 15% emissions reduction: the carbon intensity of the electricity matters. Blue-water consumption refers to water consumed from surface water or groundwater; it is not a synonym for every form of water withdrawal or the full water footprint of construction and supply chains.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstall#1 Best Overall
How the cooling systems differ
Air cooling
Fans move air through servers and heat is carried away by facility equipment. It is mature and familiar, but moving enough air to cool high-power processors takes energy and can constrain rack density. Some facilities also use evaporative cooling, which can consume water.
Direct-to-chip cold plates
A cold plate is a metal heat exchanger attached to a processor such as a CPU or GPU. Coolant circulates through it, absorbs heat close to the source, and transfers that heat to facility cooling equipment. Unlike immersion, it does not submerge the server. Other components may still need air cooling, and the facility needs compatible plumbing and coolant-distribution equipment.
One-phase immersion
The server is submerged in dielectric fluid that stays liquid. Pumps or natural convection carry heat to a heat exchanger. The approach can cool the whole server and reduce or eliminate server fans, but it requires tanks, fluid handling, and a different service routine from conventional rack systems.
Two-phase immersion
Servers sit in a low-boiling-point dielectric fluid. Heat vaporizes the fluid; vapor rises to a condenser, returns to liquid, and cycles again. The method can transfer heat effectively, but it is more specialized. Many two-phase fluids have fluorinated chemistries, raising concerns about regulation, handling, leakage, and end-of-life treatment. That concern applies particularly to many two-phase fluids, not to all liquid cooling.
Recommended Free Tools
Why electricity decarbonization has the larger carbon effect
Cooling is only one part of a data center’s electricity use. Servers, accelerators, networking, storage, and cooling equipment all draw power, and electricity use during operation is a major emissions driver in the study’s modeled systems. Microsoft reports that its modeled 100%-renewable electricity scenario reduced greenhouse-gas emissions by roughly 85% to 90%, regardless of cooling technology. That is far greater than the modeled 15% to 21% lifecycle reduction from changing cooling systems.
“100% renewable” in a model is not proof that a particular facility receives renewable electricity in every hour. Operators can pursue clean electricity through on-site generation, power-purchase agreements, grid contracts, renewable-energy certificates or other environmental attributes, batteries, and siting decisions. Those approaches differ in whether new generation is added, whether it is geographically connected to the facility, and whether clean supply matches demand by hour. Annual matching can balance a year’s consumption on paper while the facility still draws a fossil-heavy grid mix at some times.
Rank #3
Microsoft says it has pursued 100% additional renewable-energy matching for its facilities and aims to match electricity use with carbon-free resources continuously by 2030. Its sustainability reporting gives different procurement figures for different reporting contexts: the company reported contracting 19 GW of new renewable energy across 16 countries in 2024, while later 2025 report materials describe 34 GW across 24 countries. Those are not interchangeable snapshots. Microsoft’s 2025 report announcement and its sustainability reporting provide the company’s context and definitions.
Why liquid cooling is becoming important for AI
High-power AI accelerators concentrate heat in dense server racks. Liquid transfers heat more effectively than air, so cooling closer to the processor can help operators support more computing capacity in a given rack or floor area and reduce fan demand. The Nature paper says liquid cooling’s heat-transfer capability can enable greater server and virtual-machine capacity. That is an infrastructure advantage, not a guarantee that total electricity use will fall: adding more compute can outweigh efficiency gains.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Microsoft has also described new AI-focused data-center designs that use chip-level liquid cooling without evaporating water during cooling. The company estimates that, under its stated assumptions, the design could avoid more than 125 million liters of water per facility each year. It said pilot projects in Phoenix and Mount Pleasant were expected to begin coming online in late 2027. This is a company estimate about new designs, not evidence that the existing fleet is water-free. Microsoft’s description of the design explains the claim and its intended deployment.
Rank #4
Water savings can come with an energy trade-off
Eliminating evaporative cooling does not necessarily eliminate every environmental cost. Microsoft says its zero-water-evaporation design may use a nominally higher amount of annual energy than evaporative designs because mechanical cooling replaces some evaporative cooling. Higher-temperature operation and efficient economizing chillers are intended to limit that increase.
The practical balance depends on the facility. In a water-stressed region, avoiding evaporation may be valuable even if mechanical cooling uses somewhat more electricity. On a carbon-intensive grid, that extra electricity can weaken the climate benefit. Closed-loop liquid cooling is not automatically waterless across its full lifecycle: water can still be involved in electricity generation, equipment manufacturing, construction, and coolant production.
Choosing between cold plates and immersion
| Approach | Where it can fit | Operational and environmental considerations |
|---|---|---|
| Air cooling | Existing facilities and lower-density workloads | Mature and familiar, but fan and air-handling demands can constrain dense AI racks; evaporative heat rejection may consume water. |
| Direct-to-chip cold plates | New AI servers and mixed workloads where processors need targeted cooling | Does not submerge the server and can use less coolant than full immersion. Requires compatible hardware, plumbing, and coolant-distribution units; components beyond the processors may remain air-cooled. |
| One-phase immersion | High-density or specialized deployments able to adopt tank-based servicing | Cools the whole server and can reduce fan use. Tanks, fluid compatibility, contamination control, and more involved maintenance are important. |
| Two-phase immersion | Specialized high-density deployments with suitable fluid and service arrangements | Strong heat transfer, but fluid chemistry, regulatory exposure, leakage, and disposal need careful review. Microsoft previously reported 5% to 15% lower power consumption for a given server in its two-phase immersion testing; that company result is not a universal guarantee. Microsoft’s account of that testing provides the context. |
For an operator weighing a deployment, the relevant comparison is not simply “liquid or air.” It is the complete cost and performance of a system for a particular site and workload. A retrofit may require extensive plumbing and equipment changes, while a new build can incorporate liquid loops from the start. The buying case should include capital equipment, installation, pumps, controls, facility heat rejection, commissioning, maintenance, coolant replacement, and disposal—not a single equipment price.
Best Value
- Data Center Coolant
- 25% Inhibited Propylene Glycol
- JeffCool ISF 25
- High thermal conductivity
What to check before investing
- Thermal density: Determine whether the target rack load can be handled economically by air or requires chip-level liquid cooling.
- Water and climate: Assess local water stress and the cooling system’s water consumption, not just its energy efficiency.
- Electricity: Review local grid carbon intensity and the credibility, timing, and location of any clean-power procurement.
- Compatibility and service: Confirm server warranties, component coverage, technician training, leak detection, and recovery procedures for pump or connector failures.
- Fluid lifecycle: Request evidence on fluid composition, replenishment, recycling, and end-of-life handling, particularly for immersion systems.
- Total system impact: Compare emissions and water per unit of useful computing as well as per rack or facility; a more efficient system can still increase total consumption if it enables much more compute.
Why the percentages are not universal guarantees
The study is broad, but it is a lifecycle model rather than a field trial of every possible facility. Its comparison includes servers, chips, racks, buildings, support equipment, electricity, cooling fluids, transportation, manufacturing, and end-of-life treatment. Results are normalized around computing capacity, so a comparison per rack or per server could tell a different story.
The paper also identifies data limits: some fluid inputs were estimated, one-phase fluid production could not be modeled precisely with available data, cold-plate coolant additives were not fully represented, and some two-phase fluid information came from manufacturers and was confidential. Most authors were Microsoft researchers, and some inputs came from manufacturers. Those facts call for attention to assumptions and attribution; they do not by themselves negate peer review or the reported results.
There is also a scale problem. Microsoft reported that its total Scope 1, 2, and 3 emissions were 23.4% above its 2020 baseline while energy use had risen 168% over the same period. The company’s figures illustrate why efficiency improvements do not guarantee falling absolute emissions when infrastructure and demand expand. They are company-wide reporting figures, not an estimate of emissions caused only by data-center cooling. Microsoft’s sustainability report announcement describes that reporting context.
How to put cooling in a climate plan
For organizations building or operating AI infrastructure, liquid cooling is best treated as one part of a sequence: avoid unnecessary compute, improve workload and hardware efficiency, procure credible low-carbon electricity, choose cooling suited to the rack and local water conditions, and account for materials, fluids, construction, and disposal. Clean electricity generally has the larger modeled effect on emissions; liquid cooling can make dense computing practical while reducing cooling energy and water use in many designs. Neither intervention substitutes for the other.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
