Explanation: In summary, immersion cooling reduces CapEx on building infrastructure by simplifying the data hall design. The elimination of raised floors, containment, and many air handling units can cut construction and fit-out costs. Any added structural cost to support liquid tanks is usually minor compared to the air-cooling infrastructure removed. These upfront savings contribute to immersion cooling’s 20–35% lower CapEx compared to traditional designs, as observed by OCP’s data (Meeting Data Center Cooling Demands with Immersion Cooling Fluids: A Future-Ready Solution | Data Center Frontier).

Air Cooling Equipment:

A 1 MW air-cooled data center typically relies on a chiller plant (or large HVAC units) and air distribution hardware. This may include one or more chillers (for chilled water systems) plus pumps and piping, or DX (direct expansion) CRAC units with condensers. In high-density builds, a water-cooled chiller with cooling towers is common. Additionally, several CRAH/CRAC units are installed in the server room for air circulation. All these components represent significant capital cost. Schneider Electric’s analysis of a 2 MW, 10 kW/rack facility provides a reference: the air-cooled design cost about $7.02 per watt, largely driven by chillers, CRAHs, and associated electrical gear (Liquid vs. Air Cooling. Which is the Capex winner?).

‍Immersion Cooling Equipment:
Immersion setups replace most of that air hardware with liquid cooling components. Key cost items are the immersion tanks or enclosures that hold the servers and dielectric fluid, the coolant distribution units (CDUs) with pumps and heat exchangers, and the dielectric coolant fluid itself (a one-time fill for the tanks). There is also heat rejection equipment: typically dry coolers (air-cooled radiators) or plate heat exchangers connected to a warm-water loop. Notably, immersion designs do not require large chillers or CRAH units – those are removed from the design (Liquid vs. Air Cooling. Which is the Capex winner?). For example, Schneider’s study notes: going to liquid cooling “you remove the chillers, you remove the CRAHs… but you add dry coolers… additional pumps and piping…, and a premium for the liquid cooling technology (sealed chassis, dielectric fluid, etc.)” (Liquid vs. Air Cooling. Which is the Capex winner?). In other words, the cost shifts to the specialized immersion hardware and fluid. The dielectric fluids can be expensive per volume, but they are long-lived assets (one vendor notes their coolant “most likely will not need replacing over a typical data center lifecycle” (), meaning the fluid investment lasts many years).

Cost Balance:

In many cases, the net CapEx for immersion cooling is comparable to or lower than air cooling when the design is optimized. The OCP Regional Summit data suggests immersion can achieve 20–35% CapEx savings overall (Meeting Data Center Cooling Demands with Immersion Cooling Fluids: A Future-Ready Solution | Data Center Frontier).
This comes from avoiding the costly chillers, cooling towers, and large CRAC units, and also from downsizing electrical infrastructure (since cooling load and server fan load are lower, power systems like UPS and switchgear can be smaller (Liquid vs. Air Cooling. Which is the Capex winner?). Any added cost for tanks and fluid is offset by these savings. In Schneider Electric’s detailed TCO study, at equal density (10 kW/rack) the capital costs for air vs. liquid were nearly equal (~$7.0 per watt each) (Liquid vs. Air Cooling. Which is the Capex winner?). However, as rack density increases, liquid cooling shows clear CapEx advantages: e.g. at 20 kW/rack, they saw ~10% CapEx savings, and at 40 kW/rack, ~14% savings versus air cooling (Liquid vs. Air Cooling. Which is the Capex winner?). This is because higher-density immersion reduces the quantity of racks, PDUs, and building space needed, multiplying the savings in infrastructure and equipment. Table 2 summarizes the major cooling system components and their presence in each approach.

Note: The elimination of large mechanical cooling systems in immersion designs not only saves on equipment costs but also simplifies the data center, potentially accelerating deployment. Immersion cooling systems are often modular – vendors offer pre-fabricated tank systems that can be installed quickly without the extensive plumbing of traditional chillers and CRAHs.

Traditional air-cooled data centers typically achieve PUE values around 1.3–1.5 (meaning 30–50% extra power beyond the IT load is consumed by cooling, lighting, etc.), although very efficient designs can reach ~1.2. In contrast, immersion cooling significantly cuts the cooling overhead – partial PUE as low as 1.03 has been demonstrated (Meeting Data Center Cooling Demands with Immersion Cooling Fluids: A Future-Ready Solution | Data Center Frontier) (). This implies only a 3% overhead beyond IT power, an almost “bare-bones” level of energy waste. For a 1 MW IT load, an air-cooled facility with PUE 1.3 draws about 1.3 MW total, meaning 300 kW is used for cooling and infrastructure. An immersion-cooled facility might draw ~1.05 MW total at PUE ~1.05, with only 50 kW for cooling. The energy savings ~250 kW under peak load translates to roughly 2 million kWh saved annually (and lower electrical bills accordingly). In percentage terms, OCP data indicates immersion cooling can reduce overall data center energy (OpEx) by 40–50% compared to air (Meeting Data Center Cooling Demands with Immersion Cooling Fluids: A Future-Ready Solution | Data Center Frontier). A major contributor is the elimination of server fans and air movers – fans in a typical server consume about 5–10% of the server’s power (Economics of Immersion Cooling for Bitcoin Miners | Braiins). Immersion cooling removes those fans entirely, directly returning that power budget to useful computing or saving it. One immersion vendor notes that without server fans and with more efficient heat transfer, the entire immersion cooling system uses less electricity than the air-cooling fans alone (How Immersion Cooling Helps Reduce Operational Costs in Data Centers).
Additionally, immersion cooling can run at higher coolant temperatures (since electronics can operate safely in warmer liquid as long as it is below component limits). This enables more free cooling (using outside ambient air or fluid cooling without compressors) for most of the year, whereas air-cooled systems often need chillers or compressor assistance to maintain lower air temperatures.
The result is further efficiency gain. In summary, immersion cooling drastically improves PUE – e.g. one study shows advanced rear-door air cooling hit PUE ~1.2–1.3, while single-phase immersion achieved ~1.03 in the same setting (). These efficiency improvements directly reduce OpEx, as less power is drawn for cooling per month.

Routine Maintenance Tasks:

Air-cooled sites require regular maintenance of CRAC/CRAH units (changing filters, cleaning coils, checking fans/blowers and coolant levels for CRACs). Chillers and cooling towers need substantial upkeep – e.g. chiller compressors need servicing, refrigerant checks, water treatment for cooling towers, etc.
The server hardware in air cooling also faces dust accumulation, requiring periodic cleaning of server internals, and fan replacements when they fail. Immersion cooling greatly simplifies O&M. With no air handlers, there are far fewer filters or fans to service. The cooling loop in immersion has pumps and heat exchangers, but these are simpler mechanical systems with fewer failure points than compressor-based chillers. Single-phase immersion in particular is described as having “only three moving parts” (pumps), making it extremely simple and reliable (How Immersion Cooling Helps Reduce Operational Costs in Data Centers) (How Immersion Cooling Helps Reduce Operational Costs in Data Centers).
There are no air filters to replace and no dust ingress to clean out, since IT components are sealed in fluid. The dielectric fluid itself may require occasional quality monitoring (ensuring no contamination), but it does not get consumed or need frequent replacement (). Overall, the labor hours for cooling infrastructure maintenance are lower for immersion.

Equipment Lifespan and Reliability:

Air-cooled equipment is exposed to temperature swings, humidity changes, and contaminants (dust, pollutants). Fans and moving parts create vibration. These factors can shorten hardware lifespan or cause more frequent failures. Immersion cooling creates a very stable environment for IT hardware – components are not exposed to oxygen or dust, temperatures are uniform, and vibrations are dampened (no fan motors running). As a result, servers in immersion tend to last longer and fail less often, reducing replacement costs and downtime. Industry observations back this: immersion “maintains stable operating conditions” and “minimizes wear and tear” on hardware (Meeting Data Center Cooling Demands with Immersion Cooling Fluids: A Future-Ready Solution | Data Center Frontier). A blog by GRC similarly notes that with immersion, servers “last longer and require maintenance less often” (How Immersion Cooling Helps Reduce Operational Costs in Data Centers) due to the reduced risk of dust, corrosion, and thermal cycling. Even the supporting cooling equipment benefits – for example, no high-speed fans to burn out, and pumps in liquid cooling often have long MTBF (mean time between failures) since they run at moderate speeds in a clean fluid loop.

Failure Rates and Downtime:

In air cooling, common failure modes include fan failures (both in IT servers and CRAC units), leaks in CRAH coil or piping, or overheating if airflow is blocked. Immersion eliminates many of these failure modes (no IT fans, no air-cooled evaporator coils). The simpler system means fewer emergency repairs. This translates to higher uptime and potentially lower costs from unplanned outages. While hard to quantify for TCO, it’s a qualitative operational advantage.

Skill and Management:

Operating an immersion-cooled data center may require training staff on handling dielectric fluids and different procedures (like safely removing a server from a liquid tank for service).
However, these procedures are generally straightforward and can be integrated into normal operations. Many immersion systems have remote monitoring and automation to assist (e.g. fluid temperature, pump status alerts) (How Immersion Cooling Helps Reduce Operational Costs in Data Centers). On balance, the reduced complexity of infrastructure often outweighs any new skills needed, resulting in lower overall O&M effort and cost.

Air Cooling Rack Density:

Air-cooled racks are typically limited by cooling and air distribution constraints. Standard designs handle on the order of 5–15 kW per rack (some high-density air setups can go ~20–30 kW with special cooling like rear-door heat exchangers or in-row coolers, but this is the exception). To deploy 1 MW of IT with, say, 10 kW per rack, one would need about 100 racks of servers, plus aisle space between them and floor space for cooling units. Even with 20 kW racks, it’s ~50 racks for 1 MW. This consumes a large data hall area.

Immersion Rack Density:

Immersion cooling can handle extreme densities per tank/rack. It’s common to see 50–100 kW of IT load in a single immersion tank (some designs even target >100 kW per tank in HPC scenarios) (Immersion cooling systems: Advantages and deployment strategies for AI and HPC data centers) (). For example, one analysis notes that single-phase immersion can “easily cool 100+ kW per rack” whereas even advanced air (rear-door coolers) is cost-effective only up to ~15 kW/rack (). This means to deploy 1 MW, perhaps only 10–20 tanks are needed. Those tanks can be arranged densely, and no cold aisle/hot aisle spacing is needed – often tanks are placed back-to-back or in modular pods. The net result is a smaller data hall. One study by Schneider Electric showed that by doubling rack density from 10 to 20 kW with liquid cooling, they saved on “core & shell, rack, PDU, and containment” costs, yielding ~10% CapEx savings (Liquid vs. Air Cooling. Which is the Capex winner?). At 4× density (40 kW vs 10 kW per rack), savings grew to ~14% (Liquid vs. Air Cooling. Which is the Capex winner?). These savings partly come from needing fewer racks and a smaller building footprint.

Impact on Building Size:

‍A smaller footprint can translate to lower construction costs (less square footage to build out, and potentially a smaller land requirement). It may also reduce ongoing facility costs like lighting, security, and even lease costs if renting space. For example, instead of an entire large room, a 1 MW immersion system might fit in a compact area or container. Some immersion solutions are deployable in ISO shipping containers, illustrating how space-efficient they are ().

Scalability:

‍High density means future expansion of IT load is easier – one can add more servers to existing tanks or add tanks without immediately needing new building space. In contrast, an air-cooled facility might run out of cooling capacity or floor space at a lower IT load, forcing an earlier build of an expansion. This gives immersion an advantage in scaling, indirectly affecting long-term TCO (delaying or avoiding CapEx for expansion).

By contrast, immersion cooling can be designed to run with minimal or zero water consumption. Most immersion systems reject heat via dry coolers (air-cooled radiators) or closed-loop water-to-air heat exchangers, which do not consume water like cooling towers do. Any water used is in a closed loop, so operational water loss is near zero. This yields a huge reduction in Water Usage Effectiveness (WUE). Studies have found that immersion cooling can reduce water use by as much as 91% relative to conventional cooling (Infosys Knowledge Institute | Data centers look to immersion cooling as a path to sustainability – and lower costs). Essentially, the only water needed in an immersion setup might be for very occasional topping up of a closed loop or for facility safety systems – no continuous evaporation.

For operators in regions facing water scarcity or high water costs, this is a significant OpEx and sustainability benefit. It’s worth noting that some modern air-cooled designs also try to minimize water (using air-cooled chillers), but then they often pay a penalty in energy efficiency. Immersion lets you achieve high efficiency and low water usage simultaneously by using warm liquid cooling and eliminating the need for evaporative assist.

Energy & Carbon Footprint:

The improved energy efficiency (lower PUE) of immersion cooling directly translates to a lower carbon footprint for the data center (assuming a typical grid mix). Using ~30–40% less electricity for cooling means fewer emissions over time, which is increasingly important as companies strive for sustainability. OCP’s immersion initiative highlights that reducing energy and water is as much about sustainability as cost (Meeting Data Center Cooling Demands with Immersion Cooling Fluids: A Future-Ready Solution | Data Center Frontier). Immersion also avoids certain environmentally harmful substances: for example, traditional HVAC systems rely on refrigerants (HFCs, etc.) that can have high global warming potential (GWP). With immersion, large chillers and AC units are gone, and thus dependence on refrigerants is greatly reduced. This helps future-proof the facility against tightening regulations on refrigerants () and eliminates the risk of refrigerant leaks. The dielectric fluids used for immersion are generally stable and non-toxic; newer formulations are moving toward more eco-friendly blends (e.g. hydrocarbon-based fluids without PFAS chemicals) (Meeting Data Center Cooling Demands with Immersion Cooling Fluids: A Future-Ready Solution | Data Center Frontier). And since the fluid is not consumed or emitted, it doesn’t create ongoing environmental waste (eventual disposal or recycling of the fluid would happen only after many years).

Immersion cooling, on the other hand, outputs heat in a concentrated liquid form (hot water or coolant). The temperature of the coolant leaving the racks can be quite high – often in the range of 50–60°C (for single-phase immersion, the fluid can be allowed to warm to these levels safely). This temperature is much more suitable for direct heat reuse. For example, district heating systems typically operate with water temperatures of 60–70°C, which immersion-cooled data centers can help supply with minimal boost. Indeed, immersion technology can enable recovery of nearly all the server waste heat into a usable water stream – one project highlighted that using liquid cooling makes it feasible “to capture 100% of data center heat” for recycling into other uses (Data Center Heat: 4 Creative Ways to Use Rejected Heat).

District Heating

In cooler climates, an immersion-cooled facility can pipe its hot water into the municipal district heating network to warm nearby homes and offices. This offsets the use of boilers or power plants for heat, improving overall energy utilization. (Some Nordic data centers have pursued this model, citing almost 90–95% heat recovery with immersion systems feeding city heating loops.)

Agriculture (Greenhouses)

As a creative example, an immersion-cooled modular data center in the Netherlands uses its 65°C coolant output to heat an adjacent greenhouse for tomato farming (Data Center Heat: 4 Creative Ways to Use Rejected Heat).
The servers’ waste heat thus directly supports food production. Such synergy is easier with liquid cooling because you get a reliable hot water supply.

Aquaculture or Industrial Processes

A U.S. data center project used immersion cooling waste heat to support a fish farm (maintaining water temperature for breeding fish) (Data Center Heat: 4 Creative Ways to Use Rejected Heat) (Data Center Heat: 4 Creative Ways to Use Rejected Heat). Similarly, any industry that needs low-grade heat (e.g. pre-heating water for manufacturing, absorption cooling processes, etc.) could utilize the heat. The fact that immersion cooling “keeps the heat stored within a reusable liquid” makes it far superior to air cooling for these purposes (Data Center Heat: 4 Creative Ways to Use Rejected Heat).

On-site Facility Heating

A data center could use its immersion heat to provide hot water or space heating for its own campus or nearby buildings (like using it for office heating or hot water supply), reducing the need for separate boilers.

From a TCO perspective, waste heat reuse can create additional value or savings. If the heat is sold to a district heating company or used to offset heating fuel that the data center owner would otherwise buy, it effectively monetizes the waste heat. While this may not be common in all deployments, it’s a growing consideration as sustainability and energy reuse become important. Even if not directly counted in TCO, the potential for heat reuse is a strategic benefit of immersion cooling that air cooling cannot match easily. It can improve the overall ROI of the cooling solution when implemented. In short, immersion cooling turns what is traditionally a waste by-product (heat) into an opportunity – up to 99% of the heat can be recaptured with the right setup (Excess Heat Transformation with W.E District & RISE - Submer), significantly increasing the energy efficiency at a holistic level.

Initial CapEx

Immersion cooling can be installed at cost parity with air cooling (or lower, especially at high densities), as discussed. Any remaining premium at day 1 (in case of lower-density deployments or fluid costs) is generally small relative to total project cost (recall Schneider found only a $0.04/W difference at 2 MW scale for like-for-like density (Liquid vs. Air Cooling. Which is the Capex winner?). In many cases, smart design can eliminate a CapEx premium entirely.

Energy Cost Savings

Year after year, immersion cooling saves a significant percentage of energy. With perhaps ~25% or more reduction in facility power draw, the electricity cost savings compound. For example, if a 1 MW IT data center spends $0.10 per kWh, a PUE improvement from 1.30 to 1.05 would save around $200,000 every year in power bills (as earlier calculated). Over 5–10 years, that is $1–2 million saved in energy alone, directly improving TCO.

Maintenance and Operations Savings

Reduced maintenance translates to lower annual O&M budget. Fewer part replacements and less HVAC service can save tens of thousands per year in a 1 MW facility. Also, if hardware lasts longer, the capital refresh cycle can be extended – perhaps servers can run an extra year or two before replacement, meaning deferred CapEx on IT gear. Over a 10-year span, skipping even one refresh cycle or reducing hardware failure rates can noticeably improve total cost.

Resource and Utility Savings

If the site is in a location with expensive water or carbon taxes/penalties, the water savings (up to 90%) and efficiency help avoid those costs. Some regions incentivize efficiency through rebates or lower utility rates; an immersion-cooled site might qualify for energy efficiency incentives or sustainability grants (indeed, some utilities have given grants for using immersion cooling due to its efficiency (How Immersion Cooling Helps Reduce Operational Costs in Data Centers).

Overall TCO Reduction

The Open Compute Project’s immersion cooling working group has modeled these factors and found an impressive net benefit. According to OCP data, immersion cooling can yield TCO savings on the order of 30–40% over traditional air cooling (Meeting Data Center Cooling Demands with Immersion Cooling Fluids: A Future-Ready Solution | Data Center Frontier). This figure encapsulates both reduced CapEx and ongoing OpEx savings across a multi-year period. In practice, this could mean that if an air-cooled 1 MW data center cost $X million to build and $Y million to run over 5–10 years, an equivalent immersion-cooled facility might cost only ~$0.6–0.7X + 0.6–0.7Y$ over the same period (30–40% less total).

It’s important to note that exact savings will depend on the use case and implementation. For instance, extremely low power cost might make energy savings less monetarily dramatic, or a very low-density deployment might not capitalize on all of immersion’s strengths. But in scenarios where high density, efficiency, and sustainability are needed (which is increasingly the case for modern AI and cloud workloads), immersion cooling shines financially.

Long-Term Outlook

In 5–10 years, the gap is likely to widen further in favor of immersion. As liquid cooling technology matures and scales, costs for tanks and fluids are expected to come down, while air cooling will struggle with rising server densities and potential water and refrigerant constraints. Many experts now see liquid cooling not as a cost liability but as a cost advantage. Schneider Electric concluded that “capex shouldn’t be an obstacle” to liquid cooling adoption, and in fact additional savings can be expected as manufacturing scales (Liquid vs. Air Cooling. Which is the Capex winner?). The same applies to OpEx: future high-density processors will make air cooling even less efficient, whereas immersion will handle them with relatively small increases in cooling overhead.