Home / Knowledge Center / Immersion Cooling: Single-Phase vs Two-Phase

Immersion Cooling for Bitcoin Mining: Single-Phase vs Two-Phase Systems for ASIC Hosting

Air cooling has been the default thermal management strategy for Bitcoin mining since the earliest ASIC deployments. But as chip power densities climb past 40 watts per square centimeter and next-generation miners push individual unit consumption above 5,000 watts, air cooling is hitting hard physical limits. Immersion cooling -- submerging ASIC hardware directly in dielectric fluid -- solves this by removing heat at the source with dramatically higher efficiency.

The question for mining operators and ASIC hosting providers is no longer whether to adopt immersion cooling, but which type to deploy. Single-phase and two-phase immersion systems use fundamentally different thermodynamic mechanisms, carry different capital costs, and suit different operational profiles. This guide breaks down both approaches so you can make an informed decision for your facility.

Why Immersion Cooling Matters for Bitcoin Mining Operations

Before comparing the two systems, it is worth understanding why immersion cooling has moved from niche experimentation to mainstream consideration across the mining industry.

The Limits of Air Cooling for Modern ASICs

Traditional air-cooled mining facilities rely on high-velocity fans to push ambient air across heat sinks attached to ASIC hashboards. This approach works reasonably well up to a point, but it introduces several compounding problems at scale:

  • Fan failures account for the majority of ASIC downtime. In air-cooled deployments, fans are typically the first component to fail, often within 12 to 18 months of continuous operation. Each fan failure triggers thermal throttling or shutdown, reducing hashrate.
  • Dust and particulate ingress degrade hashboard performance. Even with filtration, fine particles accumulate on heat sinks and circuit boards, reducing thermal conductivity and creating hot spots that accelerate component degradation.
  • Ambient temperature sensitivity limits site selection. Air-cooled facilities in hot climates -- including the Middle East, West Texas, and parts of Southeast Asia -- face significant performance penalties during summer months unless they invest heavily in evaporative cooling or mechanical chillers.
  • Noise levels restrict site options. A single Antminer S21 produces approximately 75 dB at full speed. A facility running thousands of units generates industrial noise levels that create permitting challenges and limit proximity to populated areas.

Immersion cooling eliminates every one of these problems. Fans are removed entirely. Hashboards are sealed from airborne contaminants. Heat rejection is decoupled from ambient air temperature. And noise drops to near-zero because the loudest component -- the fan -- no longer exists.

Operational Benefits Beyond Thermal Management

Beyond solving the direct cooling problem, immersion enables operational advantages that compound over the life of the hardware:

  • Extended hardware lifespan: Operators consistently report that immersion-cooled ASICs last two to three times longer than air-cooled equivalents, primarily due to the elimination of thermal cycling and mechanical vibration from fans.
  • Overclocking headroom: With more effective heat removal, operators can push ASICs 20 to 50 percent above stock hashrate while maintaining safe junction temperatures. This directly increases revenue per unit.
  • Higher rack density: Immersion tanks support significantly more compute per square meter than air-cooled racks, reducing facility footprint and real estate costs.
  • Reduced maintenance labor: Without fans to replace, filters to change, or dust to clean, routine maintenance frequency drops substantially.

How Single-Phase Immersion Cooling Works

In a single-phase immersion system, ASIC hashboards are submerged in a tank filled with dielectric fluid that remains liquid throughout the entire cooling cycle. The fluid never boils. Heat generated by the chips transfers into the surrounding liquid through convection, and the warmed fluid is circulated -- either by natural convection or by pumps -- to an external heat exchanger where the heat is rejected to a secondary loop (typically water or glycol) connected to dry coolers or cooling towers.

Coolant Options for Single-Phase Systems

The most common single-phase coolants for Bitcoin mining applications include:

  • Engineered dielectric hydrocarbons such as BitCool from Engineered Fluids, which are purpose-designed for electronics immersion with controlled viscosity, flash point, and thermal conductivity.
  • Synthetic mineral oils offered by Shell (S5 X series) and other industrial suppliers, providing good thermal performance at moderate cost.
  • White mineral oil used by some operators as a lower-cost alternative, though with less optimized thermal properties and higher viscosity at lower temperatures.

A typical single-phase fluid costs between $5 and $15 per liter, and a standard immersion tank holding 20 to 40 ASIC units may require 500 to 1,500 liters of coolant depending on tank geometry.

Strengths of Single-Phase Immersion for Mining

  • Lower upfront cost: Tank designs are simpler. No condensers, no vapor management, no pressurized enclosures. A basic single-phase tank with external heat exchanger and pumps can be deployed for approximately $500 to $1,500 per kilowatt of IT load, depending on scale and vendor.
  • Easier maintenance: Because the fluid stays liquid, operators can open tanks, add or remove hardware, and perform repairs without specialized procedures. Splashes are manageable -- the fluid is non-toxic and non-conductive.
  • Proven at mining scale: Companies like Wattum, DCX Immersion, and LiquidCool Solutions have deployed single-phase immersion for Bitcoin mining at multi-megawatt scale. The operational track record is mature.
  • Simpler facility integration: Single-phase systems use standard plumbing components (pumps, plate heat exchangers, PVC or copper piping) and integrate into existing colocation infrastructure without exotic materials.

Limitations to Consider

  • Lower heat flux capacity: Single-phase convective cooling is inherently less efficient per unit area than phase-change (boiling) cooling. For current-generation ASICs this is not a bottleneck, but as chip densities continue climbing it may become one.
  • Pump energy overhead: Circulating dense dielectric fluid requires pumps that consume approximately 3 to 8 percent of the IT load as parasitic power, depending on flow design and fluid viscosity.
  • Fluid degradation over time: Hydrocarbon-based fluids can slowly degrade through oxidation, particularly if exposed to air. Periodic fluid testing and top-ups are necessary.

How Two-Phase Immersion Cooling Works

Two-phase immersion uses a fundamentally different thermodynamic principle. The dielectric fluid has a low boiling point -- typically between 34 and 61 degrees Celsius -- engineered to match chip operating temperatures. When the ASIC chip heats the surrounding fluid past its boiling point, the fluid vaporizes at the chip surface. This phase change absorbs a large amount of latent heat. The vapor rises to a condenser (usually a coil array at the top of the tank), where it reverts to liquid form and drips back down to the components.

This self-circulating cycle requires no pumps for the primary coolant loop. The physics of evaporation and condensation handle fluid movement automatically.

Coolant Options for Two-Phase Systems

Two-phase systems require specialized fluorocarbon-based fluids with precisely engineered boiling points:

  • 3M Novec fluids (particularly Novec 7100 and Novec 649) have been the industry standard, offering boiling points around 49 and 61 degrees Celsius respectively. Note that 3M announced it would exit PFAS manufacturing, creating supply uncertainty for these fluids.
  • Solvay Galden HT series provides fluorinated polyether fluids with a range of boiling points, serving as an alternative to Novec products.
  • Chemours Opteon and other emerging fluorocarbon alternatives are entering the market to address 3M's withdrawal.

Two-phase fluids are substantially more expensive than single-phase alternatives -- typically $50 to $300 per liter, depending on the specific chemistry. A tank of comparable capacity to a single-phase deployment may cost 10 to 30 times more in coolant alone.

Strengths of Two-Phase Immersion for Mining

  • Superior heat flux removal: Phase-change cooling removes heat at rates of 10 to 100 watts per square centimeter at the chip surface, far exceeding what convective liquid cooling achieves. This provides headroom for next-generation ASICs with even higher power densities.
  • Near-zero pump energy: The primary loop is self-circulating through natural convection of vapor and liquid. This eliminates the parasitic pump power that single-phase systems require.
  • Tighter temperature control: Because the boiling point is fixed by the fluid chemistry, chip temperatures are inherently regulated. Hot spots are virtually eliminated -- any point that exceeds the boiling point triggers immediate local cooling through vaporization.
  • Maximum overclocking potential: The aggressive heat removal enables operators to push ASICs to their absolute electrical limits, potentially extracting 40 to 50 percent more hashrate from a given unit.

Limitations to Consider

  • Dramatically higher coolant cost: Fluorocarbon fluids are the single largest cost driver. For a deployment of 500 ASICs, coolant costs alone can exceed $250,000 to $500,000, compared to $15,000 to $50,000 for a single-phase equivalent.
  • Vapor containment complexity: Two-phase tanks must be sealed or semi-sealed to prevent coolant vapor from escaping. Lost vapor means lost coolant -- at $50 to $300 per liter, even small leaks are expensive.
  • Regulatory and environmental uncertainty: Many two-phase fluids fall under PFAS regulations that are tightening globally. The EU PFAS restriction proposal, if enacted, could limit availability of some fluorocarbon coolants.
  • Maintenance complexity: Accessing submerged hardware requires more careful procedures to manage vapor, prevent fluid contamination, and minimize coolant loss during tank openings.
  • Fewer mining-scale deployments: While two-phase immersion is well-proven in high-performance computing and defense applications, large-scale Bitcoin mining deployments using two-phase are still relatively uncommon compared to single-phase.

Single-Phase vs Two-Phase: Head-to-Head Comparison

The following table summarizes the key decision factors for mining operators evaluating both approaches:

Factor Single-Phase Two-Phase
Cooling mechanism Convective heat transfer (liquid stays liquid) Latent heat of vaporization (liquid boils to vapor)
Coolant cost per liter $5 - $15 $50 - $300
Tank cost per kW (approx.) $500 - $1,500 $1,500 - $4,000+
Parasitic power (pumps) 3% - 8% of IT load Near zero (self-circulating)
Heat removal capacity Adequate for current ASICs (up to ~25 W/cm2) Exceeds current ASIC needs (up to 100+ W/cm2)
Overclocking potential 20% - 30% above stock 30% - 50% above stock
Maintenance difficulty Low -- open tank, swap hardware Moderate -- vapor management, sealed access
Noise reduction Near-silent (pump hum only) Near-silent (condenser fans only)
Environmental risk Low -- hydrocarbons are biodegradable Higher -- PFAS regulatory scrutiny
Mining-scale track record Extensive -- multi-MW deployments operational Limited -- mostly sub-MW and pilot programs
Best suited for 1 MW+ mining farms, hot climates, fleet longevity Ultra-high density, R&D, future-proofing

Which System Should You Choose for Your Mining Operation?

The right choice depends on your operational priorities, scale, and investment horizon.

Choose Single-Phase Immersion If:

  • You are deploying at 1 MW or larger scale and need proven, cost-effective cooling that works today.
  • Your primary goals are hardware longevity, noise reduction, and moderate overclocking gains.
  • You operate in a hot climate where air-cooled hosting is inefficient or impossible without mechanical chillers.
  • Your maintenance team needs to access and swap hardware frequently without specialized training.
  • You want to minimize coolant cost as a percentage of total deployment capital.

Choose Two-Phase Immersion If:

  • You are building a next-generation facility designed to handle ASICs that do not yet exist but will push power densities beyond what single-phase can efficiently manage.
  • Maximizing hashrate per unit through aggressive overclocking is your primary competitive advantage.
  • You have the capital budget to absorb significantly higher upfront fluid costs in exchange for lower long-term parasitic power consumption.
  • You are running a smaller, high-value deployment (under 500 units) where the per-unit economics of premium cooling can be justified.

The Hybrid Approach

Some forward-thinking operators deploy single-phase immersion for the bulk of their fleet while maintaining a smaller two-phase test environment for overclocking experiments and next-generation hardware evaluation. This hedges against future cooling requirements without committing the entire operation to higher-cost infrastructure.

Immersion Cooling ROI Considerations for Mining Operators

Calculating the return on investment for immersion cooling requires looking beyond the initial capital expenditure to capture the full economic picture:

  • Hardware replacement savings: If immersion extends ASIC lifespan from 2 years to 5 or more years, the avoided replacement cost of several thousand dollars per unit compounds significantly across a large fleet.
  • Overclocking revenue: A 25 percent hashrate increase on a fleet of 1,000 miners effectively gives you 250 additional miners worth of output at zero additional hardware cost.
  • Reduced cooling energy: Immersion systems typically achieve PUE (Power Usage Effectiveness) values between 1.02 and 1.10, compared to 1.15 to 1.40 for well-designed air-cooled facilities. In a 10 MW deployment, even a 0.1 PUE improvement saves approximately 1 MW of cooling power continuously.
  • Real estate efficiency: Higher density per square meter means smaller buildings, shorter cable runs, and less land -- all of which reduce capital and operating costs.
  • Noise-related site flexibility: Near-silent operation opens site options that would be impossible with air cooling, potentially accessing cheaper land or more favorable power agreements.

For most commercial mining operations above 1 MW, single-phase immersion cooling reaches positive ROI within 18 to 30 months compared to air cooling, driven primarily by reduced hardware failures and overclocking gains. Two-phase systems have a longer payback period -- typically 30 to 48 months -- but may deliver superior total returns over a 5-year horizon if the operation sustains aggressive overclocking throughout.

Deploying Immersion Cooling with a Hosting Partner

Building immersion cooling infrastructure in-house requires specialized engineering expertise in fluid dynamics, heat exchanger sizing, electrical integration, and environmental containment. For operators who want the benefits of immersion cooling without the facility engineering burden, partnering with a managed ASIC hosting provider that offers immersion-cooled capacity is the most practical path.

When evaluating hosting partners for immersion-cooled deployments, prioritize:

  • Demonstrated immersion experience: Ask for operational history, not just marketing materials. How many units are currently running in immersion? What is the measured failure rate versus air-cooled fleets?
  • Fluid management protocols: Professional operators test coolant quality regularly, maintain fluid reserves, and have documented procedures for fluid replacement and disposal.
  • Overclocking support: Immersion without overclocking captures only part of the value. Confirm that the hosting partner supports and monitors overclocked firmware.
  • Transparent SLAs: Uptime guarantees, power delivery commitments, and clearly defined responsibilities for hardware maintenance should all be in writing.

Rax Data and Energy provides enterprise-grade colocation and managed hosting infrastructure engineered for high-density compute workloads, including both Bitcoin mining and AI/GPU deployments. Contact our team to discuss cooling infrastructure requirements for your deployment.

Frequently Asked Questions

Is immersion cooling worth it for Bitcoin mining?

Yes, for operations running at scale. Immersion cooling eliminates fans (the most failure-prone ASIC component), extends hardware lifespan by two to three times, reduces noise to near-zero, and enables overclocking that can increase hashrate by 20 to 50 percent. The capital investment is higher than air cooling, but total cost of ownership over a three to five year period is typically lower for deployments above approximately 1 MW.

What is the difference between single-phase and two-phase immersion cooling?

In single-phase immersion cooling, the dielectric fluid remains liquid throughout the cooling process. Heat is absorbed by the fluid and removed via external heat exchangers. In two-phase immersion cooling, the fluid boils at the chip surface and the vapor rises to a condenser where it returns to liquid form. Two-phase systems remove more heat per unit volume but are significantly more complex and expensive.

What coolant is used for immersion cooling Bitcoin miners?

Single-phase systems typically use engineered dielectric fluids such as mineral oil derivatives, synthetic hydrocarbons, or purpose-built fluids from manufacturers like Engineered Fluids (BitCool) and Shell (S5 X). Two-phase systems use fluorocarbon-based fluids such as 3M Novec or Solvay Galden, which have low boiling points that enable phase change at chip operating temperatures.

How much does immersion cooling reduce ASIC failure rates?

Industry operators report that immersion cooling reduces ASIC failure rates by approximately 50 to 90 percent compared to air-cooled deployments. The primary mechanism is eliminating fan failures, dust ingress, and thermal cycling stress. Hashboards submerged in dielectric fluid are protected from humidity, corrosion, and particulate contamination, which are leading causes of premature failure in air-cooled environments.

Can I convert existing air-cooled ASICs to immersion cooling?

Yes. Most modern ASIC miners from Bitmain, MicroBT, and Canaan can be converted to immersion cooling by removing fans and heat sinks, then submerging the boards in a compatible immersion tank. Some operators also modify firmware to remove fan speed alerts. Conversion adds approximately $200 to $500 per unit in preparation costs, plus the cost of the tank and coolant infrastructure.

Ready to Upgrade Your Mining Infrastructure?

Our engineering team can help you evaluate immersion cooling options for your ASIC fleet -- whether you are planning a new deployment or retrofitting an existing facility.

Get a Custom Cooling Assessment