The impact of free cooling on chiller service revenue is more medium- to long-term and limited in magnitude
AI summary card
The impact of free cooling on chiller service revenue is more medium- to long-term and limited in magnitude
Bernstein believes free cooling will reduce compressor wear in data center chillers, mainly affecting repair labor and parts revenue. However, there is no clear near-term impact, and by after 2030 it may create an annualized 1%-2% headwind to data center service revenue.
- Chiller service revenue can be broken into maintenance, modernization upgrades, repairs, parts, and additional services; the parts truly affected by lower compressor utilization are mainly repair labor and related parts.
- The report estimates that repair labor plus related parts account for about 25%-30% of the service-revenue pool, but this does not mean the entire portion disappears; rather, lower repair frequency creates a revenue headwind.
- The near-term impact is limited because newly installed chillers typically need to run for 5-10 years before enough wear accumulates; pressure from the current build cycle is more likely to show up after 2030.
- The medium-term impact is estimated at roughly a 1%-2% annual drag on data center service revenue; if data centers account for 40% of the business and services account for 50% of data center revenue, that implies about a 20-40 bp drag on companywide revenue.
- Bernstein keeps TT, JCI, and VRT at Outperform and CARR at Market-perform; the market is currently more focused on equipment sales than on service-revenue risk.
Report interpretation
Overview
This report is the third installment in Bernstein's data center chiller series, and its core question is whether free cooling will erode chiller service revenue. The report argues that free cooling reduces compressor utilization, thereby lowering wear and failure probability, so it does affect repair-related revenue. However, maintenance contracts, modernization upgrades, and most additional services are unlikely to be impacted to the same degree. The overall conclusion is that the impact exists, but it is not a near-term risk and is small relative to companywide revenue.
Core views
The report's main views are as follows: first, free cooling primarily lowers compressor operating load, thereby reducing failure and repair frequency; second, only repair labor and related parts in chiller service revenue are truly exposed to this risk, accounting for about 25%-30% of the service-revenue pool; third, because repair revenue usually rises meaningfully only later in the equipment lifecycle, the current wave of data center chiller installations is more likely to pressure service revenue after 2030; fourth, the medium-term drag on data center service revenue is about 1%-2% annually, which translates to roughly 20-40 bps of companywide revenue; fifth, original equipment manufacturers are not especially worried right now, and the market remains focused on equipment sales and data center order growth.
Analysis framework
The report first breaks down chiller service revenue composition, then explains the technical pathways for free cooling, including air-side economizers, water-side economizers, and refrigerant migration. It then assesses repair-revenue risk through a framework based on compressor wear and failure probability, and combines chiller life cycles of roughly 25 years, data center chiller installation timing, and the aging window around 2030 to estimate the effect on service revenue and companywide revenue.
Methodology notes
Failure risk rises nonlinearly with usage intensity
The report argues that a linear failure model does not reflect real equipment wear. A more realistic Weibull distribution means failure probability is lower in early usage stages and becomes more concentrated later; free cooling lowers compressor utilization, keeping it in a lower-failure-risk zone for longer.
Split chiller service revenue into maintenance, upgrades, repairs, parts, and additional services
The report divides chiller service revenue into roughly 30% maintenance labor, 20% upgrade/modernization labor, 15% repair labor, 30% parts, and 5% additional services, and concludes that repair labor and some parts are the most exposed to free cooling.
Assess revenue drag over the next 5 years, 5-15 years, and beyond 15 years
The report believes service revenue will be largely unaffected over the next 5 years; in the 5-15 year period, as early data center chillers enter a higher-repair phase, there may be an annual 1%-2% headwind to data center service revenue; the long-term impact eventually stabilizes or becomes marginal.
Asset mapping & comparison
Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).
- Trane Technologies PLC (TT)Coverage name for data center chillers and related services
- Strengths
- Rated Outperform, target price US$550; the report sees data center demand and the service-revenue base as still attractive.
- Weaknesses
- Both chiller sales and services may face marginal drags from free cooling, air-cooling substitution, or other cooling-technology shifts.
- Comparison
- Compared with CARR, the report is more constructive on TT; compared with VRT, TT has more direct exposure to chiller equipment and service economics.
- Risks
- Escalation of pricing-manipulation lawsuits, competitors catching up in CDU or broader liquid-cooling innovation, further deterioration in residential or transportation businesses, and chiller sales and service headwinds exceeding the liquid-cooling tailwind.
- Carrier Global Corporation (CARR)Coverage name for chillers and HVAC-related businesses
- Strengths
- Target price US$75; upside could come from an improvement in the U.S. residential and light-commercial cycle, continued data center growth, and better heat-pump policy in the U.K. and Germany.
- Weaknesses
- Rated Market-perform, indicating a more balanced risk/reward; chiller service revenue in data centers may still be affected by free cooling in the midterm.
- Comparison
- Compared with TT, JCI, and VRT, CARR has the lowest rating in this report, at Market-perform.
- Risks
- Escalation of pricing-manipulation lawsuits, slower capex from hyperscale cloud data centers, tighter-than-expected R-410A supply, or a lengthening cycle.
- Johnson Controls International PLC (JCI)Coverage name for data center cooling and building equipment
- Strengths
- Rated Outperform, target price US$176; data center demand and operational improvement remain potential supports.
- Weaknesses
- Chiller-related service revenue may face reduced repair revenue after 2030 due to free cooling.
- Comparison
- Also an Outperform name like TT, but JCI is additionally affected by its own lean transformation and operating leverage realization.
- Risks
- Failure to effectively embed the lean transformation across the organization, weaker data center business due to chiller headwinds, and operating leverage below management's 50% guidance.
- Vertiv Holdings Co (VRT)Coverage name for data center power and thermal management, indirectly affected by cooling-demand changes
- Strengths
- Rated Outperform, target price US$416; highly tied to data center expansion, liquid cooling, and high-power computing infrastructure demand.
- Weaknesses
- If significantly better cooling efficiency reduces the amount of cooling required per unit, some growth logic could be weakened.
- Comparison
- Unlike TT, CARR, and JCI, VRT is not a direct bearer of traditional chiller service-revenue risk, but it is highly linked to the data center thermal-management investment cycle.
- Risks
- Cooling-efficiency gains reduce required cooling volume, unexpected slowdown in data center expansion, faster-than-expected migration from NVIDIA chips to custom silicon, and pressure if it cannot respond quickly to shifts such as DTC or 800 VDC.
Key data
- Service revenue relative to equipment capexAbout 8-10x in nominal terms; about 4-5x after inflation adjustmentTrane has previously cited that chiller service, parts, and modernization revenue can reach 8-10x the original capex.
- Maintenance labor as a share of the service-revenue poolAbout 30%Mainly preventive maintenance and scheduled service agreements; the report believes this is not heavily affected by free cooling.
- Upgrade/modernization labor as a share of the service-revenue poolAbout 20%Typically occurs once or twice over a chiller's roughly 25-year life cycle, and the report believes it is basically unaffected by lower compressor utilization.
- Repair labor as a share of the service-revenue poolAbout 15%Related to equipment operating issues and failures, and is the core exposure to free cooling risk.
- Parts revenue as a share of the service-revenue poolAbout 30%Parts are more tied to repairs and upgrades, so part of this segment will be affected by lower repair frequency.
- Additional services as a share of the service-revenue poolAbout 5%Includes software monitoring and similar offerings, and the share is relatively small.
- Service revenue pool exposed to riskAbout 25%-30%Mainly repair labor plus related parts; not all of this will disappear.
- Medium-term drag on data center service revenueAbout 1%-2%/yearThis mainly shows up after 2030, once early installed chillers enter a higher-repair-demand phase.
- Drag on companywide revenueAbout 20-40 bpsAssumes data centers account for 40% of the business and services account for 50% of data center revenue in 2030.
- Base-case data center chiller market size2026E about US$7.9B; 2030E base case about US$16.6BThe base case implies about 20% CAGR.
Impact & implications
From an investment perspective, free cooling is more of a medium- to long-term dilution factor for the data center chiller service-revenue pool than a near-term shock to orders or equipment sales. The report remains broadly constructive on the covered companies because data center orders are still growing strongly, and the service-revenue drag has only a limited impact when translated into companywide revenue. More important differentiators are each company's data center exposure, service-revenue mix, opportunities in liquid cooling and CDU products, and differences in free-cooling availability across geographies.
Risks
- Free cooling penetrates more than expected in low-temperature or specific geographic regions, causing a larger-than-estimated drag on repair revenue.
- The shift from chillers to dry coolers, air cooling, or other architectures happens faster than expected, affecting equipment sales and service revenue.
- Repair demand for components other than compressors is weaker than expected, making the report's assumption about residual repair revenue too optimistic.
- Capex from hyperscale cloud data centers slows, weakening demand for chillers and related equipment.
- Large differences in electricity, water resources, and climate conditions across regions could cause free-cooling economics and equipment choices to deviate from the base-case view.
What to watch
- Actual repair rates and service-revenue changes for the early data center chiller installation cohorts around 2030.
- Actual free-cooling usage duration and compressor run hours in different climate regions such as NoVa and Phoenix.
- OEM disclosures or management guidance changes for service revenue, parts revenue, and repair attach rates.
- Adoption speed of air cooling, water-side economizers, refrigerant migration, dry coolers, and liquid cooling solutions.
- Whether data center order growth, chiller equipment sales growth, and service-revenue mix continue to support the valuations of TT, JCI, CARR, and VRT.