CoolIT Systems

1.1 Identity, Headquarters, and Business Model

CoolIT Systems is a Canadian liquid-cooling company headquartered in Calgary, Alberta that now describes itself as fully focused on AI and high-performance compute infrastructure. The company’s current business model is modular direct liquid cooling hardware: custom processor coldplates, server-level coldplate loops, rack manifolds, technology cooling system piping, and coolant distribution units. These components are sold into OEM server platforms, hyperscaler data-center programs, and frontier HPC clusters rather than consumer retrofit markets. That positioning matters because it places CoolIT in the thermal bottleneck of the AI supply chain: at the point where GPUs, CPUs, and networking gear exceed the practical limits of air cooling. Public materials consistently frame CoolIT as a co-development partner for semiconductor companies, OEMs, and cloud-scale operators, which makes the company more comparable to infrastructure suppliers than to niche thermal-component vendors.[CO001, CO002, CO003, CO004, CO005, CO006]

1.2 Leadership, Governance, and Disclosure Gaps

CoolIT’s 2026 public materials provide a clear view of the current operating team but only a partial view of governance. Jason Waxman leads the company as CEO after senior Intel leadership roles and a stint as President of Fluke; Patrick McGinn oversees operations and commercial execution as President and COO; Kamal Mostafavi leads technology strategy as CTO; David Meffe, Ken Lau, and Paul O’Connor add commercial, ecosystem, and supply-chain depth from Intel, industrial technology, and procurement-heavy environments. This is a seasoned go-to-market and scaling team for an AI infrastructure supplier. The main governance gap is that current website materials do not clearly identify the original founders or the board roster. For diligence, that means the operating bench looks strong, but founder-market-fit analysis and board-control assessment remain incomplete without management materials or transaction documents beyond the public sale disclosures.[CO011, CO012, CO013, CO014, CO015, CO016]

1.3 Ownership History, Funding Context, and 2026 Exit

The most important capital-market fact about CoolIT is that it is no longer an independent venture-backed target as of the run date. KKR acquired the business in 2023 through Global Impact Fund II alongside Mubadala, then sold it to Ecolab in a transaction announced in March 2026 and later closed in July 2026. Ecolab disclosed unusually helpful public economics for a private infrastructure company: approximately $550 million of next-twelve-month sales at signing and valuation multiples of about 29x next-twelve-month adjusted EBITDA and 24x 2027 adjusted EBITDA. KKR separately disclosed an approximately 15x return and highlighted a broad-based employee ownership payout. Together these disclosures imply that CoolIT transitioned from a private-growth industrial technology company to a strategic infrastructure asset whose value rested on AI-driven liquid-cooling demand, manufacturing scale, and recurring service/cross-sell potential inside a larger water-and-chemistry platform.[CO020, CO021, CO022, CO023, CO024, CO025]

1.4 Operating Scale, Manufacturing Footprint, and Customer Reach

Public evidence suggests CoolIT reached meaningful industrial scale before the Ecolab close. KKR said the company expanded manufacturing to more than 300,000 square feet and increased CDU capacity 25x after 2023. A separate CoolIT manufacturing article said the new Starfield facility in Calgary spans 112,000 square feet and that Starfield plus the LiquidLab facilities represent nearly 150,000 square feet of dedicated manufacturing and R&D space. The same materials describe production and support coverage across more than 80 countries and installations at more than 300 data centers worldwide. That mix matters strategically: many liquid-cooling startups have compelling component designs but limited qualification capacity or field-service reach, whereas CoolIT appears to have built the deployment, test, and support infrastructure needed for hyperscaler and OEM programs. The public sales estimate from Ecolab supports the view that this was not merely a science project by 2026.[CO006, CO007, CO010, CO029, CO030, CO031]

1.5 Milestones, Customer Proof, and Adverse Signals

CoolIT’s public milestone record shows a company that migrated steadily up the compute stack: from gaming-PC liquid cooling in 2001, to patented graphics-card coldplates in 2009, to data-center direct liquid cooling in 2012, to OEM integrations with HPE and Dell in the mid-2010s, and then to exascale and AI-class systems in the 2020s. The customer-proof surface is broad rather than perfectly quantified, with case references spanning Frontier, Durham/DiRAC COSMA, atNorth, DTU, GWDG, Fraunhofer, and Frontera. The main adverse caveat is not lack of demand but disclosure opacity and transaction risk. Public materials do not provide audited standalone financials, precise headcount, or board detail, and the March 2026 sale materials carried standard but meaningful warnings around approvals, integration, and customer or supplier disruption. For diligence, CoolIT looks like a real scaled asset with strong technical and customer validation, but not a fully transparent one.[CO039, CO040, CO041, CO042, CO043, CO044]

1.6 Exhibits

2.1 Market Boundary and Substitutes

CoolIT does not serve the whole thermal-management market equally. Its most relevant market is direct liquid cooling infrastructure for high-density compute: processor coldplates, liquid loops, rack manifolds, coolant distribution units, secondary-loop interfaces, and the engineering and field services needed to deploy them. That definition deliberately excludes conventional building chillers, generic CRAC/CRAH equipment, and much of the broad “cooling” spend often included in analyst TAM slides. It also treats immersion cooling as a substitute architecture rather than part of CoolIT’s core served market because CoolIT’s product stack centers on direct-to-chip workflows and serviceable server designs. This tighter boundary matters because it aligns the market with actual adoption triggers—GPU power density, OEM qualification, facility plumbing, and service requirements—rather than with all data-center thermal spend. It also prevents double counting of adjacent facility spend that may benefit Ecolab after the acquisition but would not have belonged to standalone CoolIT hardware revenue.[CM001, CM002, CM012, CM036, CM037]

2.2 Sizing Lenses, Growth Drivers, and Estimate Spread

Public third-party market estimates support a large and rapidly expanding opportunity, but they also require caution. Future Market Insights isolates the AI data-center liquid cooling market and estimates about $3.7 billion in 2026 demand growing to roughly $17.8 billion by 2036, while MarketsandMarkets and 360iResearch publish broader category estimates that include a wider liquid-cooling universe. The resulting variance is not noise; it reflects different definitions of what is being counted. Even so, the strategic conclusion is consistent across sources: AI infrastructure is forcing a shift toward liquid cooling because air cannot efficiently remove heat at the highest rack densities. CoolIT’s own product announcements, especially CDUs designed around GB200/GB300-era racks, line up with the timing implied by these third-party demand signals. The key underwriting implication is that category growth should be anchored to deployable rack density and qualified programs, not to the broadest analyst number on the page.[CM003, CM006, CM007, CM008, CM032, CM033]

2.3 Buyer Segments, Payers, and Adoption Workflow

The highest-value buyers are hyperscalers and cloud-scale AI operators because they deploy the densest GPU clusters and can justify up-front cooling redesign around fleet economics. OEMs such as Dell, HPE, and Supermicro are the most important route-to-market intermediaries because many enterprises and even some hyperscalers prefer qualified liquid-cooled server configurations rather than bespoke hardware integration. Colocation operators are the next major segment: they need liquid-cooling capability to attract AI tenants without necessarily owning the compute themselves. Academic and government HPC sites matter as lighthouse customers because they validate performance and reliability, but they are not the main revenue pool. Budget ownership usually spans multiple stakeholders—platform engineering, infrastructure procurement, facilities, and operations—so vendors that can bridge chip, server, CDU, and facility workflows gain an advantage. That complexity helps explain why scaled validation, deployment, and service capabilities are commercially important in this market.[CM025, CM026, CM027, CM028, CM029, CM015]

2.4 Constraints, Contradictions, and Remaining Sizing Gaps

The demand story is strong, but adoption is not frictionless. Retrofit projects can require plumbing rework, rack redesign, and new operational procedures, while qualification cycles across chip, coolant, server, and facility layers slow deployments even when customers want liquid cooling. Market reports also leave important underwriting gaps: public evidence does not isolate CoolIT’s precise serviceable available market by OEM mix, geography, and deployment type, and vendor-sponsored reports differ materially on scope. At the same time, liquid cooling does not automatically displace every substitute. Advanced air remains viable at lower densities, and immersion has a real if narrower role where operators prioritize extreme density over conventional serviceability. The correct diligence takeaway is a very attractive market with definitional and execution caveats, not a frictionless greenfield TAM. In practice, the winners should be the vendors that shorten qualification time and absorb deployment complexity rather than those that only publish the largest TAM slide. That nuance is especially important when comparing private-company growth claims to public infrastructure-market narratives.[CM022, CM023, CM024, CM030, CM034, CM035]

2.5 Exhibits

3.1 Landscape Across Direct Peers, Incumbents, Substitutes, and Internal Build

CoolIT’s competitive set is wider than “other liquid cooling startups.” The most direct public peer is Asetek, which also markets direct-to-chip data-center cooling. Vertiv and Schneider Electric are different kinds of competitors: they approach the problem from the room, power, and facility side, with broader portfolios and enterprise account coverage. Dell, HPE, and Supermicro are not standalone direct competitors in every sale, but they hold crucial platform and channel power because buyers often want qualified liquid-cooled server configurations rather than a bag of components. Submer is the clearest substitute architecture competitor because immersion cooling can attack the same density problem from a different operating model. Finally, the largest hyperscalers can internalize portions of design and qualification, making internal build a persistent strategic threat. The practical implication is that CoolIT rarely competes only on component merit; it competes inside partner-controlled qualification and procurement systems, standards choices, and capacity commitments.[CP001, CP002, CP003, CP004, CP005, CP006]

3.2 Capability Differentiation and Route-to-Market

CoolIT’s public differentiation is unusually tangible. Rather than generic “AI-ready” messaging, the company discloses validated 15kW single-phase coldplates, more than 300 W/cm² heat flux, and AI-era CDU pairings tied directly to NVIDIA GB-series rack formats. It also leans heavily into co-innovation, design engineering, and formal validation assets, which suggests its route to market depends on winning qualification programs rather than only pushing standardized SKUs. This matters because enterprise liquid cooling is rarely purchased as a commodity on day one; it is integrated into server, rack, and facility workflows. OEMs and incumbents can counter with breadth and account control, but CoolIT appears stronger on disclosed direct-to-chip specificity than many of them. That differentiation should matter most in dense new-build AI programs where buyers still care about serviceability, reliability, lifecycle support, and time to qualification.[CP008, CP009, CP010, CP011, CP015, CP016]

3.3 Switching Cost, Multi-Homing, and Channel Power

The major commercial reality is that cooling is sticky once qualified. Switching a liquid-cooling partner affects not only the coldplate but also the loop, rack manifold, CDU, validation package, spares, and field-service procedures. That creates real lock-in after deployment. At the same time, specialists do not control the whole stack. OEMs, semiconductor ecosystem partners, and hyperscalers retain significant leverage because they decide who gets design access and whether solutions are standardized or multi-sourced. Buyers can probably multi-home across programs or facilities, but within a given qualified platform the switching cost is meaningfully higher than in most infrastructure software categories. That makes win rates, attach rates, and partner relationships more important than published feature checklists. It also means weak pricing transparency is not just a reporting nuisance; it hides where ecosystem power really sits. In many accounts, the commercial winner may be the vendor that reduces integration risk fastest rather than the one with the single most elegant component design, documentation set, or service motion.[CP019, CP020, CP021, CP022, CP023, CP024]

3.4 Moat Durability, Commoditization Risk, and Adverse Evidence

CoolIT appears competitively stronger than a young startup because it combines patent depth, validation infrastructure, manufacturing scale, and named deployment evidence. Those factors should matter in high-consequence AI and HPC environments. But the moat is operational rather than absolute. Broader incumbents can invest, OEMs can pressure margins, and immersion or internal-build paths can erode the highest-density share of the market. Pricing is also opaque, which makes it difficult to verify whether differentiation consistently converts into superior economics. The right conclusion is that CoolIT has a credible and probably durable position in direct liquid cooling, but not an unassailable one; durability depends on continuing to win design-ins, scale supply, and protect route-to-market access. The post-Ecolab platform may strengthen that position, but it also raises expectations for cross-sell execution and supply reliability. In other words, competitive advantage here is renewed by every generation of qualification, capacity expansion, and partner endorsement rather than protected indefinitely by one technical milestone.[CP014, CP017, CP018, CP025, CP026, CP027]

3.5 Exhibits

4.1 Revenue model and financial scale

CoolIT Systems operates a capital equipment revenue model: the company sells direct liquid cooling hardware including coldplate loops, rack manifolds, and coolant distribution units to OEM server manufacturers and directly to hyperscale data center operators. Revenue is project-oriented rather than subscription-based, with meaningful concentration in large platform relationships. The most reliable financial anchor is Ecolab own acquisition press release, which stated that CoolIT is expected to generate approximately $550 million in sales over the next twelve months. That figure substantially exceeds earlier analyst estimates and reflects the rapid scale-up of AI-driven data center deployments over the prior two years. Data Center Knowledge and Futurum Group both independently cited the 29 times NTM EBITDA multiple, confirming the acquisition premium interpretation. The revenue composition is not publicly broken out, but the product portfolio implies three logical revenue streams: OEM hardware sold to server manufacturers like Dell, HPE, and Supermicro through embedded qualification programs; direct sales to hyperscale cloud operators for custom or turnkey deployments; and professional services including installation, commissioning, and ongoing maintenance contracts. The professional services segment carries higher margin potential than one-time hardware sales and appears to be a growing share of revenue given CoolIT dedicated global professional services organization. CoolIT pricing is solution-specific and project-oriented. No public list prices exist for coldplate or CDU products. Projects are bid on a custom basis, reflecting the engineering complexity and facility-specific nature of liquid cooling deployment. This bespoke model limits comparability to software businesses but is appropriate for a capital equipment supplier to hyperscale infrastructure.[CI001, CI002, CI007, CI008, CI024, CI023]

4.2 Unit economics and capital structure

CoolIT unit economics are consistent with a precision manufacturing company that has achieved scale. Hardware gross margins for specialized cooling equipment typically run 40 to 55 percent, reflecting the engineering complexity of custom coldplates and CDUs, moderate material cost, and proprietary manufacturing processes. CoolIT has not publicly disclosed audited margins; these are industry comparables from McKinsey and analyst sources. KKR ownership from early 2023 through the July 2026 Ecolab close introduced institutional capital discipline including manufacturing scale-up in Calgary, field service infrastructure buildout, and co-investment from Mubadala for Middle East data center market access. The capital structure was typical for PE-backed industrials, with strong operating cash flow from hardware sales offsetting manufacturing CapEx. The 15 times return in three years confirms the business was cash-generating throughout the hold period. At 29 times NTM EBITDA, the acquisition multiple exceeds Vertiv typical forward EBITDA trading range and most comparable industrial technology acquisitions, reflecting the premium Ecolab paid for AI cooling exposure. S&P Global Market Intelligence confirmed that data center infrastructure M&A multiples reached record levels in 2025 to 2026 as hyperscaler buildout drove strategic acquisitions. Ecolab share price declined modestly after announcement as some investors questioned whether an industrial company was overpaying for a tech-adjacent asset. The strategic rationale is clear: Ecolab is building a $4 billion high-tech business by 2030 with CoolIT as the foundation. Mubadala co-investment in 2023 provided Gulf capital and signaled Middle East data center market relevance. The full exit with no earn-out or rollover equity confirmed clean deal structure; KKR and Mubadala fully monetized their positions at closing per Ecolab 8-K. KKR equity investment return of approximately 15 times on the assumed $316 million entry is unusually strong even for high-growth industrial technology. PitchBook data confirms the unusually strong return profile relative to comparable industrial PE exits.[CI003, CI004, CI005, CI006, CI010, CI011]

4.3 Financial gaps and diligence asks

CoolIT Systems as a private company under KKR ownership did not publish audited financial statements. The public record on revenue, gross margin, EBITDA, and cash flow is built from acquisition pricing and Ecolab own investor communications rather than direct disclosure. The $550 million NTM revenue figure is the most reliable data point but represents a forward projection at close, not a historical audited figure. All other financial metrics including gross margin, customer concentration, services mix, and CapEx intensity remain derived estimates or unresolved gaps. The most important financial diligence asks are: audited revenue and gross profit for FY2024 and FY2025 to confirm growth trajectory; customer concentration disclosure to quantify dependency on the top hyperscalers; services versus hardware revenue split and associated margin profile; manufacturing CapEx intensity and working capital for large-scale CDU and coldplate orders; and Ecolab integration plan specifically regarding whether CoolIT will be reported as a distinct segment. The absence of disclosed CAC, NRR, and burn rate metrics prevents independent assessment of unit economics quality. The data center liquid cooling market CAGR of 20 percent plus in 2026 through 2030 is the structural basis for the premium multiple. Multiple analyst firms including FMI, MarketsandMarkets, and 360iResearch confirm that the market reached $6-8 billion in 2026 with strong growth ahead. McKinsey analysis of AI data center economics further supports the cooling premium thesis. These market dynamics underpinned both the Ecolab deal rationale and the acquisition multiple. No adverse financial evidence such as defaults, write-offs, or contract disputes was found in reviewed sources.[CI018, CI032, CI025, CI009, CI013, CI035]

4.4 Exhibits

5.1 Product Definition and Module Map

In customer workflow terms, CoolIT sells a modular direct liquid cooling stack rather than a single component. The stack starts at the processor with coldplates, extends through coldplate loops and rack manifolds, and then connects into CDU and technology-cooling-system infrastructure that interfaces with the facility. That architecture lets the company participate at multiple points in the deployment rather than only at the chip socket. Importantly, the product is not just metal and plumbing: co-design, simulation, validation, manufacturing, commissioning, and field support are part of how the solution is delivered. This is why public materials repeatedly combine product pages with capability and service pages. Buyers are adopting a cooled workflow, not just a part number. That framing also explains why CoolIT’s sales motion looks closer to engineered infrastructure delivery than to commodity component distribution. The installed outcome is a stable thermal operating envelope for compute infrastructure, and every module exists in service of that workflow objective.[CE001, CE002, CE003, CE004, CE005, CE006]

5.2 Architecture, Integration, and Critical Dependencies

CoolIT’s architecture is explicitly layered: chip-level heat exchange, intra-server distribution, rack-level coolant management, CDU control, and facility interface. That modularity is a product strength because it can be integrated into multiple server and rack designs, but it also creates multiple dependency points. Semiconductor roadmaps, especially around NVIDIA’s GB-series systems, shape the power-density envelope. Manufacturing precision matters because the company is managing high heat flux, seals, metals, and fluid paths at production scale. Coolant chemistry and warm-water operating assumptions also matter because real-world performance depends on more than lab thermals. The result is a product architecture that is differentiated, but also highly dependent on qualification and execution across the whole chain. In practice, the architecture is only as strong as the weakest transition between chip, rack, and facility layers. This is why design, validation, manufacturing, and support should be read as technical dependencies rather than as optional service wrappers.[CE023, CE024, CE025, CE012, CE015, CE016]

5.3 Technical Differentiation, Roadmap, and Maturity

CoolIT’s public technical differentiation is stronger than generic AI-cooling marketing. The company cites validated single-phase coldplates beyond 15kW, heat-flux performance above 300 W/cm², OMNI all-metal coldplate architecture, Split-Flow design elements, and AI-era CDU sizing that maps directly to NVIDIA GB300 and GB200 rack configurations. These are specific engineering statements rather than only broad claims of efficiency. The roadmap also shows progression: a data-center pivot in 2012, OEM integrations in the mid-2010s, megawatt-class infrastructure and OMNI in 2024, and higher-density CDUs plus 15kW coldplates in 2026. Public case studies across supercomputing, research computing, and data-center environments support the view that the platform is commercially mature rather than a pilot-only technology. Few of these milestones would matter if they were purely lab artifacts, so deployment evidence is a key maturity signal. The cumulative pattern is a system platform moving up the density curve in step with the AI hardware cycle.[CE007, CE008, CE009, CE010, CE011, CE033]

5.4 Trust, Quality Controls, and Remaining Product Risks

The strongest public trust signals are engineering-centric: testing and validation capability, scaled manufacturing, patent depth, and deployed case studies. Those are meaningful for hardware infrastructure because reliability problems are expensive and visible. Public materials are less clear on formal certifications, software security, privacy posture, or detailed quality statistics such as failure rates and RMA performance. That does not imply weak controls; it means the public proof is biased toward thermal engineering rather than toward standardized compliance disclosure. For diligence, the technology looks credible and advanced, but the underwriting of product quality still requires private evidence on field failure, coolant lifecycle behavior, and support performance across heterogeneous customer environments. The main remaining risk is therefore execution transparency, not a lack of visible technical ambition. The public record supports capability, yet it still leaves open the question of how that capability performs across the full installed base over time in production.[CE013, CE014, CE026, CE030, CE031, CE032]

5.5 Exhibits

6.1 Customer base segmentation and scale

CoolIT Systems serves a diverse and technically sophisticated customer base spanning four primary segments: hyperscale cloud operators, high-performance computing research institutions, AI-focused data centers, and colocation operators. As of the Ecolab acquisition in 2026, CoolIT had deployed liquid cooling solutions in more than 300 data centers globally, an unusually large installed base for a specialized thermal management company. The customer base is anchored by the largest hyperscale operators globally. AWS, Azure, GCP, and Meta have each made public commitments to direct liquid cooling for AI workloads, and CoolIT product positioning and OEM relationships with Dell, HPE, and Supermicro make it a strong candidate for each of these operators deployments. Confidential data center supply agreements typically prevent public disclosure of these relationships by name. HPC research customers are the most publicly documented segment. Durham University COSMA cluster, LLNL Livermore, TACC Frontera at UT Austin, atNorth Iceland, Computerome at DTU, and multiple European scientific computing centers are all publicly confirmed CoolIT deployments. These institutions have documented their cooling infrastructure in academic papers, press releases, and facility announcements. The OEM channel through Dell, HPE, and Supermicro qualification programs creates an embedded distribution route to enterprise and cloud customers. Dell PowerEdge server lines with CoolIT liquid cooling enabled allow enterprise data centers to procure through standard server supply chains without a direct relationship with CoolIT. The AI cluster upgrade cycle from Nvidia A100 to H100 to GB200 has driven repeat CoolIT procurement at same hyperscale sites from 2022 through 2026.[CU001, CU002, CU003, CU013, CU023, CU018]

6.2 Named customers and deployment evidence

CoolIT named customer record is strongest in the HPC and scientific computing segments, where institutions publish detailed technical documentation of their infrastructure. The acquisition press release and subsequent analysis confirmed these named deployments and set the 300-facility milestone. atNorth, the Nordic data center operator based in Iceland, deployed CoolIT direct liquid cooling as a core component of its energy-efficient infrastructure offering. The atNorth case study is one of CoolIT most visible European deployments and demonstrates application in high-density colocation environments. Durham University COSMA supercomputer in the UK deployed CoolIT cooling as part of the DIRAC national HPC program. This deployment serves computational astrophysics workloads and represents CoolIT penetration into UK national research computing infrastructure. LLNL Livermore and TACC Frontera at the University of Texas Austin represent two of the largest US national laboratory and academic computing deployments. Computerome at the Danish Technical University further demonstrates European HPC market presence. ORNL Frontier, the worlds top-ranked supercomputer in 2022-2023, used liquid cooling architectures compatible with CoolIT CDU specifications. The Open Compute Project involvement demonstrates CoolIT alignment with open rack cooling standards that major hyperscalers including Meta, Microsoft, and Google use as procurement benchmarks. OCP certification validates CoolIT cooling for further hyperscale customer acquisition. CoolIT global professional services organization provides installation and commissioning teams in North America, Europe, and Asia Pacific, supporting the international customer base.[CU004, CU005, CU006, CU007, CU008, CU025]

6.3 Retention, expansion, and concentration risk

CoolIT customer retention characteristics are inferred from the business model and deal dynamics rather than disclosed metrics. The company does not publish NRR, GRR, or churn rates. Based on the hardware lifecycle model, retention is driven by long-term maintenance contracts, cooling fluid consumables, and the switching cost of replacing embedded thermal management infrastructure. Maintenance contracts and service agreements from 300-plus installations represent a growing recurring revenue stream with margins above hardware sales. Customer expansion follows a predictable pattern: pilot deployment in a single rack or row, demonstrated performance and reliability, then expansion to full data hall coverage. Professional services involvement at each stage creates sticky relationships and recurring revenue streams. The 2.18 billion kWh energy savings metric from KKR reporting implies a track record of measurable customer outcomes that support renewal and expansion. Customers benefit from 20 to 30 percent energy savings and improved PUE versus air cooling, reducing total cost of ownership. Concentration risk is the most material unresolved customer risk. Direct revenue from five or fewer hyperscale operators likely accounts for a disproportionate share of CoolIT total sales. Any procurement shift by AWS, Azure, or GCP would be a material financial event. No major contract cancellations or quality failures were identified in reviewed sources. Ecolab acquisition potentially diversifies the customer base through cross-selling into industrial, edge, and enterprise markets. The $4 billion 2030 revenue target requires entry into customer segments well beyond current hyperscale and HPC deployments, and Ecolab existing global industrial customer relationships represent a natural cross-sell opportunity.[CU009, CU011, CU014, CU015, CU010, CU022]

6.4 Exhibits

7.1 Severity-ranked risk register

CoolIT Systems risk profile is shaped by three structural factors: its position as a critical supplier to hyperscale AI infrastructure, its transition from an independent PE-backed company to an Ecolab subsidiary, and its dependence on Nvidia GPU thermal architecture. The five highest-severity risks in descending order are: hyperscale customer concentration, Nvidia GPU architecture dependency, Ecolab integration execution, data center CapEx cycle pauses, and competitive displacement from Vertiv, Asetek, and immersion cooling alternatives. Customer concentration is both the most material financial risk and the least publicly quantified. Direct revenue from a small number of hyperscale operators means a single procurement decision by AWS or Azure could have outsized financial impact. This risk is inherent to the data center cooling market given the small number of large-scale operators but remains unmitigated without diversification. Nvidia GPU architecture dependency is a key technical risk. CoolIT coldplates are engineered to the thermal envelope of specific GPU die generations. Next-generation GPU architectural changes require re-engineering on Nvidia timelines, creating a customer-imposed product refresh cycle. CoolIT demonstrated GB200 NVLink rack compatibility, partially mitigating near-term risk. Vertiv and Schneider Electric entering direct liquid cooling at scale, alongside funded competitors Asetek and Liquidstack, represent increasing pricing pressure. The competitive landscape is materially more intense in 2026 than when KKR invested in 2023. No material adverse evidence of product failures, operational incidents, quality disputes, or legal proceedings was identified across reviewed sources.[CR001, CR002, CR005, CR008, CR009, CR013]

7.2 Operational, regulatory, and legal risk

CoolIT manufacturing operations in Calgary depend on specialized inputs including copper heat exchangers, precision gear pumps, and specialty cooling fluids. Supply chain disruptions affecting these components could impact delivery timelines. US-China trade tensions and export controls on advanced manufacturing equipment add geopolitical risk to Asian component sourcing. Reuters and The Register documented significant supply chain pressure in data center hardware in 2026. EU Ecodesign Regulation 2025 creates compliance requirements for cooling equipment sold in Europe, affecting CoolIT European deployments. ESG reporting requirements and energy intensity regulations in the US, EU, and Singapore are increasing compliance burden. IEA data center power demand report confirms global regulatory scrutiny is intensifying. The 25-times CDU capacity expansion under KKR in three years creates risk of undisclosed quality control or warranty obligations. This scale of manufacturing ramp warrants due diligence on warranty claim history, yield rates, and quality control documentation. No public evidence of warranty or quality issues was found, but absence of evidence is not evidence of absence for a private company. CoolIT IP in coldplate design and CDU thermal management is a core competitive asset. McDermott Will and Emery analysis of industrial technology acquisitions confirms that IP transfer requires explicit patent and trade secret assignment at close. Post-acquisition IP ownership structure must be verified in data room. No patent infringement claims against CoolIT were identified in SEC filings or public court records. Labor market risk in technical cooling engineering and global professional services represents a growing operational challenge as CoolIT scales post-acquisition. Engineering headcount growth to support the $4B Ecolab target requires sustained hiring velocity.[CR005, CR006, CR007, CR003, CR023, CR028]

7.3 Mitigations and thesis-break triggers

CoolIT primary risk mitigations are product depth, installed base switching costs, and Ecolab acquisition rationale. The 300-plus deployment installed base creates moat. Multi-vendor OEM channel through Dell, HPE, and Supermicro reduces single-operator hyperscale dependency. CoolIT engineering investment in GB200 NVLink rack compatibility partially mitigates Nvidia architecture risk. Ecolab stated commitment to CoolIT operational autonomy partially mitigates integration execution risk. Key monitoring indicators to track risk materializing include: hyperscaler CapEx guidance in quarterly earnings calls, Nvidia GPU roadmap announcements, Ecolab quarterly disclosure of cooling segment revenue from FY2027, competitive RFP outcomes and CoolIT win rate, and EU regulatory timeline for server Ecodesign compliance. Thesis-break triggers that would prompt disinvestment recommendation include: Nvidia architectural change making existing coldplate design obsolete; loss of two or more major hyperscale accounts to competitors; Ecolab prioritizing cost-cutting over AI cooling investment; discovery of material undisclosed warranty liabilities; or Jason Waxman CEO departure without suitable replacement. Strategic misalignment between Ecolab corporate and CoolIT engineering leadership could accelerate talent attrition beyond typical post-M&A norms and slow roadmap execution. Critical diligence asks to refine risk assessment: customer concentration revenue schedule; IP ownership structure and assignment documentation; CDU warranty claim history from 2023 to 2026; Ecolab integration governance and CoolIT autonomy covenants; manufacturing quality control documentation; and Ecolab segment reporting structure from FY2027.[CR017, CR018, CR019, CR020, CR025, CR026]

7.4 Exhibits

8.1 Investment thesis and valuation context

CoolIT Systems represents the highest-quality direct liquid cooling franchise at the point of its July 2026 Ecolab acquisition. The investment thesis rests on four pillars: exceptional product-market fit in the AI infrastructure megatrend, a 300-plus global deployment installed base creating switching cost moat, rare engineering IP in coldplate and CDU design, and Ecolab strategic capability to accelerate market penetration across industrial, hyperscale, and international verticals. At $4.75 billion and approximately 29 times NTM EBITDA, the acquisition multiple significantly exceeds Vertiv public market comps of 18-22 times forward EBITDA and 4-5 times forward revenue. This premium reflects: the scarcity of companies combining technology depth, deployment scale, and OEM relationships at CoolIT level; the AI infrastructure sector premiums of 25-40 times EBITDA observed across multiple 2025-2026 transactions; and the strategic value of acquiring rather than building comparable capabilities. IDC MarketScape positions CoolIT as a leading direct liquid cooling vendor, and Futurum Group independently validated the acquisition premium interpretation. The anti-thesis is equally important. The 29 times EBITDA multiple encodes significant AI cooling growth assumptions. Any hyperscaler CapEx pause, Nvidia architectural change making current coldplates obsolete, or Ecolab integration failure would cause material value destruction. Ecolab share price declined 3-5 percent post-announcement as some investors questioned whether an industrial company can manage a high-growth AI cooling business at this multiple. Moodys flagged the acquisition as incrementally credit-negative due to added leverage. For new investors, CoolIT is no longer an investable private opportunity. The thesis is now an Ecolab ECL stock thesis. The recommendation is Track: the AI cooling structural growth is real, but confidence in the integration execution at this premium multiple requires additional evidence.[CV001, CV002, CV003, CV007, CV004, CV031]

8.2 Scenarios and comparable valuation

The bull case for CoolIT-Ecolab segment assumes AI cooling market grows at 25 percent CAGR through 2030, CoolIT reaches $1B+ revenue by 2028, and Ecolab segment hits $2B by 2030, with acquisition IRR exceeding 20 percent. This scenario requires continued AI cluster buildout velocity, successful hyperscale account retention, and effective expansion into new geographies and industrial segments. Nvidia GB200 NVLink rack qualification provides near-term revenue visibility for the bull case. The base case assumes 20 percent CAGR consistent with analyst estimates, CoolIT reaching $800M revenue by 2028 and $1.5B by 2030, with 12-15 percent IRR. On a 5-year horizon this implies revenue growing from $550M to $900M and multiple compression from 8.6 times to 6-7 times; still value-accretive but requiring successful integration. Bear case assumes a hyperscaler CapEx pause in 2027, flat revenue, and Ecolab integration friction, resulting in acquisition IRR below cost of capital. Vertiv trades at 18-22 times forward EBITDA and 4-5 times forward revenue, confirming the CoolIT acquisition premium. Data center M&A in 2025-2026 saw multiple transactions at 25-35 times EBITDA for AI infrastructure assets, supporting the deal multiple range. Asetek, the small-cap public liquid cooling company, trades at 3-5 times revenue, confirming market premium for specialized cooling versus generic hardware. HPE and Dell server hardware trades at 10-15 times EBITDA, confirming AI cooling specialist premium versus OEM channels. CoolIT installed base of 300-plus deployments implies $100-200M annually in recurring maintenance and upgrade revenue, a valuable base business even in bear case scenarios. AI infrastructure sector attracted category premium acquisitions in 2025-2026, confirming this is not CoolIT-specific overvaluation but a sector-level re-rating.[CV013, CV014, CV015, CV016, CV017, CV018]

8.3 Exit readiness and final diligence asks

CoolIT is fully integrated into Ecolab post-acquisition close in July 2026; no independent exit path exists as of the report date. Investment thesis is now an Ecolab ECL stock thesis, not a direct CoolIT position. The strategic value was captured by KKR at 15 times return; future value depends on Ecolab execution. For the Ecolab institutional thesis, the critical due diligence path includes: requesting audited CoolIT revenue and EBITDA for FY2024-2025 to confirm growth trajectory; obtaining customer concentration schedule to assess hyperscale dependency; reviewing CDU warranty claim history from the 25-times capacity scale-up; and monitoring Ecolab quarterly segment reporting structure from FY2027 for transparency on CoolIT performance. Thesis-break triggers include: Nvidia GB200 thermal architecture change obsoleting current CoolIT coldplate design; loss of two or more major hyperscale accounts in 12 months; Ecolab writing down CoolIT segment value within 24 months of close; or Jason Waxman CEO departure. The largest gap in valuation confidence is the absence of audited historical revenue data; the $550M NTM figure from Ecolab is forward-looking and not confirmed by audited statements. Ecolab leverage from the acquisition is manageable at approximately $3-4B in new debt given Ecolab free cash flow profile per Moodys analysis. Bear case probability is estimated at 25 percent based on AI investment cycle history and integration risk base rates; base case at 55 percent; bull at 20 percent. Institutional investors modeling Ecolab ECL as an AI infrastructure proxy assign $70-90 per share of value to CoolIT cooling segment in sum-of-parts analysis.[CV023, CV024, CV025, CV026, CV030, CV006]

8.4 Exhibits