KKR’s 2023 thesis was that rapid compute growth would expose the structural limits of air cooling and create durable value in liquid cooling infrastructure

Decision Lens

The CoolIT deal is not primarily a cooling story. It is a market signal that the energy cost structure of AI-era data centers is forcing a reclassification of what counts as power infrastructure. Liquid cooling systems reduce cooling energy consumption by roughly 30–40% compared to air-cooled equivalents, while closed-loop operation cuts water draw through recirculation rather than evaporation. For a Global Head of Data Center Energy managing a multi-GW portfolio, that efficiency differential is a lever on the power budget, PPA sizing logic, and sustainability reporting. The $4.75 billion price tag is an external valuation of that leverage — and the question is whether internal procurement and vendor strategy has kept pace with what the market is now willing to pay for it.

90-Second Brief

In recent days, ecolab has agreed to acquire CoolIT, a Calgary-based direct liquid cooling provider, from KKR in a deal valuing the company at $4.75 billion. KKR, which acquired CoolIT in 2023 through its Global Impact Strategy, generated approximately 15 times its original equity investment. Ecolab intends to integrate CoolIT’s technology with its water chemistry, fluid management, and digital monitoring capabilities to expand its Cooling-as-a-Service offering for AI data centers. Liquid-cooled systems use roughly 30, 40% less energy for cooling than air-cooled alternatives and reduce water consumption through closed-loop design.

What’s Actually Happening

KKR’s 2023 thesis was that rapid compute growth would expose the structural limits of air cooling and create durable value in liquid cooling infrastructure. The 15x return validates that thesis at a pace few anticipated — reflecting not just AI-driven demand acceleration, but the degree to which liquid cooling has moved from pilot deployments to production-scale adoption across hyperscale and colocation operators.

The mechanism matters for energy planning. Direct liquid cooling removes heat at the chip or server level using coolant, eliminating a significant portion of the thermodynamic work conventional air systems require. The 30–40% cooling energy reduction follows directly from that physics. Closed-loop design recirculates rather than evaporates coolant, cutting water draw — a material consideration in water-stressed markets or where water rights are a site-selection constraint.

Ecolab’s strategic logic is to combine CoolIT’s engineered hardware with its own chemistry, fluid management expertise, and digital monitoring to deliver a managed service rather than a product. That shift transforms the procurement relationship: from capital expenditure on cooling equipment to a performance-guaranteed service contract with a large industrial counterparty.

Why It Matters for Global Heads of Data Center Energy?

The energy implication is direct and quantifiable. Cooling typically represents 30–40% of total facility energy consumption in conventional air-cooled data centers. A 30–40% reduction in that portion produces a measurable shift in total facility PUE and, therefore, in contracted power demand. Energy leaders who have established interconnection positions and PPA volumes based on air-cooled PUE assumptions may need to recalibrate power forecasts for liquid-cooled build-outs — with downstream implications for offtake agreement sizing and grid reservation commitments.

The Cooling-as-a-Service model also introduces a contract layer that intersects directly with energy procurement. Performance-based cooling contracts create efficiency guarantees that interact with utility rate structures and sustainability reporting obligations. An operator committing to a specific PUE for tariff negotiations or 24/7 CFE reporting needs clarity on what a service provider is actually underwriting and over what time horizon.

Vendor consolidation at this scale also warrants scrutiny. Concentrating a critical thermal management layer under a single large industrial provider changes the risk profile for long-term infrastructure planning. Contract terms, sole-source dependency, and service continuity provisions deserve the same rigor applied to generation counterparties.

The Forward View

The consolidation of liquid cooling under Ecolab changes the competitive dynamics for this technology class. A $4.75 billion acquisition by a global industrial services company with established data center relationships suggests adoption will be operationalized at a pace and through distribution channels that smaller independent vendors cannot replicate. That likely compresses adoption timelines — but also implies that pricing power over the next 12–24 months may shift toward service providers rather than operators.

For energy strategy specifically, the operative question is whether liquid cooling becomes a standard specification for new AI-density builds within the current planning cycle. If rack power densities continue increasing — and there is no current evidence that trajectory is reversing — the thermal management constraints that created this deal’s value will intensify. Energy leaders should expect liquid cooling adoption to feed back into power density assumptions in capacity planning, and potentially to alter the cost-benefit calculus for behind-the-meter generation commitments as total facility load profiles evolve.

What We’re Uncertain About?

  • Integration timeline for the combined Ecolab-CoolIT service offering: Ecolab has stated intent to expand its Cooling-as-a-Service platform, but no commercial availability timeline for the integrated product has been confirmed. What would resolve this: Ecolab product roadmap communications post-transaction close.

  • Efficiency range across operational contexts: The 30–40% cooling energy reduction is a category-level comparison between liquid-cooled and air-cooled facilities. Whether that range holds across varying rack densities, mixed infrastructure environments, and legacy retrofit scenarios is not established in available evidence. What would resolve this: independent operational benchmarking across multi-site, multi-density deployments.

  • Contract economics of Cooling-as-a-Service: It is not confirmed whether CaaS pricing will be structured to pass efficiency savings to operators or absorb them as service margin, nor what performance risk-sharing mechanisms will apply. What would resolve this: published commercial terms or reference contracts from early adopters.

  • Jurisdictional friction for closed-loop deployment: Coolant chemistry, wastewater treatment requirements, and water rights regulations vary materially across data center markets. The degree to which these create deployment friction in specific geographies is unconfirmed. What would resolve this: operator experience data from multi-jurisdiction deployments.

One Question to Bring to Your Team

If liquid cooling becomes the default specification for AI-density new builds in this planning cycle, do our current interconnection reservations, PPA volumes, and power demand forecasts reflect updated PUE assumptions — or are they still priced against air-cooled baselines that may no longer represent our actual buildout?

Sources

  • Esgtoday — KKR Impact Fund Earns 15x Investment on Sale of Data Center Cooling Provider CoolIT to Ecolab (Link)

Tags