Investment Opportunities Rising in Liquid Cooling for AI Data Centers
November 24, 20256 min read
Investment Opportunities Rising in Liquid Cooling for AI Data Centers
How liquid cooling is reshaping AI data‑center growth.
Tendrill
Investment Angles Emerging in Liquid Cooling for AI Data Centers
Liquid cooling has rapidly shifted from a niche technology to a core pillar of next‑generation AI infrastructure. Surging model sizes, denser GPU clusters, and the rollout of NVIDIA’s Blackwell architecture are pushing thermal and power limits to the edge, accelerating demand for more efficient cooling systems. Recent reporting from IEEE Spectrum, CIO.inc, TrendForce, and global market‑research firms highlights how liquid cooling is shaping the capex cycle of hyperscalers — and where investors may find opportunity as adoption moves from experimentation to standard practice.
Why Liquid Cooling Is Becoming Essential
AI data centers are running hotter than ever, with rack power densities now routinely surpassing 50–100 kW and edging toward 150 kW in advanced training clusters. According to an August 2025 whitepaper summarized by IEEE Spectrum (source), liquid cooling is emerging as the only scalable solution capable of handling high‑density deployments driven by Blackwell‑class GPUs and custom accelerators.
CIO.inc reports that the shift reflects a structural power‑and‑thermal inflection point, noting that hyperscalers and enterprise AI deployments are increasingly designing facilities around direct‑to‑chip loops and immersion tanks rather than air‑first layouts (source). This represents a multi‑year infrastructure refresh cycle.
Efficiency Gains Driving Adoption
Key performance drivers appealing to operators include:
Lower Power Usage Effectiveness (PUE) and more predictable energy costs
Higher rack density without expanding real‑estate footprints
Improved reliability and reduced throttling under sustained AI workloads
Compatibility with recirculating heat‑reuse systems improving sustainability metrics
Market Forecasts: Cooling Demand Is Accelerating
TrendForce expects the introduction of NVIDIA’s Blackwell platform to push liquid‑cooled server penetration from roughly 10% in 2024 to over 20% in 2025 (source). Blackwell’s 1,200–1,500W thermal profile for top‑end GPUs makes this shift unavoidable for large‑scale training clusters.
Meanwhile, the global data‑center cooling market — including both liquid and advanced air systems — is projected to grow from $16.3B in 2024 to more than $40B by 2030, according to a Research and Markets report (). Liquid cooling is expected to represent the fastest‑growing segment of that expansion.
Try Tendrill for free
Want to generate your own public shares? Try Tendrill for free.
Share this article
We're building Tendrill to be the smartest, most accurate agent out there.
This approach is rapidly becoming the preferred solution for AI‑training racks. Key beneficiaries include:
Asetek
nVent
Vertiv
Schneider Electric
These firms are scaling manufacturing for cold plates, manifolds, controls, and integrated cooling modules designed specifically for Blackwell and next‑generation accelerators.
2. Immersion Cooling
Immersion solutions from companies such as Asperitas, Submer, and LiquidStack are drawing fresh attention as hyperscalers pilot multi‑megawatt immersion suites. IEEE Spectrum highlights new immersion‑ready rack formats and dielectric fluid technologies enabling safer, modular installations.
3. Liquid Cooling‑as‑a‑Service
A growing number of integrators are offering turnkey solutions that bundle facility retrofits, coolant management, monitoring software, and ongoing maintenance. This model appeals to enterprises building private AI clusters without hyperscale expertise.
4. Waste‑Heat Recovery
Liquid systems unlock the potential for:
District heat networks
Industrial co‑generation
On‑site energy recycling
Sustainability‑linked financing and regulatory incentives in Europe and parts of Asia are expanding this as an investable theme.
Key Investment Themes for 2025 and Beyond
Infrastructure Providers Enter a Multi‑Year Upgrade Cycle
Hyperscalers (Amazon, Microsoft, Google, Meta) are driving 30–40% annual increases in data‑center capex tied to AI. Liquid cooling becomes a structural line item rather than a marginal cost — creating stable, long‑duration revenue opportunities for vendors.
Semiconductor‑Driven Demand
NVIDIA, AMD, Intel, and custom AI‑chip designers are all delivering hotter, denser accelerators. Blackwell is just the beginning; 2026 and 2027 roadmaps imply further TDP increases, locking in demand growth for liquid‑cooling OEMs.
Supply‑Chain Consolidation
Large equipment makers (Vertiv, Schneider, Johnson Controls) are acquiring specialized cooling startups to expand their AI thermal portfolios. Investors should expect more M&A as demand scales.
Real‑Estate and REIT Implications
Liquid‑ready AI campuses command premium rents and attract long‑term hyperscale tenants. Digital Realty, Equinix, and regional operators are accelerating retrofits to stay competitive, which may influence valuation multiples in data‑center REITs.
Sustainability and Regulation
EU energy‑efficiency rules and U.S. state‑level emissions requirements could mandate liquid‑cooling deployments in high‑density racks. This regulatory tailwind benefits providers whose systems improve PUE and reduce water consumption compared with evaporative cooling.
Conclusion
Liquid cooling is transitioning from a technical upgrade to a foundational requirement for AI data‑center design. With reporting from IEEE Spectrum, CIO.inc, TrendForce, and market‑research firms all pointing to a rapid acceleration in adoption — driven by Blackwell GPUs and the escalating heat profile of AI computation — the investment landscape is expanding across equipment vendors, integrators, REITs, and sustainability‑focused infrastructure plays. As capex cycles deepen and thermal limits tighten further, liquid cooling stands out as one of the most durable and strategically important growth themes in AI infrastructure.