HPE’s Liquid Cooling Strategy: How Data Center Partnerships Will Define the 2025 AI Market
Industry Adoption: How Hewlett Packard Enterprise Data Centers Are Dominating the AI Cooling Race
Between 2021 and 2024, Hewlett Packard Enterprise (HPE) laid the strategic groundwork to transition data center cooling from a niche HPC concern to a mainstream enterprise necessity. The period was defined by foundational moves: extending liquid cooling options to its high-volume HPE ProLiant Gen11 servers, forming key technology partnerships with specialists like Danfoss for heat reuse and Iceotope for immersion cooling, and securing initial commercial wins that showcased tangible efficiency gains. Projects with QScale in Canada, for instance, validated the approach by achieving a 20% reduction in energy consumption and a Power Usage Effectiveness (PUE) of less than 1.2. This era was characterized by proving the economic and environmental viability of liquid cooling, with deployments in pioneering data centers like Green Mountain in Norway and the UAE’s first managed Direct Liquid Cooling (DLC) service with Khazna, establishing proof points for the technology’s effectiveness.
The year 2025 marks a dramatic inflection point, shifting from proving the technology to aggressively scaling it for the global AI build-out. The adoption has broadened from specialized supercomputers to major enterprise and hyperscale applications. This acceleration is evident in massive commercial agreements, such as a greater than $1 billion AI server deal with X (formerly Twitter) and a landmark partnership with Japanese telecom giant KDDI to build a large-scale AI data center in Osaka by 2026. The application range has expanded significantly, now encompassing automotive AI development with Subaru, environmental science in New Zealand with the “Cascade” supercomputer, and national-level AI infrastructure in South Korea with KISTI. This shift is driven by the extreme thermal demands of new technologies like NVIDIA’s Blackwell GPUs, making HPE’s integrated liquid cooling not just an option, but a core enabler for the next wave of AI. The strategy has matured from technology integration to full-scale market capture, solidifying HPE’s position as a critical player in the AI infrastructure ecosystem.
Table: Hewlett Packard Enterprise’s Strategic Investments in Liquid Cooling and AI Infrastructure
| Partner / Project | Time Frame | Details and Strategic Purpose | Source |
|---|---|---|---|
| Server Business Growth | October 2025 | Reported record server revenues of $4.9 billion, a 16% year-over-year increase, primarily driven by the conversion of AI-related orders. This demonstrates a direct financial return on its strategic focus on high-performance, liquid-cooled servers. | HPE Gains From Server Refresh Cycle: Sign of More … |
| Acquisition of Juniper Networks | July 2025 | Completed the acquisition of Juniper Networks to create a comprehensive, cloud-native, AI-driven portfolio. The move integrates AI-native networking with HPE’s servers and liquid cooling, enhancing its high-margin business for end-to-end AI data center solutions. | Hewlett Packard Enterprise closes acquisition of Juniper … |
| Leibniz Supercomputing Center (LRZ) | December 2024 | Commissioned to build a €250 million “Blue Lion” supercomputer in Germany. The system features a 100% direct liquid-cooled architecture, serving as a flagship project to showcase the capabilities of its liquid cooling technology in exascale and AI computing. | HPE to Build €250M Liquid Cooled HPC at Leibniz … |
| Intellectual Property Development | June 2024 | Leveraged a substantial R&D investment resulting in over 300 patents in liquid cooling. This IP portfolio created a competitive moat and was instrumental in securing $4.5 billion in AI server orders. | HPE takes $4.5bn in enterprise IT AI server orders |
| QScale QO1 Campus | May 2023 | Became an anchor tenant at QScale’s QO1 campus in Quebec, Canada. This provides a large-scale environment powered by renewable energy to deploy and showcase its liquid-cooled AI and HPC infrastructure, demonstrating a commitment to sustainable data center operations. | HPE becomes anchor tenant at QScale’s QO1 campus in … |
Table: Hewlett Packard Enterprise’s Key Alliances in Data Center Liquid Cooling
| Partner / Project | Time Frame | Details and Strategic Purpose | Source |
|---|---|---|---|
| nVent, NVIDIA, Lenovo, and AMD | November 2025 | Participated in a joint panel at the SC25 conference to discuss the future of liquid cooling, signaling a shared industry-wide consensus on adopting liquid cooling as the standard for AI and HPC. | nVent Unveils New Liquid Cooling and Power Portfolio at … |
| Oak Ridge National Laboratory | October 2025 | Contracted to build two new systems, “Discovery” and “Lux,” for ORNL. The “Discovery” supercomputer will be fully liquid-cooled, reinforcing HPE’s leadership in building next-generation exascale and AI systems for premier government labs. | HPE to build two systems for Oak Ridge National Laboratory |
| UNICOM Engineering | September 2025 | As part of the HPE Partner Ready OEM program, UNICOM Engineering began offering immersion cooling-ready versions of HPE ProLiant Gen11 servers, expanding the channel availability of specialized cooling solutions. | HPE immersion-cooled servers by UNICOM Engineering |
| Subaru | July 2025 | Subaru is leveraging HPE’s AI solutions, including the DLC-equipped HPE Cray XD670 system, to accelerate the development of its EyeSight Driver Assist System, showcasing liquid cooling’s application in the automotive industry. | Subaru selects HPE to accelerate AI development for next- … |
| KDDI Corporation | June 2025 | Partnered with the Japanese telecom giant to build the Osaka Sakai Data Center by early 2026. The facility will use HPE’s DLC technology to cool NVIDIA Blackwell GPUs, targeting the $5.35 billion Japanese hyperscale market. | KDDI and HPE join forces to launch AI data center … |
| Nexalus | May 2025 / Dec 2024 | Announced a collaboration to integrate its specialized liquid cooling and heat recycling technology into core HPE ProLiant server models (DL360, DL365, DL380a), aiming to bring advanced efficiency to mainstream enterprise servers. | Nexalus brings liquid cooling technology to HPE servers |
| Khazna Data Centers | August 2024 | Partnered to launch the first managed data center hosting service with DLC for AI in the UAE, establishing a strategic foothold for high-density computing in the Middle East’s sovereign AI market. | Hewlett Packard Enterprise introduces first managed data … |
| Danfoss | June 2024 | Collaborated to deliver an off-the-shelf heat recovery solution, combining HPE’s Modular Data Centers with Danfoss’s heat reuse technology to transform a major operational expense into a potential asset. | Hewlett Packard Enterprise and Danfoss partner to curb … |
| Iceotope, Intel, and nVent | February 2023 | Partnered to develop a sustainable Open RAN solution using Iceotope’s precision immersion liquid cooling, demonstrating applicability in telco and edge computing environments. | Sustainable Open RAN Partnership |
Geography of Hewlett Packard Enterprise Data Centers
Between 2021 and 2024, HPE’s liquid cooling activities were concentrated in North America and Europe, establishing key reference sites. In Canada, the anchor tenancy and subsequent project with QScale showcased sustainable, high-density computing. In Europe, deployments were prominent in the HPC sector, with the €250 million “Blue Lion” supercomputer in Germany and AI infrastructure for Green Mountain’s colocation customers in Norway. The partnership with Khazna Data Centers in the UAE marked a strategic, but singular, entry into the Middle East. This geographic footprint was about establishing credibility and proving the technology in mature, high-tech markets.
The landscape has shifted dramatically in 2025, signaling a deliberate and aggressive expansion into the Asia-Pacific region. This pivot is not speculative; it is backed by major commercial agreements for national-scale infrastructure. The partnership with KDDI to build an advanced AI data center in Osaka is a move to capture a significant share of Japan’s $5.35 billion hyperscale market. Simultaneously, HPE was selected to build South Korea’s largest supercomputer for KISTI, deploying its most advanced fanless liquid cooling architecture. The “Cascade” supercomputer project in New Zealand further cements this APAC focus. This geographical expansion indicates that APAC is viewed as the primary growth engine for AI, and HPE is positioning its liquid cooling technology as the foundational enabler for this growth, creating a strong first-mover advantage in the region.
Technology Maturity of Hewlett Packard Enterprise Data Centers
In the 2021–2024 timeframe, HPE’s liquid cooling technology transitioned from a niche, custom solution for high-performance computing (HPC) toward commercialization for the broader enterprise. The key validation point was the launch of liquid cooling options for the mainstream HPE ProLiant Gen11 server line in September 2023. This move signaled that the technology was mature and reliable enough for general-purpose data centers. During this period, more advanced concepts, like the 100% fanless architecture, were announced (October 2024), but remained in the “emerging technology” phase, serving to demonstrate HPE’s innovation pipeline rather than being widely deployed. The focus was on piloting and initial commercial deployments, such as with QScale and Khazna, to build a portfolio of successful case studies.
From 2025 onwards, the technology has rapidly moved into the “scaling” and “maturation” phases. The once-emerging 100% fanless DLC architecture is now a commercial reality, selected for South Korea’s largest supercomputer. Direct liquid cooling is no longer just an “option” but an integral, non-negotiable component of HPE’s flagship products, including the new HPE ProLiant Gen12 servers and systems shipping with NVIDIA’s power-hungry GB200 NVL72 architecture. The introduction of patented technologies like Adaptive Cascade Cooling—a hybrid system supporting both air and 100% liquid cooling—demonstrates a maturing portfolio. This flexibility shows HPE is not just offering a single extreme solution but is providing a sophisticated suite of thermal management tools to meet diverse customer needs, from enterprise to exascale, validating its readiness for mass-market adoption.
Table: SWOT Analysis of HPE’s Liquid Cooling Strategy
| SWOT Category | 2021 – 2023 | 2024 – 2025 | What Changed / Resolved / Validated |
|---|---|---|---|
| Strengths | Established leadership in HPC with liquid-cooled Cray systems. Growing IP portfolio and partnerships with specialists like Iceotope to explore different cooling modalities (immersion). | Dominance in TOP500 and Green500 supercomputer lists (e.g., #1 system El Capitan). Deep IP portfolio with over 300 patents. Integrated portfolio from servers (ProLiant) to supercomputers (Cray) with built-in DLC. | Validated: HPE’s long-term R&D in liquid cooling, evidenced by its patent portfolio and consistent wins in building the world’s most powerful and efficient supercomputers, has translated into a significant competitive moat in the burgeoning AI market. |
| Weaknesses | Liquid cooling was still perceived as a niche technology primarily for HPC, with limited adoption in mainstream enterprise servers. | Some investor caution reported regarding fiscal 2026 guidance, suggesting potential concerns about long-term growth trajectory or margin pressures despite strong current performance. | Unresolved: While revenue from AI server sales is at a record high ($4.9B), market concern over future guidance indicates that HPE still needs to prove it can sustain this high-growth momentum and translate its technological leadership into consistent, long-term financial outperformance. |
| Opportunities | The growing power consumption of CPUs and GPUs presented an opportunity for more efficient cooling. The liquid cooling market was projected to grow, with a high CAGR predicted through 2026. | The data center liquid cooling market is projected to grow from $2.84B in 2025 to $21.14B by 2032 (33.2% CAGR). Surging demand for AI servers powered by energy-intensive GPUs (NVIDIA Blackwell) makes liquid cooling a necessity. | Validated & Expanded: The theoretical market opportunity from the earlier period has become a concrete, urgent market need. The surge in AI workloads has made liquid cooling essential, not optional, validating HPE’s strategic focus and dramatically expanding its addressable market. |
| Threats | Competition from other server manufacturers and specialized cooling vendors. Slower-than-expected enterprise adoption of a new, complex technology. | The need to successfully integrate the large acquisition of Juniper Networks to realize its full strategic value. The rapid pace of AI hardware innovation requires continuous R&D investment to ensure cooling solutions can handle next-generation thermal loads (e.g., beyond Blackwell). | Evolved: The threat has shifted from convincing the market to adopt liquid cooling, to executing at scale. The new risks involve complex M&A integration (Juniper) and staying ahead in a hyper-competitive technology race where falling behind on a single product cycle could be detrimental. |
Forward-Looking Insights and Summary
The data from 2025 clearly signals that Hewlett Packard Enterprise has successfully positioned its liquid cooling technology as a foundational pillar of the modern AI data center. Looking ahead, market actors should closely monitor several key signals that will determine if this strategic advantage translates into sustained market dominance. The first is the general availability of HPE systems with the NVIDIA Blackwell Ultra option, scheduled for January 2026; their market uptake will be a direct measure of HPE’s ability to capitalize on the bleeding edge of AI hardware. Secondly, the launch of the KDDI Osaka Sakai Data Center in early 2026 will serve as a crucial, large-scale commercial proof point for HPE’s technology in the competitive Asia-Pacific hyperscale market.
Internally, the successful integration of Juniper Networks’ AI-native networking into HPE’s server and GreenLake portfolio is paramount. If executed well, this synergy could create an unparalleled end-to-end, high-margin data center offering that is difficult for competitors to replicate. Finally, with the liquid cooling market projected to grow at a staggering 33.2% CAGR through 2032, HPE’s quarterly server revenue and margin performance will be the ultimate barometer of its success. While the company has built a significant competitive moat through its integrated technology, its ability to execute on these large-scale deployments and strategic integrations will define its trajectory in the year ahead.
Frequently Asked Questions
Why has liquid cooling become so critical for data centers in 2025?
According to the analysis, the extreme thermal demands of new AI technologies, particularly powerful GPUs like NVIDIA’s Blackwell series, have made traditional air cooling insufficient. This has shifted liquid cooling from a niche solution for supercomputers to a core, necessary component for mainstream enterprise and hyperscale AI applications.
What is the core of HPE’s strategy to dominate the AI cooling market?
HPE’s strategy is built on three pillars: 1) Integrating its proprietary liquid cooling technology, backed by over 300 patents, directly into its high-volume server products (like ProLiant and Cray). 2) Forming key partnerships with both technology specialists (Danfoss, Nexalus) and major customers (KDDI, QScale) to scale deployment. 3) Acquiring complementary companies like Juniper Networks to offer a complete, end-to-end AI data center solution that includes servers, cooling, and networking.
The article highlights a geographic shift in HPE’s focus. Where is HPE concentrating its efforts in 2025?
While HPE established its technology in North America and Europe between 2021-2024, the article shows a dramatic and deliberate expansion into the Asia-Pacific (APAC) region in 2025. This is evidenced by major agreements like the KDDI partnership to build an AI data center in Japan, the contract for South Korea’s largest supercomputer (KISTI), and the “Cascade” project in New Zealand.
How has HPE’s technology evolved from the 2021-2023 period to now?
In the earlier period, HPE was focused on commercializing liquid cooling by offering it as an option for its mainstream ProLiant servers. By 2025, the technology has matured significantly. Advanced concepts like the 100% fanless architecture are now being commercially deployed, and Direct Liquid Cooling (DLC) is an integral component of its flagship AI systems, demonstrating its readiness for mass-market adoption.
What are the main risks or challenges facing HPE’s liquid cooling strategy?
The SWOT analysis points to two key evolving threats. First, HPE must successfully execute the complex integration of its $14 billion acquisition of Juniper Networks to realize its strategic value. Second, it must keep pace with the rapid innovation in AI hardware to ensure its cooling solutions can handle the thermal demands of chips that will come after NVIDIA’s Blackwell, requiring continuous R&D investment.
Experience In-Depth, Real-Time Analysis
For just $200/year (not $200/hour). Stop wasting time with alternatives:
- Consultancies take weeks and cost thousands.
- ChatGPT and Perplexity lack depth.
- Googling wastes hours with scattered results.
Enki delivers fresh, evidence-based insights covering your market, your customers, and your competitors.
Trusted by Fortune 500 teams. Market-specific intelligence.
Explore Your Market →One-week free trial. Cancel anytime.
