Case Study
Delivering Ireland’s First Liquid-Cooled Supercomputer
How AlloComp supported CloudCIX in bringing the Boole Supercomputer from strategic ambition to live infrastructure in under six months
Client: CloudCIX
Location: Cork, Ireland
Platform: Boole Supercomputer
Project’s Phase: Initial POC deployment of liquid-cooled high-density nodes
Liquid-cooled NVIDIA HGX B200 platform, Dell Technologies hardware, high-density AI infrastructure
Cooling: Direct-to-chip liquid cooling
independent, vendor-agnostic procurement partner and project coordinator
August 2025 to January 2026
AI FORWARD > Supercomputing the Future, 27 January 2026
Client: CloudCIX
Location: Cork, Ireland
Platform: Boole Supercomputer
Project’s Phase: Initial POC deployment of liquid-cooled high-density nodes
Liquid-cooled NVIDIA HGX B200 platform, Dell Technologies hardware, high-density AI infrastructure
Cooling: Direct-to-chip liquid cooling
independent, vendor-agnostic procurement partner and project coordinator
August 2025 to January 2026
AI FORWARD > Supercomputing the Future, 27 January 2026
- Client: CloudCIX
- Platform: Boole Supercomputer
- Location: Cork, Ireland
- Project’s Phase: Initial POC deployment of liquid-cooled high-density nodes
- Technology focus: Liquid-cooled NVIDIA HGX B200 platform, Dell Technologies hardware, high-density AI infrastructure
- Cooling: Direct-to-chip liquid cooling
- Timeline: August 2025 to January 2026
- Launch: AI FORWARD > Supercomputing the Future, 27 January 2026
- AlloComp’s Role: independent, vendor-agnostic procurement partner and project coordinator
About the Project
CloudCIX wanted to expand the AI and high-performance computing capability of its Boole Supercomputer platform through a proof-of-concept deployment of next-generation, liquid-cooled GPU infrastructure capable of supporting real-world AI, research and enterprise workloads.
The result was the Boole Supercomputer upgrade: Ireland’s first liquid-cooled NVIDIA B200 deployment and one of the earliest systems of its kind in Europe. Hosted in CloudCIX’s Cork data centre, the platform gives organisations access to advanced AI compute within Irish and EU jurisdiction, supporting growing demand for sovereign, high-performance AI infrastructure.
The Irish Examiner described the system as Ireland’s first liquid-cooled NVIDIA B200 deployment and one of the earliest of its kind in Europe. (Irish Examiner)
AlloComp acted as CloudCIX’s independent, vendor-agnostic procurement and delivery coordination partner, supporting the project from technical specification and provider evaluation through to commercial coordination, deployment planning, stakeholder alignment and ongoing optimisation support.
From AlloComp’s first formal engagement in August 2025, the project progressed to public launch on 27 January 2026, creating a new sovereign AI infrastructure reference point for Ireland.
Key Outcomes
- First liquid-cooled NVIDIA B200 platform deployed in the country
- CloudCIX established a sovereign AI infrastructure capability in Ireland
- High-density, liquid-cooled infrastructure integrated into a live data centre environment
- Reduced procurement burden through independent vendor comparison and coordination
- Commercially strong outcome without compromising technical requirements
- Deployment converted into a market proof point, driving engagement and visibility
Technical Snapshot
The initial proof-of-concept deployment of liquid-cooled high-density nodes included:
- Dell PowerEdge platform with NVIDIA HGX B200 systems
- Direct-to-chip liquid cooling
- Rack-based design with liquid cooling manifold and cooling distribution unit
- Integration into CloudCIX’s existing Cork data centre environment
- Roadmap for future expansion
Moving beyond previous-generation GPU capacity
CloudCIX already had experience operating GPU infrastructure, including NVIDIA A100 and H100 systems. However, the organisation needed to take the next step.
The strategic requirement was clear: CloudCIX needed a platform capable of supporting more demanding AI training, inference and high-performance computing workloads, giving customers a local alternative to relying solely on global cloud providers.
This was not simply a hardware refresh. It was a strategic move into next-generation accelerated compute, giving CloudCIX a stronger platform for AI, research and enterprise workloads.
As CloudCIX put it during the project review: “It enables us to be in the market.”
That short statement captures the wider significance of the deployment. The Boole Supercomputer upgrade was not only about technical capability. It was about market relevance, national AI infrastructure, and creating a foundation for future growth.
Deploying advanced liquid cooled servers within the limits of an existing data centre
Greenfield AI infrastructure projects are often more straightforward – this one had to be delivered within the practical constraints of an existing live data centre environment. CloudCIX needed to balance:
Facility constraints
- Available power and cooling capacity
- White space and physical access limitations for large pre-integrated rack
- Integration into a live data centre environment
High-density deployment requirements
- Rack-level power and cooling demands
- Direct-to-chip liquid cooling integration
- Facility readiness for liquid-cooled GPU infrastructure
Commercial and procurement constraints
- Budget discipline
- Multi-vendor comparison and procurement
- Lead time and supply chain management
Long-term planning
- Future scalability
- Operational support requirements
- A roadmap for expanding
- AI and HPC capacity over time
Liquid cooling was a deliberate strategic decision. CloudCIX recognised that liquid cooling would become increasingly important for high-density AI infrastructure and used the project to build that capability within its own data centre environment.
The system also had to support future expansion, not just the initial deployment. The first phase was designed as a proof-of-concept NVIDIA HGX B200 deployment, with a roadmap to scale further over time.
This first phase of the Boole Supercomputer upgrade required careful coordination across power, existing and new cooling infrastructure, multiple vendors, specialist contractors and logistics.
That is exactly the type of environment where buyers need more than a reseller. CloudCIX needed a partner who could compare options, coordinate suppliers, manage risk and translate ambition into deployable infrastructure.
Independent, vendor-agnostic procurement partner and project coordinator
Technical Procurement
AlloComp supported the technical specification and procurement approach for the high-density AI platform, including the requirements for compute, cooling, power, integration and future growth.
Sourcing and Evaluation
AlloComp identified, engaged and evaluated multiple providers, delivering a clear, evidence-based comparison of leading technical and commercial options to CloudCIX.
Project Coordination
AlloComp coordinated across OEM, GPU platform, data centre, logistics and engineering partners, supporting alignment across technical and commercial workstreams throughout delivery.
Operational support
Post-deployment, AlloComp provides ongoing operational support and optimisation to maintain performance, stability and efficiency over time.
Why this mattered to CloudCIX
During the project review, CloudCIX highlighted AlloComp’s market testing and comparative vendor work as one of the most valuable parts of the engagement:
“The Boole Supercomputer was a strategic requirement for CloudCIX. AlloComp helped us turn that ambition into a delivered, operational platform.”
“For a project of this scale, having AlloComp as a procurement and coordination partner simplified the process for us. AlloComp handled the cross-vendor comparisons and much of the evaluation work we would otherwise have had to do internally.”
This had a direct impact on how the project was delivered. CloudCIX did not need to run parallel vendor evaluations, manage multiple commercial discussions or reconcile technical differences across providers.
Instead, CloudCIX had a clear, structured view of the available options and a partner to guide the decision-making process.
From first contact to launch in under six months
From initial project involvement in August 2025 to official launch in January 2026, AlloComp helped CloudCIX move from procurement planning to delivered, commissioned and publicly launched AI infrastructure in under six months.
August 2025
Initial engagement with CloudCIX on the Boole Supercomputer project
September 2025
Purchase order received
The project moved from option evaluation to committed procurement.
December 2025
System delivered to CloudCIX
January 2026
Commissioning, integration and readiness work completed.
Boole Supercomputer officially launched at AI FORWARD >Supercomputing the Future event
For a first-of-its-kind liquid-cooled AI platform involving next-generation GPUs, supplier coordination, facilities planning and live data centre deployment, that timeline is a strong signal of execution capability.
The hardest challenge was not what was expected
While liquid cooling was expected to be the main challenge for a first-of-its-kind deployment in the country, the most difficult practical aspect proved to be the physical placement of the rack.
Positioning the system within the existing Boole Supercomputer environment required careful planning and execution. The rack was exceptionally large and heavy, and moving it into place involved specialist handling, constrained access routes and close coordination on site.
This highlights an important reality of modern AI infrastructure. As compute density increases and liquid cooling becomes more common, the boundary between IT and facilities becomes less defined. Deployment is no longer just about installing equipment, it involves the building itself, from access and structural constraints to power, cooling and layout.
CloudCIX noted that while liquid cooling was initially seen as a major challenge, it proved manageable with the right approach. In contrast, the physical installation required the most careful coordination.
For organisations planning similar deployments, this is a valuable insight. The complexity is not only in the technology, but in how it is physically delivered and integrated into a live environment.
Lessons from a first-of-kind deployment
-
Facilities planning must start early
Power and cooling are only part of the readiness picture. Physical access, rack dimensions, floor loading, delivery routes and structural constraints are equally critical to successful deployment. -
Liquid cooling is manageable with the right partners
With the right expertise, support model and commissioning approach in place, liquid cooling can be deployed in a controlled and predictable way. -
Multi-vendor coordination requires structured governance
Delivery across OEM, data centre, cooling, logistics and commercial stakeholders benefits from clearly defined roles, responsibilities and proactive communication pathways from the outset. -
Procurement and installation are tightly coupled
In high-density AI infrastructure, procurement decisions directly influence site readiness, installation complexity, commissioning and long-term operations.
Turning infrastructure into market capability
The system is positioned to support industry, start-ups, scale-ups and applied research teams running data-intensive and compute-heavy workloads. Irish Examiner reported that the platform is intended for sectors including industry, start-ups, scale-ups and applied research, with medtech, pharma, manufacturing, robotics and other compute-heavy sectors among those expected to benefit.
A central theme of the Boole project is control: where data sits, who governs it, how workloads are managed and how organisations can avoid relying exclusively on global cloud platforms. CloudCIX’s infrastructure is hosted and operating in Ireland within Irish and EU jurisdiction and governed by Irish and EU law.
CloudCIX described the impact simply:
“It enables us to be in the market.”
That is a powerful outcome – it shows the project was not just technically successful. It shows anyone with sensitive – data regulated sectors, public bodies, research organisations and enterprises, that high-performance AI infrastructure is available locally, under Irish and EU governance.
The deployment and launch changed CloudCIX’s perception in the market – it created a visible proof point that they were not just discussing sovereign AI infrastructure but operating it and gave CloudCIX a platform for further engagement with industry, academia, government and specialist sectors.
The launch event was the value multiplier
AI FORWARD >Supercomputing the Future event brought more than 100 attendees to the CloudCIX Cork data centre, including industry leaders, researchers, infrastructure specialists, technology partners, public sector representatives and the Lord Mayor of Cork.
The event gave attendees the opportunity to see next-generation liquid-cooled GPU infrastructure operating in a live data centre environment and opened a wider discussion about the future of AI capability in Ireland.
CloudCIX feedback on the event:
Public engagement: exceeded expectations
Media coverage: exceeded expectations
Ecosystem response: exceeded expectations
Internal reaction: very positive
The launch helped turn a technical milestone into a market-facing moment – it shifted the conversation from concept to capability.
What this project proves for future customers
AlloComp can support complex AI infrastructure decisions
The project required evaluation of multiple vendor and OEM options, technical trade-offs, commercial comparison, supply chain realities and future growth planning. AlloComp helped CloudCIX make informed decisions without being locked into a single-vendor view.
AlloComp understands high-density, liquid-cooled deployment realities
This was not a standard IT procurement exercise. The project involved direct-to-chip liquid cooling, CDU integration, facility readiness, white space, power, cooling, access, physical installation and operational considerations.
AlloComp can reduce procurement burden and secure better value
CloudCIX highlighted the value of AlloComp’s vendor evaluations and managing commercial discussions. AlloComp engaged multiple OEMs, tested commercial options, reconcile technical differences across providers and helped CloudCIX secure a financially strong deal without compromising the technical requirements of the project.
AlloComp can help accelerate delivery
The project moved quickly from initial engagement to deployment and launch, requiring close coordination across procurement, OEM engagement, supply chain, logistics, delivery planning, installation and commissioning.
AlloComp can coordinate across technical, commercial and operational stakeholders
The project required alignment between CloudCIX, OEMs, the data centre environment, logistics partners, specialist engineering teams and launch stakeholders. AlloComp acted as a bridge between customer requirements, vendor capabilities and deployment realities.
AlloComp can support strategic outcomes, not just equipment supply
The deployment helped CloudCIX enter the AI infrastructure market with a sovereign, locally hosted platform. The launch campaign then helped turn the completed deployment into a visible market proof point, supporting engagement with industry, academia, public sector stakeholders and specialist user communities.
