If you are a data centre operator struggling with rising energy bills and hardware that sits idle during off-peak hours — this project developed modular microserver appliances demonstrated at TRL 7 that let you tailor computing resources to actual workload demands. The system was tested in a real data centre environment with 15 consortium partners, combining embedded CPUs, GPUs, and FPGAs into a single flexible architecture. This means you deploy only the compute power you need, cutting both energy consumption and total cost of ownership.
Modular Microserver Systems That Cut Data Centre Costs and Energy Use
Imagine your data centre is like a kitchen where every appliance is a massive industrial oven — even when you just need to toast bread. M2DC built a system of small, specialized computing modules that you can mix and match like building blocks — plug in exactly the processors you need for each job, whether that's crunching images, running cloud services, or heavy scientific calculations. The modules fit together using standard connectors, so you can swap parts in and out without redesigning the whole setup. The result is a server system that uses less power, costs less to own, and adapts to whatever workload you throw at it.
What needed solving
Data centres today waste enormous amounts of energy and money running general-purpose servers that are over-provisioned for most workloads. When your computing needs vary — from image processing spikes to steady cloud hosting — fixed hardware means you are either paying for capacity you do not use or struggling with bottlenecks during peak demand. The inability to mix and match specialized processors (CPUs, GPUs, FPGAs) in a single flexible system forces operators into costly one-size-fits-all infrastructure.
What was built
The project delivered final prototypes of resource-efficient microserver modules and a next-generation microserver system, along with turnkey appliances optimized for specific workloads. These are modular computing units using standard COM Express connectors that combine heterogeneous processors — embedded CPUs, GPUs, FPGAs, and manycore processors — into a single, energy-efficient server architecture with built-in management and monitoring capabilities.
Who needs this
Who can put this to work
If you are running image processing at scale and your current servers waste power on general-purpose hardware — M2DC built turnkey appliances specifically optimized for image processing workloads. The architecture integrates hardware accelerators like GPUs and FPGAs through standard COM Express modules, letting you match hardware to your exact processing needs. With 10 industry partners involved in development, the appliances were designed for real commercial deployment.
If you are an engineering firm or research lab that needs HPC capability but cannot justify the cost of a full supercomputer cluster — M2DC created a next-generation microserver system that delivers heterogeneous computing power in a compact, energy-efficient package. The final prototypes include resource-efficient microserver modules that combine manycore processors with reconfigurable hardware. The system was backed by EUR 7,998,935 in EU funding and validated across multiple real-life application domains.
Quick answers
What does this cost compared to traditional server infrastructure?
M2DC focused on Total Cost of Ownership (TCO) optimization as a core design goal. The modular approach means you buy and power only the compute resources you actually need, rather than over-provisioning with general-purpose servers. Exact pricing depends on configuration, but the architecture was specifically designed to reduce both capital and operating expenses.
Can this scale to a full production data centre?
Yes. The project targeted TRL 7, meaning the system was demonstrated in a real data centre environment running real-life applications. The architecture uses established standards like COM Express for module integration, making it compatible with existing data centre infrastructure. With 15 partners across 7 countries validating the design, it was built for production-scale deployment.
What is the IP situation — can I license or buy this technology?
The project involved 15 partners including 10 industry organizations and 2 SMEs. IP ownership and licensing terms would need to be negotiated with the consortium, coordinated by Instytut Chemii Bioorganicznej Polskiej Akademii Nauk in Poland. Based on available project data, the open architecture approach suggests components may be available through multiple consortium members.
How does this integrate with my existing data centre management software?
M2DC was specifically designed with well-defined interfaces to the surrounding software ecosystem — this was one of the three main pillars of the project. The system uses the latest middleware for resource management and provisioning, and was built to seamlessly interact with existing data centre infrastructure. Integration was validated in real data centre deployment.
What types of workloads does this actually support?
The project demonstrated support for image processing, cloud computing, and HPC workloads. The heterogeneous architecture — combining embedded CPUs, GPUs, FPGAs, and manycore processors — means the system can be tailored to various application domains. Each turnkey appliance was optimized for specific workload profiles.
Is there regulatory compliance for data centre deployments?
Based on available project data, M2DC focused on energy efficiency and dependability as built-in system enhancements. The project addressed security and behaviour monitoring at the system level. Specific regulatory certifications would need to be confirmed with the consortium partners directly.
What is the timeline to deploy this in my facility?
The project ran from 2016 to 2019 and produced final prototypes of both the microserver modules and the complete appliance system. With TRL 7 achieved, the technology was validated in an operational environment. Deployment timelines would depend on your specific configuration needs and discussions with consortium partners.
Who built it
The M2DC consortium is strongly industry-driven with 10 out of 15 partners (67%) coming from industry, complemented by 2 universities and 3 research organizations across 7 countries (CH, DE, FR, IT, PL, SI, UK). This heavy industry presence signals that the technology was designed with commercial viability in mind from the start. The consortium includes 2 SMEs, and the project was coordinated by Instytut Chemii Bioorganicznej Polskiej Akademii Nauk in Poland. The geographic spread across major European technology markets — particularly Germany, France, Italy, and the UK — suggests strong potential for deployment across the continent. With EUR 7,998,935 in EU funding backing the effort, the consortium had substantial resources to move from concept to operational prototypes.
- INSTYTUT CHEMII BIOORGANICZNEJ POLSKIEJ AKADEMII NAUKCoordinator · PL
- XLAB RAZVOJ PROGRAMSKE OPREME IN SVETOVANJE DOOparticipant · SI
- COMMISSARIAT A L ENERGIE ATOMIQUE ET AUX ENERGIES ALTERNATIVESparticipant · FR
- HUAWEI TECHNOLOGIES DUESSELDORF GMBHparticipant · DE
- CHRISTMANN INFORMATIONSTECHNIK + MEDIEN GMBH & CO KGparticipant · DE
- OFFIS EVparticipant · DE
- ARM LIMITEDparticipant · UK
- POLITECNICO DI MILANOparticipant · IT
- UNIVERSITAET BIELEFELDparticipant · DE
Coordinated by Instytut Chemii Bioorganicznej Polskiej Akademii Nauk (Poland). Contact details can be found through the project website or CORDIS page.
Talk to the team behind this work.
Want to explore how M2DC's modular microserver technology could reduce your data centre costs? SciTransfer can connect you directly with the right consortium partner for your specific use case.