Max Planck Computing and Data Facility (MPCDF)

What Max Planck Computing and Data Facility (MPCDF) does

Max Planck Computing and Data Facility (MPCDF) is a non-profit, cross-institutional competence centre of the Max Planck Society (Max-Planck-Gesellschaft zur Förderung der Wissenschaften e.V.) headquartered at the Garching Research Campus near Munich, Germany. It operates as a centralized HPC and data-management facility serving the computational and data-science needs of Max Planck Institutes across astrophysics, fusion research, materials science, bio sciences, polymer research, and theoretical chemistry. The organization is funded through institutional cost recovery from the Max Planck Society budget and operates without commercial external revenue.

MPCDF operates a unified infrastructure platform anchored by three flagship supercomputers — Viper-CPU (768 AMD EPYC Genoa nodes, 98,304 cores, 4.3 PFlop/s HPL, operational June 2024), Viper-GPU (600 AMD Instinct MI300A APUs, 31.1 PFlop/s HPL, operational February 2025), and Raven (768 NVIDIA A100 GPUs, 16 PFlop/s peak) — plus an IBM TS4500-based tape archive holding 300+ Petabytes of long-term scientific data. Supporting services include high-performance data analytics (HPDA), scientific visualization, AI inference (LLM service introduced 2026), HPC application support, hosting, and structured training programs. MPCDF is the founder and core developer of the ELPA eigensolver library and contributes to FHI-aims, GENE, DFTB+, Octopus, and ESPResSo++.

The go-to-market is a captive, sales-led motion: authorized users at Max Planck Institutes apply for accounts through an electronic approval process requiring director sign-off, with usage restricted to scientific purposes. MPCDF participates in national and European research-data and exascale consortia (FAIRmat, NFDI-MatWerk, Base4NFDI, BioExcel, Plasma-PEPSC, MaMMoS, PUNCH4NFDI, RDA Deutschland), and runs joint training with HLRS and LRZ. The organization has 39 employees and is governed under the MPG Board of Directors (President Prof. Dr. Patrick Cramer; Secretary General Dr. Simone Schwanitz). It was formerly known as RZG (Rechenzentrum Garching) before being rebranded to reflect its expanded data-science mandate.

Name

Max Planck Computing and Data Facility (MPCDF)

Legal name

Max Planck Computing and Data Facility (MPCDF)

Website

https://mpcdf.mpg.de

Company type

Private

Operating status

Operating

Headcount range

11–50 employees

Short description

Max Planck Computing and Data Facility (MPCDF) is a non-profit, centralized HPC and data-management facility of the Max Planck Society that operates flagship supercomputers and 300+ PB of archival storage for Max Planck Institutes' computational and data-science research.

Ownership category

akta.pro rank

Product category

Research Computing Services

NAICS

Computer Facilities Management Services (541513), Computer Systems Design and Related Services (5415), Computer Systems Design and Related Services (54151)

SIC

Services-Computer Programming, Data Processing, Etc. (7370), Electronic Computers (3571), Services-Computer Processing & Data Preparation (7374)

akta.pro primary industry

HPC Supercomputers & Turnkey Systems (HDACAOAA)

akta.pro secondary industries

GPU-Accelerated HPC Systems (HDACAOAC), AI/HPC Converged Systems (Hybrid Training/Simulation) (HDACAOAD), HPC System Software & Resource Management (Schedulers, MPI, OS Images) (HDACAOAH), HPC / AI-Optimized Data Center Facilities (HDABACAJ)

Keywords

• High-performance computing
• Scientific data management
• Supercomputing services
• Research data archival
• HPC application support

Headquarters

HQ city

Garching

HQ country

Germany

HQ region

Europe

Offices1 record

Markets served

GTM type

B2B

Offering type

Services

Cost components

Infrastructure, Personnel, Technology or R&D, Operations

Revenue model

1. Institutional Funding: MPCDF operates as a central facility of the Max Planck Society, funded through institutional budgets. Costs for DP facilities and services are calculated from running costs based on utilization. Direct costs for compute/storage systems are cleared with respective institutes. The Board approves cost distribution yearly.

Go-to-market motion1 record

Distribution channels1 record

Marketing channels5 records

Core offering

MPCDF operates state-of-the-art supercomputers, mid-range compute clusters, and data repositories for Max Planck Institutes, providing high-performance computing, data management and long-term archival services. It also delivers specialized HPC application development and optimization support (e.g., for ELPA, FHI-aims, GENE, GROMACS, DFTB+, Octopus) together with AI/data analytics, scientific visualization, hosting, training and helpdesk services for fundamental scientific research.

Product overview

MPCDF is a cross-institutional competence centre of the Max Planck Society that operates a unified platform of HPC and data management services for Max Planck Institutes. The core infrastructure consists of flagship supercomputers (Viper-CPU, Viper-GPU, Raven) providing computing, data storage, AI/analytics, and visualization capabilities. These are complemented by user-facing services including a Selfservice portal, Helpdesk, Technical Documentation, and Training programs. The platform integrates specialized offerings such as the ELPA eigensolver library, LLM inference service, scientific visualization, and long-term data archival via tape libraries. Application support covers optimization of scientific codes (FHI-aims, GROMACS, DFTB+, GENE, Octopus) for domains including astrophysics, fusion research, materials science, and bio sciences.

Differentiator

Problem solved

Functional benefit

Products and services

• Computing Services (Supercomputing) High-performance computing service providing Max Planck Institutes with access to Viper-CPU, Viper-GPU and Raven supercomputers, CPU/GPU compute clusters, application development and optimization support for scientific research.
• Data Services Data management infrastructure including IBM TS4500 tape libraries with 105,000+ LTO slots and 200+ LTO drives, HPSS-based long-term archives and IBM Spectrum Protect backups storing 300+ petabytes for Max Planck scientists.
• AI & Data Analytics High-Performance Data Analytics (HPDA) service offering distributed training of neural networks, hyperparameter optimization, machine-learning analysis of HPC performance data, and LLM inference workloads for Max Planck researchers.
• Scientific Visualization Scientific visualization service for simulation and experimental data, including VTK-based visualization and VR-based analysis of large-scale turbulence simulations for Max Planck researchers.
• HPC Application Support High-level consulting and support for development, optimization, analysis and visualization of HPC and HPDA applications, covering scientific codes including ELPA, FHI-aims, GROMACS, DFTB+, GENE and Octopus for Max Planck domain scientists.
• Hosting & Housing Virtual-machine hosting and housing services for MPCDF users, with self-administered services available under maintenance agreement.
• Training & Education Training programs for Max Planck researchers including monthly Meet MPCDF seminars, semi-annual Introduction to MPCDF services, GPU/OpenMP programming workshops, and joint courses with HLRS and international HPC summer schools.
• LLM Inference Service Large language model inference service hosted on MPCDF infrastructure and offered to Max Planck researchers for AI workloads.
• Viper-CPU Supercomputer CPU-based supercomputer based on AMD EPYC Genoa 9554 processors used by Max Planck scientists for large-scale scientific simulations.
• Viper-GPU Supercomputer GPU-accelerated supercomputer based on AMD Instinct MI300A APUs used by Max Planck scientists for AI, HPDA and simulation workloads.
• Raven Supercomputer GPU-accelerated supercomputer based on Intel Xeon IceLake-SP processors and NVIDIA A100 GPUs: 1,592 CPU nodes, 114,624 CPU cores, 457 TB DDR4 RAM, 192 GPU nodes with 768 NVIDIA A100 GPUs, 16 PFlop/s theoretical peak performance.

Quantifiable outcome

• 31.1 PFlop/s peak performance on Viper-GPU supercomputer
• +2 more outcomes

Named customers7 records

Segments1 record

Ideal customer profiles1 record

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

AI capability4 records

Feature6 records

Partnerships

14 partnerships are on record, tiered core and minor.

• MaMMoS Project Consortium (Magnetic Multiscale Modeling Suite)coreStrategic or Co-development PartnerMPCDF participates in the MaMMoS project focused on developing integrated tools for characterization and modeling of magnetic materials at multiple scales. The project involves industrial partners (Robert Bosch GmbH, SIEMENS AG) and academic institutions. AI and ML play central roles in integrating experimental and simulation data for materials discovery.
• Plasma-PEPSC (Plasma Exascale-Performance Simulations Centre of Excellence)coreStrategic or Co-development PartnerMPCDF takes a leading role providing the technical director for this EuroHPC Centre of Excellence. The project enables scientific breakthroughs in plasma science through exascale computing, focusing on fusion plasmas, plasma accelerators, and space plasma dynamics.
• BioExcel (European Centre of Excellence for Computational Biomolecular Research)coreStrategic or Co-development PartnerLeading European Centre of Excellence developing popular applications for computational biomolecular research. MPCDF collaborates on software development, training, and providing HPC infrastructure access.
• FAIRmat ConsortiumcoreStrategic or Co-development PartnerGerman Research-Data Infrastructure consortium focused on making data Findable, Accessible, Interoperable, and Re-purposable for condensed-matter physics and chemical physics of solids.
• NFDI-MatWerk (National Research Data Infrastructure for Materials Science)coreStrategic or Co-development PartnerConsortium establishing data infrastructure for Materials Science and Engineering in Germany, enabling seamless access to high-quality data and fostering collaborative research.
• Base4NFDI (Basic Services for NFDI)coreStrategic or Co-development PartnerMPCDF is a partner in this DFG-funded project, contributing to integration and ramp-up of NFDI-wide basic services for the German science system.
• FAIR-DI e.V.coreStrategic or Co-development PartnerFAIR Data Infrastructure association for physics, chemistry, materials science, and astronomy. MPCDF is a founding member supporting Max Planck Institutes involved.
• RDA Deutschland (Research Data Alliance)coreStrategic or Co-development PartnerMPCDF is a founding and leading member of RDA Deutschland, the German chapter of the Research Data Alliance, which builds social and technical infrastructure for open data sharing.
• PUNCH4NFDIminorStrategic or Co-development PartnerNFDI consortium for particle, astro-, astroparticle, hadron and nuclear physics, representing about 9,000 scientists in Germany.
• MUDS (Munich School for Data Science)minorStrategic or Co-development PartnerMPCDF is a partner of MUDS, training next-generation data scientists at the interface of data science and application domains including plasma physics, in collaboration with MPI for Plasma Physics.
• HLRS (High Performance Computing Center Stuttgart)coreTechnology or IntegrationJoint training partner for GPU programming workshops, specifically AMD GPU training. HLRS organizes workshops together with AMD focusing on OpenMP for GPU programming using AMD MI300A APUs.
• LRZ (Leibniz Supercomputing Centre)coreTechnology or IntegrationCo-location of HPC systems (Viper-CPU deployed at LRZ) and collaboration on training courses. Courses held at LRZ Garching facility.
• DaREXA-F ProjectcoreStrategic or Co-development PartnerBMBF-funded project on data reduction for exascale applications in fusion research. MPCDF coordinates the project developing methods for reducing data traffic between compute nodes and storage systems.
• Movebank PartnersminorStrategic or Co-development PartnerMovebank animal tracking data platform hosted in coordination with North Carolina Museum of Natural Sciences, Ohio State University, and University of Konstanz. Long-term funding through MPG and University of Konstanz.

Scale indicators10 records

Recent moves6 records

Expansion highlights5 records

Direct peers

• JSC (Jülich Supercomputing Centre): German national HPC centre operating JUWELS and JURECA systems for European research. Directly comparable as a peer Tier-0/Tier-1 German HPC provider serving similar academic workloads.
• CSCS (Swiss National Supercomputing Centre): Swiss national HPC centre operating the Alps system based on NVIDIA GH200. Directly comparable as a peer European national HPC centre serving publicly-funded research.
• HLRS (High Performance Computing Center Stuttgart): National HPC center at the University of Stuttgart operating the Hazel Hen and Hunter systems, providing supercomputing for German academia. Directly comparable to MPCDF as a peer German HPC center, and an active joint training partner with MPCDF on AMD GPU programming.
• IT4Innovations National Supercomputing Center: Czech national HPC centre operating the Karolina system. Comparable as a peer European national supercomputing centre serving academic research workloads.
• CINECA: Italian non-profit HPC consortium operating Leonardo, a EuroHPC pre-exascale system. Comparable as a peer European national HPC provider serving academic research with similar institutional funding model.
• NCSA (National Center for Supercomputing Applications): US national HPC centre at the University of Illinois operating Delta and ACCESS allocations. Comparable as a peer academic-research supercomputing centre providing compute and data services to a federated research community.
• NHR (Nationales Hochleistungsrechnen / NHR Alliance): German national HPC alliance bundling compute resources from multiple university computing centres. Comparable as the broader coordinating entity for tier-2 HPC across German academia, parallel to MPCDF's tier-0/tier-1 role for Max Planck.
• LRZ (Leibniz Supercomputing Centre): German national supercomputing centre in Garching operating SuperMUC-NG. Directly comparable to MPCDF as a German HPC provider for research, and physically co-locates MPCDF's Viper-CPU system.

Broad incumbents

• Microsoft Azure HPC: Commercial cloud provider offering HPC and AI compute at large scale. Comparable as a broad incumbent that competes for scientific HPC workloads where researchers have flexibility to choose commercial versus in-house national centres.

Others

• EuroHPC Joint Undertaking: EU public partnership that pools investment in pre-exascale and exascale supercomputers across Europe. Comparable as the broader European ecosystem in which MPCDF participates through CoEs and shared compute allocations.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat6 records

Key highlights1 record

Customer concentration

• LinkedInlinkedin.com/company/rzg-max-planck-society

Revenue estimate

Valuation estimate

Number of profiles

Profiles6 records

Funding overview

Funding rounds

Investors

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Max Planck Computing and Data Facility (MPCDF)