Germany

HammerHAI provides computing infrastructure and comprehensive services to facilitate the adoption and advancement of artificial intelligence in industry and research. At the core of its offerings is a high-performance, AI-optimized supercomputer designed for training and inference using complex AI models. 

The new HammerHAI supercomputer will support medium- and large-scale applications in artificial intelligence, machine learning, and data science. Hosted by the High-Performance Computing Center Stuttgart (HLRS), the system gives European industry, small and medium-sized enterprises, startups, and scientific research access to powerful AI capabilities, particularly in the engineering, manufacturing, automotive, and mobility sectors.

As a secure and scalable AI resource, the HammerHAI supercomputer will support the same types of AI workloads typically handled by commercial cloud services. Its cloud-native software stack makes it easy to migrate or scale applications from local systems or commercial cloud environments.

Beyond infrastructure and technical support, HammerHAI places significant emphasis on knowledge transfer and skills development. Through tailored training programs, workshops, and university collaborations, it fosters an AI-ready workforce, equipping professionals and researchers with the expertise needed to implement AI effectively. These educational initiatives and hands-on experimentation opportunities ensure that users can apply AI technologies in a practical and impactful manner. 

Delivery of the system is scheduled for the second quarter of 2026 and it is expected to go into operation in the second half of 2026. The HammerHAI supercomputer will be manufactured and installed by HPE, based on the liquid-cooled NVIDIA GB200 NVL4 architecture. 

Model Development Services 

• High-Speed Data Transfer, Storage, and Management: Ensuring secure and efficient management and transfer of large datasets for AI tasks 
• Virtual Development Spaces: Providing configurable, cloud-based HPC environments suitable for a range of AI workflows 
• Development of AI MVPs: Offering pre-configured architectures that facilitate rapid prototyping and initial AI product development 
• Foundation Model Catalog & Custom Hosting: Delivering access to foundation models and other pre-trained models along with custom model hosting capabilities 
• Task-Specific Model Derivation and Fine-Tuning: Supporting the development and refinement of customized ML models derived from large-scale foundation models 
• HPC Model Training: Providing resources and infrastructure for training of ML models requiring significant computational power 
• Reinforcement Learning: Supplying efficient execution of simulation environments and training for reinforcement learning experiments 
• Model Performance Improvement: Implementing structured processes and tools to systematically analyze and improve model effectiveness 
• Hyperparameter Optimization & Meta Experiments: Offering automated experimentation frameworks for extensive hyperparameter tuning 

Data Services 

• Data Acquisition: Guiding the process of identifying and sourcing relevant datasets for AI applications powered by JAIFs data spaces 
• Data Annotation & Preparation: Automating data annotation, preprocessing and augmentation with high performance resources to minimize manual effort 
• Synthetic Data Generation: Producing synthetic datasets through generative AI and simulation techniques to enhance available data resources 

Productisation Services 

• Model Complexity Reduction: Providing methods and tools to streamline AI models for effective deployment on target hardware 
• Collaborative Model Benchmarking & Validation: Offering benchmarking frameworks to objectively assess model performance against competing applications 
• Model Deployment: Creating standardized, automated deployment workflows applicable across various hardware platforms 
• Monitoring and Performance Tracking: Establishing systems for ongoing monitoring and real-time analysis of AI applications 
• Legal Assessment for AI Products: Assisting with regulatory compliance through advisory services and automated compliance checking tools 

Onboarding Services 

• Introductory Self-Learn Resources: Providing educational materials to facilitate understanding and initial use of AI technologies 
• User-Centered Development Interfaces: Designing user-friendly interfaces accessible to varying levels of technical expertise 
• Co-working with AI Experts: Offering structured collaboration opportunities including physical co-working spaces with AI experts for knowledge sharing and joint development  

Consulting Services 

• Exploration Consulting: Conducting consultations to identify suitable opportunities for AI application within specific business contexts 
• Feasibility Assessment & Roadmapping: Evaluating technical feasibility and developing clear implementation roadmaps 
• Lifecycle Consulting: Providing comprehensive strategic guidance through each phase of AI product development 

HPC Optimization Services 

• First-Level Support: Offering initial guidance to facilitate user access to HPC systems 
• HPC Deployment and Monitoring: Supporting deployment and performance monitoring of AI workflows on HPC infrastructure 
• Parallelisation of Training and Inference: Advising on techniques to effectively parallelize computational workloads for ML training and inference 
• Reduction of Memory Usage: Addressing memory limitations through specialized optimization techniques 
• HPC-Efficient Application Scale-Out: Providing solutions for effectively scaling applications on HPC systems