Launched in 2024, the Alps supercomputer at the Swiss National Supercomputing Centre (CSCS) has undergone a fundamental shift, evolving from simply providing computing time to a comprehensive research infrastructure. CSCS now describes Alps as “more like an entire airport, with several runways and many terminals,” reflecting its capacity to support diverse research needs beyond traditional simulations.
This transition marks a completed move to a cloud-native system, offering tailored tools and services to researchers at ETH Zurich and across Swiss universities. “Researchers used to adapt to the supercomputer; today we adapt the supercomputer to researchers and their needs,” says Maria Grazia Giuffreda, associate director at CSCS.
Alps Supercomputer: Transition to Cloud-Native Research Infrastructure
The Alps supercomputer now facilitates the development of foundation models, AI systems trained on extensive datasets and serving as a base for specialized applications. CSCS embraced graphics processing units (GPUs) early on to accelerate computing operations, a decision that positioned the centre to capitalize on the recent surge in artificial intelligence research. The centre’s approach prioritizes adaptability; rather than requiring scientists to modify their work to fit the supercomputer’s limitations, CSCS now customizes the infrastructure to meet individual research needs.
The versatility of Alps is not solely defined by its hardware, but also by the development of algorithms, software, and integrated infrastructure. “What sets us apart is that we don’t just use the hardware – we also develop algorithms, software and infrastructure that are designed to work together,” says Torsten Hoefler.
This approach is crucial for Switzerland’s digital sovereignty, enabling the creation of new knowledge and technologies within the country. The Swiss AI Initiative, launched by ETH Zurich and EPFL in 2023, is using Alps to develop these foundation models, demonstrating the supercomputer’s central role in advancing AI research. “The technology behind high-performance computing should be smooth and streamlined, empowering researchers to focus on their research rather than the underlying infrastructure,” says Joost VandeVondele, highlighting the ultimate goal of this infrastructure transformation.
It’s not just the hardware infrastructure that makes Alps so special. The supercomputer enables CSCS and researchers to create new knowledge and new technologies.
Maria Grazia Giuffreda, associate director at CSCS, responsible for the user programme and for collaboration with
Versatile Architecture Adapts to Researcher-Specific Workflows
The Alps supercomputer’s architecture prioritizes adaptability through vClusters, a software layer dividing the hardware into flexible areas allowing researchers at ETH Zurich and institutions like MeteoSwiss to tailor their workflows. This approach moves beyond simply allocating computing time, enabling users to operate within customized environments suited to their specific needs. CSCS, the Swiss National Supercomputing Centre, designed this system to accommodate the increasing diversity of data, software, and workflows inherent in modern large-scale research.
Joost VandeVondele, deputy director for science at CSCS responsible for technical development, emphasizes that Alps was intentionally built as an integrated system where hardware, software, and specialized services function across multiple layers. Cloud services, web portals, data platforms, and AI applications build upon this foundation, simplifying access to computing power, data, and scientific applications for researchers.
Since July 2026, a new service has allowed researchers to utilize the Apertus 1.5 large language model, with Alps automatically providing the necessary computational resources. This illustrates a shift from a machine responding to requests to a proactive system anticipating and fulfilling them. Their work on Sailor, an automated AI training platform, further optimizes resource utilization by helping researchers select the most appropriate hardware and maximize efficiency.
“The more effectively we design software to make use of the hardware resources and available power in Alps, the more quickly each job in the cluster completes and hence the more jobs we get to run on Alps to explore all kinds of research goals,” Klimovic explains. This layered approach, combining adaptable hardware with intelligent software, positions Alps as a cornerstone of Switzerland’s AI research infrastructure and a driver of digital sovereignty.
The more effectively we design software to make use of the hardware resources and available power in Alps, the more quickly each job in the cluster completes and hence the more jobs we get to run on Alps to explore all kinds of research goals.
Ana Klimovic, ETH Zurich professor of computer science
Grace Hopper Superchips and GPUs Fuel AI Development
The Grace Hopper Superchips at the heart of the Alps supercomputer are important in enabling a new era of artificial intelligence research, allowing scientists to actively shape the infrastructure itself. CSCS was an early adopter of this technology, becoming the launch customer of the Grace Hopper Superchips and positioning Switzerland in a leading position for AI development as a result. This proactive approach extends beyond hardware acquisition; CSCS actively develops algorithms and software designed to maximize the potential of the Grace Hopper Superchips and the accompanying GPUs.
Torsten Hoefler, an ETH professor of computer science specialising in high-performance computing and chief architect for AI and machine learning at CSCS, highlights this integrated strategy. The development of Sailor, an automated AI training platform, exemplifies this commitment to resource optimization.
This platform assists researchers in selecting the most appropriate computing resources for their AI tasks and maximizing hardware utilization, effectively acting as a tool for the supercomputer’s processing capabilities. The Swiss AI Initiative, launched in 2023, brings together researchers from ETH Zurich, EPFL and other Swiss universities. Together, they are developing what are known as foundation models: AI models that are trained on large datasets and is the basis for more specific AI applications.
The GPUs are a key component of these chips. CSCS embraced GPU technology early on to accelerate computing operations.
Maria Grazia Giuffreda, associate director at CSCS, responsible for the user programme and for collaboration with
See today’s quantum computing news on Quantum Zeitgeist for the latest breakthroughs in qubits, hardware, algorithms, and industry deals.
