The next phase of artificial intelligence may depend less on bigger models and more on a fundamental redesign of computing itself. That message sits at the centre of preparations for ISC High Performance 2026 in Hamburg, where researchers, engineers and industry leaders will gather to confront what many now see as the defining challenge of the post-Moore era.
Organisers have confirmed that Professor Dr Martin Schulz, chair of Computer Architecture and Parallel Systems at the Technical University of Munich and board member at the Leibniz Supercomputing Centre, will deliver the opening keynote of the conference in June. His address, titled HPC A heterogeneous future, will argue that the traditional path of performance improvement that underpinned modern computing for decades has reached its limits.
For much of the past half century, advances in computing relied on Moore’s Law and Dennard Scaling, predictable gains that allowed processors to become faster and more efficient with each generation. According to the conference programme, those assumptions are now breaking down. Energy consumption, memory bandwidth and data movement have emerged as the primary constraints, creating what Schulz describes as a “complexity wall” facing high performance computing.
The implications extend far beyond scientific simulation. As artificial intelligence systems grow in scale and ambition, their performance increasingly depends on the same architectural foundations that underpin supercomputing.
The post Moore computing challenge
Schulz is expected to argue that incremental improvements, including the industry’s transition towards graphics processing units, are no longer sufficient to sustain progress. Instead, computing must move towards integrated heterogeneity, combining multiple architectures rather than relying on a single dominant processor design.
Emerging technologies such as quantum computing, neuromorphic systems and photonics will form part of this discussion. Rather than viewing architectural diversity as a complication, the keynote proposes that combining different computing paradigms through unified software stacks could unlock new performance gains while maintaining accessibility for researchers and developers.
This shift reflects how AI workloads are evolving. Training and inference increasingly demand specialised hardware, advanced memory systems and tightly integrated software environments. The convergence between AI and high performance computing means architectural decisions once confined to research laboratories now influence industrial competitiveness and scientific discovery alike.
Schulz brings extensive experience to the debate, having previously held positions at Lawrence Livermore National Laboratory and Cornell University. His research spans parallel architectures, performance modelling, power-aware computing and quantum integration, and he currently chairs the MPI Forum, the international body responsible for standardising the Message Passing Interface used widely in supercomputing.
AI drives a broader computing community
Alongside the keynote, ISC 2026 organisers have unveiled a programme of 24 workshops covering high performance computing, artificial intelligence and quantum computing. The sessions are designed to bring together academic researchers, infrastructure engineers and industry practitioners to address technical challenges emerging from increasingly complex computing environments.
Some workshops will produce peer-reviewed papers published in Springer’s Lecture Notes in Computer Science series, while others will focus on operational experience and collaborative discussion. The structure reflects a growing recognition that AI progress depends not only on theoretical advances but also on shared engineering practice.
The workshops will take place at the Congress Center Hamburg and are intended to serve a wide audience, from researchers presenting new work to practitioners evaluating emerging architectures and students entering the field.
The emphasis on skills development has also led to the creation of a new ISC Next Generation Committee, an advisory and operational body aimed at strengthening engagement with students and early-career professionals. Organisers say the committee responds to sustained feedback from younger attendees seeking clearer pathways into the high performance computing community.
The group will focus on improving visibility for emerging talent, reducing participation barriers and ensuring conference content remains relevant to the next generation of practitioners. Members include postgraduate students, early-career researchers and industry analysts, signalling an effort to formalise intergenerational collaboration within the field.
Taken together, the announcements highlight a moment of transition for advanced computing. As AI systems push beyond the limits of conventional architectures, supercomputing conferences such as ISC are increasingly becoming forums not just for scientific computing, but for defining the infrastructure foundations of the AI era itself.
The question facing the sector is no longer simply how to build faster machines. It is how to design computing systems capable of sustaining intelligence at scale once the assumptions that powered decades of progress no longer apply.
