News & Events

Supercomputing Asia / HPCAsia 2026 brought together leading experts, policymakers, and infrastructure providers from across the Asia-Pacific and beyond to discuss the future of high-performance computing and artificial intelligence. As one of the region’s key annual gatherings for HPC and AI, the conference serves as a platform for strategic dialogue between academia, industry, and governments. Within this framework, the EU–India collaboration project GANANA organized a dedicated workshop on “Bridging continents: multilateral cooperation between EU, Japan, India, Singapore, Korea and Australia”. The workshop was organized jointly with INPACE, the Indo-Pacific European Hub for Digital Partnerships.

The workshop made one thing clear: international cooperation in HPC and AI is no longer primarily a hardware challenge but a human and institutional one. Across all participating regions, investments in high-performance computing and AI infrastructure are accelerating. GPU-powered systems, AI-focused platforms, and even quantum–HPC hybrids are coming online at an unprecedented pace. Yet despite this progress, the discussions showed that access to compute is not the main bottleneck to impact. Instead, three structural challenges dominate.

First, capacity.

Supercomputing centres across continents are seeing a rapid influx of AI-driven users, many coming from cloud-native environments. While AI frameworks are technically easier to deploy than traditional HPC codes, efficient and scientifically meaningful use of large-scale systems remains difficult. Training, workflow optimisation, and domain integration are now critical. At the same time, public infrastructures face increasing difficulty retaining skilled staff, as industry offers significantly higher salaries and longer-term stability.

Second, sustainability and effectiveness.

Measuring success purely by system utilisation is no longer sufficient. Centres are increasingly looking at effective GPU usage, energy efficiency, and scientific or industrial outcomes. The conversation is shifting from “Are the machines full?” to “Are they being used well, and to what end?”

Third, cooperation models.

Bilateral collaboration is common and often productive. Scaling to multilateral cooperation, however, introduces complexity: fragmented funding mechanisms, regulatory constraints, visa regimes, and short project cycles limit continuity. Informal agreements without dedicated funding rarely produce lasting results.

Rather than advocating for a single global infrastructure, the workshop pointed toward a pragmatic model: interconnected but sovereign regional ecosystems. Cooperation works best when focused on capacity building, shared methodologies, benchmarking, and training. Those are the areas where collaboration strengthens all partners without compromising autonomy.

The overall conclusion was that bridging continents in HPC and AI will not be achieved by connecting machines alone. It requires aligning incentives, investing in people, and building trust over time. Infrastructure is essential but ecosystems are what ultimately determine impact.