How ODISSEE Is Preparing Europe for Exabyte-Scale Scientific Computing

Key Takeaways

• CERN LHCb generates ~1 PB/s, SKA 2 PB/s raw data.
• ODISSEE develops flexible, energy‑efficient data‑centric architectures.
• Consortium shifts from coordination to real‑world workload integration in 2026.
• Open‑source RISC‑V and sovereign hardware reduce reliance on imports.

Summary

The EU‑funded ODISSEE project, launched in 2025 under Horizon Europe, aims to create data‑centric, exabyte‑scale computing solutions for CERN’s LHCb and the SKA Observatory. A diverse consortium—including CERN, SKAO, CNRS, SURF, EPFL, ETH Zurich, SiPearl, Energy Aware Solutions and NextSilicon—spent its first year aligning goals and building baseline infrastructure. In 2026 the effort moves to real‑world integration, leveraging NextSilicon’s Maverick‑2 accelerators and exploring open‑source RISC‑V hardware. SiPearl’s recent €130 million (~$143 million) Series A underscores Europe’s push for sovereign HPC.

Pulse Analysis

Exabyte‑scale data generation is no longer a theoretical future; CERN’s LHCb already streams roughly one petabyte per second, and the upcoming SKA Observatory will double that rate. Traditional CPU‑GPU clusters cannot sustain such volumes without prohibitive energy costs, prompting a shift toward data‑centric designs that move computation closer to storage and exploit burst‑driven workloads. This paradigm shift is critical for maintaining scientific throughput and for meeting Europe’s climate‑aware computing targets.

ODISSEE’s consortium blends top‑tier research institutions with cutting‑edge technology firms to prototype these new architectures. NextSilicon contributes Maverick‑2 accelerators that claim four‑times the performance‑per‑watt of leading GPUs, while partners explore reconfigurable dataflow engines and RISC‑V based processors that promise a sovereign, open‑hardware stack. The project’s €130 million (~$143 million) funding round, highlighted by SiPearl’s Series A, signals strong investor confidence in Europe’s ability to compete globally in high‑performance computing.

Looking ahead, 2026 will see ODISSEE transition from planning to deployment, with a June hackathon focused on tightly coupling fast interconnects with compute kernels. Early integration results on LHCb and SKA workloads will provide concrete evidence of energy‑efficient exascale performance. Success will not only accelerate discoveries in particle physics and radio astronomy but also set a template for other data‑intensive domains worldwide, reinforcing Europe’s role as a leader in sustainable, sovereign scientific computing.