// ITS FOSS — LINUX & OPEN SOURCE
openEuler and the Race to Build Asia's AI Infrastructure
When it comes to open source and AI, we almost always look West. While a lot happens in the United States and some in Europe, the East is not lagging either.
In the recently concluded LEAP East conference, several startups and enterprises came together to showcase their innovations and share what they were building.
Out of them, openEuler caught our attention for its focus on Open Source AI.
At LEAP East 2026, openEuler showcased its SuperPoD OS, which extends operating-system support for UnifiedBus-based SuperPoD computing systems.
Think of a SuperPoD as a cluster of several powerful computers that need to act like one giant machine. Running AI workloads takes computing power and doing it across hundreds of such machines is messy as they need to share memory, pass data around quickly, and not conflict with each other.
UB Service Core is the open-source software layer, part of the openEuler ecosystem, that makes coordination across those machines actually work. It provides five cluster-level system services for UnifiedBus-based SuperPoDs: Engine, Memory, Communication, I/O, and Virtualization.
The Engine pools memory and DPU resources across the cluster and schedules them dynamically. It also handles automatic failover so the system keeps running if a node goes down.
The Memory component lets machines within a UnifiedBus-based SuperPoD share and pool memory using unified-memory programming, meaning software can treat distributed memory as a single addressable space.
The Communication piece provides high-performance networking over UnifiedBus while staying compatible with standard Socket and Verbs interfaces, so existing applications don't need to be rewritten to benefit from it.
The I/O service handles global data read/write caching at a high level, which matters for workloads like AI inference that repeatedly access large shared datasets.