• Lablup will present Backend.AI’s integration with the Kubernetes ecosystem at Booth 216
• The company continues to strengthen integrations with related ecosystems, including Slurm Frontend and K8s Bridge

Lablup (KOSDAQ: 0147B0), an AI infrastructure software company led by CEO Jeongkyu Shin, will participate as a Silver Sponsor in KubeCon + CloudNativeCon North America 2026, taking place in Salt Lake City, Utah, from November 9 to 12.

During the event, Lablup will operate Booth 216 and present the current state of integration between Backend.AI, its AI infrastructure management platform, and the Kubernetes (K8s) ecosystem. Backend.AI supports AI accelerators from 12 vendors, including NVIDIA, AMD, Intel, Rebellions, and FuriosaAI. Kubernetes is an open source platform for deploying and operating container-based services.

Backend.AI has isolated container-based work environments without compromising the native performance of servers, even before Kubernetes released version 1.0. Building on this foundation, Lablup developed Sokovan, its proprietary resource management engine for AI workloads. Sokovan supports high-performance computing (HPC) and AI workloads that combine multiple types of AI accelerators and GPUs, as well as multiple servers. In an independent AI foundation model project with Upstage, it has reliably supported AI model training across 1,016 GPUs.

The industry has traditionally operated large-scale AI model training across multiple servers using Slurm, the de facto standard job management tool in supercomputing environments, while running web services on Kubernetes. Over the past two years or so, AI inference using GPUs has emerged as a major component of services in the Kubernetes ecosystem, accelerating efforts to standardize related technologies. In line with this standardization trend, Backend.AI has continued to strengthen its integration with both ecosystems.

For integration with supercomputing environments, Lablup provides Slurm Frontend. Researchers can continue using familiar Slurm commands, while Slurm Frontend executes the required workloads in Backend.AI’s container environment. Lagrange, the K8s Bridge, registers Backend.AI as a virtual server in Kubernetes and synchronizes the two systems bidirectionally, allowing Backend.AI resources to be viewed through Kubernetes management tools. Lablup is improving the user experience across both environments: running Backend.AI on top of Kubernetes and installing it directly on customers’ own servers or virtual machines (VMs). Many customers currently use Backend.AI directly on their own servers or VMs.

Lablup joined the Cloud Native Computing Foundation (CNCF) as a Silver Member in 2020 and has participated in KubeCon as a Silver Sponsor for three consecutive years since 2024.

When infrastructure providers that lease resources like cloud services and user organizations that consume them operate AI workloads under separate management systems, issues often arise over how much direct hardware control should be granted for performance optimization and security. Through its activities in the TTA and AIPP standardization groups, Lablup is examining this hardware access control issue in greater depth.

“Even with Kubernetes, this issue is not fully resolved,” said Joongi Kim, Chief Technology Officer at Lablup. “The industry needs to think more deeply about how to reduce information asymmetry between providers and tenants while maintaining security isolation.” He added, “KubeCon is a place to see what is happening at the forefront of the CNCF ecosystem. Based on the experience Lablup has accumulated operating AI infrastructure, we will contribute to helping the Kubernetes ecosystem address this challenge.”

Lablup was listed on the KOSDAQ market on November 3 and supplies its Backend.AI platform to more than 120 customer sites in Korea and overseas.