Lablup Inc. develops AI infrastructure software for organizations that need to share, schedule, and govern accelerated compute at scale. Its portfolio centers on Backend.AI, a software-defined platform for orchestrating heterogeneous GPU and NPU clusters with secure multi-tenancy, container-level GPU virtualization, and multi-node workload management.
Around that core, Lablup Inc. has built an ecosystem spanning MLOps pipelines, model serving, air-gapped software distribution, local AI clients, and browser-based AI creation experiences. It positions its software to reduce infrastructure friction for teams running training, inference, and high-performance computing workloads across on-premises, cloud, and hybrid environments.
Offerings, Capabilities, and Integrations
Lablup Inc. focuses on the operational layer of AI infrastructure. Its software abstracts heterogeneous accelerators and storage into a unified control plane, applies policy-based governance, and supports secure multi-tenant execution from fractional GPU sharing to large multi-node jobs.
The broader stack supports end-to-end AI operations, including workflow orchestration, model serving, cluster monitoring, and software supply for isolated environments. Lablup Inc. also aligns its platform with hardware, storage, and cloud ecosystems so customers can deploy AI workloads on the infrastructure mix that best fits their performance, security, and procurement requirements.
Products and Services
- Backend.AI: Flagship AI infrastructure operating platform for orchestrating heterogeneous GPU and NPU clusters. It combines secure multi-tenancy, container-level GPU virtualization, multi-node workload management, and support for on-premises, cloud, hybrid, and air-gapped deployments.
- Backend.AI FastTrack 3: MLOps pipeline platform built on Backend.AI clusters. It defines and runs AI and ML workflows as DAGs, covering data preparation, training, validation, deployment, and monitoring with both GUI and CLI-driven operation.
- Backend.AI:GO: Free desktop AI application for running models locally on macOS, Windows, and Linux. It supports offline use, local model management, agent capabilities, multi-device mesh operation, and optional offload to Backend.AI clusters or cloud APIs.
- Sokovan: Acceleration-aware container orchestrator and scheduler used within the Backend.AI stack. It is designed for multi-tenant, batch-oriented AI workloads and dynamically allocates resources across cluster and node levels.
- Reservoir: Local package repository and update service for Backend.AI environments. It is designed for private cloud, offline, and air-gapped deployments that need controlled delivery of validated packages inside internal networks.
- All-SMI: Unified monitoring tool for heterogeneous accelerator and server environments. It gives operators a single interface for observing resource usage and infrastructure health across mixed hardware estates.
- PALI: Inference software for fast deployment and operation of deep learning and generative AI models. It is positioned for organizations that want packaged model serving with support for open model sources and deployment in on-premises or air-gapped environments.
- AI:DOL: AI-native platform for creating and prototyping AI services in a browser using models deployed on Backend.AI. It emphasizes natural-language interaction, low-code creation, and integration of on-premises and cloud resources.
Target Customers
Lablup Inc. serves organizations that operate shared AI infrastructure rather than isolated single-user systems. Its software fits enterprises, research institutes, universities, public-sector bodies, and cloud or service providers that need to allocate accelerator resources efficiently, enforce tenancy controls, and support both experimentation and production workloads.
It is especially relevant for teams in sectors such as telecom, finance, manufacturing, healthcare, defense, and advanced research that run mixed training, inference, and HPC-style jobs across varied hardware. Lablup Inc. also addresses developers and smaller teams that want local-first AI tooling or a simpler path from prototype to deployable AI service.
Cloud Integrations and Marketplace
- Microsoft Azure Marketplace: Lablup Inc. has a verified marketplace presence through its PALI listing, providing Azure buyers with a packaged route to evaluate and procure its inference software.
Key People
- Jeongkyu Shin: CEO
- Joongi Kim: CTO
- Jiwon Hwang: CFO
- Jeongmook Kim: CRO
- Jonghyun Park: Research Director
- Eunjin Hwang: Principal Researcher and Research Team Lead
- Jongeun Lee: FrontendDev Team Lead
- Hyeokjin Kim: CoreDev Team Lead
Key Facts
- Headquarters: Gangnam-gu, Seoul, Republic of Korea
- Employees: About 40 employees
- Annual Revenue: ₩6.8B
- Parent Company: None
- Subsidiaries: Lablup USA, Inc.
- Publicly Listed: No
