Lablup
Lablup builds Backend.AI, a software-defined AI infrastructure platform that orchestrates heterogeneous GPU and NPU clusters for ~120 enterprise, government, and research customers across Korea, North America, and EMEA, selling cloud subscriptions, on-premises licenses, and professional services.
- Company typePrivate
- Founded2015
- HeadquartersSeoul, South Korea
- Headcount11–50
- GTM typeB2B
- OfferingSoftware
What Lablup does
Lablup Inc. is a South Korea-based AI infrastructure software company founded in 2015 that builds Backend.AI, a software-defined orchestration platform for heterogeneous GPU and NPU clusters. The platform's core technical assets include patented container-level GPU virtualization (which partitions a single GPU across multiple containers to maximize utilization), the in-house Sokovan workload orchestrator, and the MLxcel inference engine optimized for Apple Silicon. Backend.AI is validated as NVIDIA DGX-Ready Software, supports Intel Gaudi 2/3 and Intel Arc Pro B70 accelerators, and integrates with high-performance storage systems from VAST Data, NetApp, WEKA, and Pure Storage, enabling a single control plane to manage workloads from a single workstation to hyperscale clusters.
The company sells to enterprises, research institutions, government agencies, and universities across Korea, North America, EMEA, and Southeast Asia through a mix of direct enterprise sales, channel partnerships (Boston Limited for EMEA, KT Cloud for Korea), and a free self-serve desktop product (Backend.AI:GO) that runs small language models offline on personal PCs. Revenue streams include Backend.AI Cloud subscriptions (credit-based, 1 Credit = KRW 1), on-premises software licenses, professional services delivered by Solution Architects, and a GPU-as-a-Service joint offering with KT Cloud. Named customers span Samsung Electronics, Samsung Research, Samsung Medical Center, Hyundai Mobis, LG Electronics, KT, KT Cloud, Shinhan Bank, Bank of Korea, Korean Air, LIG Defense & Aerospace, ROK Navy, ETRI, KISTI, HIRA, SKKU, USC, and others, with the platform orchestrating 16,000+ GPUs across 110+ sites for roughly 120 organizations globally. The company is privately held, has raised approximately $9.7M in disclosed funding, employs 11-50 staff, and operates from Seoul with a U.S. subsidiary in San Jose established January 2025.
Lablup firmographics
Firmographics- Name
- Lablup
- Legal name
- Lablup Inc.
- Website
- https://lablup.com
- Company type
- Private
- Founded year
- 2015
- Operating status
- Operating
- Headcount range
- 11–50 employees
- Short description
- Lablup builds Backend.AI, a software-defined AI infrastructure platform that orchestrates heterogeneous GPU and NPU clusters for ~120 enterprise, government, and research customers across Korea, North America, and EMEA, selling cloud subscriptions, on-premises licenses, and professional services.
- Ownership category
- akta.pro rank
Lablup industry classification
Industry- Product category
- AI Infrastructure Platform
- NAICS
- Computer Systems Design and Related Services (54151), Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (5182)
- SIC
- Services-Prepackaged Software (7372), Services-Computer Integrated Systems Design (7373)
- akta.pro primary industry
- AI Compute Virtualization & Scheduling (GPU virtualization, cluster schedulers) (HDAAAAAG)
- akta.pro secondary industries
- Model Hosting, Serving & Inference Platforms (HDAAACAB), AI Compute Cloud & GPU-as-a-Service (HDAAAAAK), End-to-End Enterprise AI Platforms (MLOps & Model Lifecycle Management) (HDAEANAA)
Keywords
Where Lablup is headquartered
LocationHeadquarters
- HQ city
- Seoul
- HQ country
- South Korea
- HQ region
- Asia
Offices2 records
Markets served
Lablup business model
Business model- GTM type
- B2B
- Offering type
- Software
- Cost components
- Technology or R&D, Personnel, Marketing or Sales, Infrastructure, Operations
Revenue model
- Backend.AI Cloud Subscription: Cloud-based AI infrastructure services where users pay for telecommunication resources through hardware and software accessed via external cloud servers. Members can use prepaid credits (1 Credit = KRW 1) or monthly billing based on usage.
- On-premises Software Licenses: Organizations can install and operate Backend.AI on-premises through software licenses. Boston Limited distributes on-premises licenses directly to end customers in EMEA region along with installation, configuration, and consulting services.
- Professional Services: Solution Architects and Solution Engineers provide deployment consulting, technical support, installation, configuration, and ongoing monitoring services for enterprise clients. Boston Limited offers installation, configuration, and consulting services for EMEA customers.
- GPU as a Service (GPUaaS): Joint development with KT Cloud for GPU as a Service solutions, enabling customers to access GPU resources on-demand through cloud delivery model.
Go-to-market motion4 records
Distribution channels6 records
Marketing channels8 records
Lablup product offering
Product offeringCore offering
Lablup develops and sells Backend.AI, a software-defined AI infrastructure operating platform that orchestrates heterogeneous GPU and NPU clusters at hyperscale with secure multi-tenancy and multi-node workload management. The company extends this enterprise platform with Backend.AI:GO, a free desktop application that runs small language models offline on personal PCs, and Backend.AI:DOL, a browser-based cloud solution for accessing open AI models, providing full-stack coverage from individual workstations to hyperscale data centers.
Product overview
Lablup offers Backend.AI as its core AI infrastructure platform—a software-defined orchestration system that manages heterogeneous GPU and NPU clusters at hyperscale with secure multi-tenancy and multi-node workload management. The portfolio extends to personal AI solutions: Backend.AI:GO (free desktop application for running SLMs offline on personal PCs) and Backend.AI:DOL (cloud-based model access via web browser), completing full-stack coverage from AI PCs to hyperscale GPU clusters. Key add-on modules include Backend.AI FastTrack 3 (data preparation to serving), Backend.AI Continuum (fault tolerance and model routing), Sokovan (proprietary workload orchestrator), and MLxcel (Apple Silicon inference engine). The platform supports NVIDIA, AMD, Intel, and domestic NPU accelerators through a single control plane, validated as NVIDIA DGX-Ready Software.
Differentiator
Problem solved
Functional benefit
Brands
- Backend.AI: Software-defined AI infrastructure platform that orchestrates heterogeneous GPU and NPU clusters at hyperscale with secure multi-tenancy and multi-node workload management.
- Backend.AI:GO
- Backend.AI:DOL
- FastTrack
- Sokovan
Products and services
- Backend.AI Software-defined AI infrastructure operating platform that orchestrates heterogeneous GPU and NPU clusters at hyperscale with secure multi-tenancy and multi-node workload management. Features patented container-level GPU virtualization for NVIDIA GPUs and integrates with leading high-performance storage systems including VAST Data, NetApp, and WEKA. Targets enterprises, research institutions, and government agencies requiring scalable AI infrastructure.
- Backend.AI:GO Free desktop application that runs Small Language Models (SLMs) entirely offline on personal PCs with no cloud connection required. Supports macOS, Windows, and Linux with native Apple Silicon optimization through the MLxcel inference engine. Features Mesh Mode for distributed inference across multiple devices, agent marketplace, and Cowork mode for multi-agent collaboration. Targets individual developers, researchers, and AI PC users.
- Backend.AI:DOL Cloud-based solution that provides web browser access to various open AI models. Deployable Omnimedia Lab that allows users to access and run AI models directly through a browser without local installation, complementing on-premises deployment options. Targets users seeking easy cloud access to open-source AI models.
- Backend.AI FastTrack 3 Workflow acceleration tool that streamlines the data preparation to model serving pipeline, enabling faster AI model deployment and experimentation on the Backend.AI platform. Targeted at data scientists and ML engineers needing rapid iteration cycles.
- Backend.AI Continuum
Quantifiable outcome
- 47% reduction in average recovery time for distributed training failures, with failed processes restarting in under 3 seconds
- +2 more outcomes
Companies that use Lablup
Customer profileNamed customers23 records
Segments4 records
Ideal customer profiles3 records
Lablup technology and API
TechnologyTechnology focussed Yes
API detail
- Has API
- No
- API docs
- API detail
Core technology
AI maturity
App detail
Integration15 records
AI capability9 records
Feature8 records
Lablup partnerships and signals
Strategic signalPartnerships
19 partnerships are on record, tiered flagship, strategic, minor and core.
- Upstage ConsortiumflagshipNamed one of Top 5 elite teams in Korea's Sovereign AI Foundation Model Project. Lablup serves as the AI infrastructure partner for Upstage's Solar Open 100B training, operating a 60+ node B200 cluster (504 GPUs) for 73 days.
- Agentic AI Foundation (AAIF)strategicSilver Member of the Agentic AI Foundation, co-founded by Anthropic, Block, and OpenAI under Linux Foundation in December 2025. Lablup contributes AI infrastructure and workload orchestration expertise. Plans to adopt MCP standard and help grow open agent ecosystem in South Korea.
- Boston LimitedflagshipStrategic distribution partnership to bring Backend.AI platform to EMEA market. Boston Limited offers cloud services (SaaS) and on-premises software licenses with installation, configuration, and consulting services. Partnership signed at MWC 2026 in Barcelona. Combines Boston's HPC infrastructure expertise with Lablup's software platform.
- Knovel Engineering Pte LtdminorMOU signed for AI infrastructure collaboration in Singapore region. Partnership to distribute and support Backend.AI in Southeast Asian markets.
- PTC System (S) Pte LtdminorMOU signed for AI infrastructure collaboration in Singapore region. Partnership to support Backend.AI deployment and operations in Southeast Asian markets.
- Python Software Foundation (PSF)strategicParticipating Sponsor since 2026. Backend.AI and entire tech stack built on Python. Contributing to stewardship of Python language and ecosystem.
- VAST DatacoreJoined VAST Cosmos Partner Program as Technology Alliance Partner. Integration between Backend.AI and VAST AI Operating System enables end-to-end AI workload orchestration from training to deployment and inference. High-throughput data access through RDMA and GPU-direct data paths powered by VAST DataStore. Validated architectures for AI-ready solutions.
- KT CloudstrategicMOU signed for joint development of GPUaaS (GPU as a Service) solutions. Collaboration to offer cloud-based GPU services leveraging Backend.AI platform.
- Linux FoundationstrategicSilver Member since 2020. Contributing to open-source infrastructure projects that power AI servers. Home of the Linux kernel and critical open-source infrastructure.
- Cloud Native Computing Foundation (CNCF)strategicSilver Member since 2020. Kubernetes and cloud-native projects integrated with Backend.AI at every layer. Foundation home for container orchestration standards.
- NVIDIAcoreNVIDIA Inception member since 2018. Backend.AI achieved DGX-Ready Software certification (first in Asia-Pacific) and NVIDIA AI Accelerated Program membership. Container-level GPU virtualization technology works with NVIDIA GPUs for intelligent resource distribution. Collaboration includes GPU orchestration, multi-node multi-tenant workload management, and GPU network auto-control.
- IntelcorePartnership to bring Intel Gaudi AI Accelerators to Backend.AI. Sokovan orchestrator distributes workloads across Gaudi compute engines. Support for Gaudi 2 and Gaudi 3 platforms. Container-level virtualization for fine-grained accelerator resource management. Joint benchmarking and solution briefs for Gaudi performance optimization.
- NetAppcorePartner for high-performance storage integration. Organizations can integrate with leading high-performance storage systems for more efficient infrastructure operation.
- WEKAcorePartner for high-performance storage integration. WEKA NeuralMesh acceleration available for HPC & AI workloads.
- MicrosoftcoreTechnology partner for AI infrastructure solutions. Integration and compatibility with Microsoft cloud and enterprise platforms.
- AWScoreTechnology partner. Backend.AI cloud services available and integrated with AWS cloud infrastructure.
- Dell TechnologiescoreTechnology partner for AI infrastructure hardware. Integration with Dell GPU servers and HPC infrastructure solutions.
- VultrcoreCloud partner. Backend.AI hands-on workshops conducted using Vultr GPU instances. Scalable AI and HPC solutions available on Vultr.
- Everpure (Pure Storage)coreStorage partner for high-performance all-flash storage integration with AI infrastructure.
Scale indicators9 records
Recent moves12 records
Expansion highlights6 records
Lablup competitors and assessment
Company assessmentDirect peers
- Run:ai: Run:ai (acquired by NVIDIA in 2024 for ~$700M) provided GPU orchestration and fractional GPU sharing for AI workloads—directly comparable to Lablup's Backend.AI platform and patented container-level GPU virtualization.
- Anyscale: Anyscale operates the Ray-based distributed compute platform for AI workloads, providing orchestration, scheduling, and multi-node training similar to Backend.AI's Sokovan orchestrator. Both target enterprise AI infrastructure at hyperscale.
- Domino Data Lab: Domino offers an enterprise AI platform with MLOps, model lifecycle management, and multi-cloud/hybrid GPU orchestration—closely comparable to Backend.AI's enterprise AI platform capabilities targeting regulated industries.
- Modal Labs: Modal provides serverless AI infrastructure for running models on GPUs, with similar developer-facing deployment and orchestration abstractions to Backend.AI:GO/DOL and Backend.AI Enterprise.
- Determined AI: Determined AI (acquired by HPE) provides an open-source AI training platform with distributed hyperparameter tuning and cluster management—directly comparable to Backend.AI's distributed training and fault-tolerant scheduling capabilities.
- Vast.ai: Vast.ai operates a distributed GPU compute marketplace and orchestration platform for AI training and inference—comparable to Backend.AI's heterogeneous GPU/NPU scheduling across distributed resources.
Broad incumbents
- CoreWeave: CoreWeave is a large GPU cloud provider offering both raw compute and orchestration layers (including acquired Valohai/run.ai assets). A broader incumbent than Lablup but operating in the same AI infrastructure category with similar enterprise customer base.
- Lambda Labs: Lambda provides GPU cloud infrastructure and GPU workstations for AI training and inference. A broader incumbent offering hardware-plus-software stack similar to Lablup's full-stack AI PC to hyperscale positioning.
- Paperspace: Paperspace (acquired by DigitalOcean) offers GPU cloud infrastructure with notebook and orchestration tooling. A broader incumbent in the same AI compute infrastructure category as Lablup.
Emerging players
- Weights & Biases: Weights & Biases provides MLOps tooling including experiment tracking, model management, and increasingly infrastructure orchestration. Adjacent but partial overlap with Backend.AI's enterprise AI platform capabilities.
Market position
Strengths5 records
Weaknesses5 records
Competitive moat5 records
Key risks5 records
Key highlights7 records
Customer concentration
Lablup social profiles
Digital presenceLablup compliance and trust
Trust signalCompliance5 records
Lablup financial estimates
Financial estimateRevenue estimate
Valuation estimate
Lablup leadership team
Management profileNumber of profiles
Profiles9 records
Lablup subsidiaries and ownership
Company hierarchySubsidiaries1 record
Lablup funding detail
Funding detailFunding overview
Funding rounds3 records
Investors7 records
Funding detail is available on the Subscription and Enterprise plan.Contact sales →
Lablup M&A and investment
M&A and investmentM&A
Investments
M&A and investment is available on the Subscription and Enterprise plan.Contact sales →
Frequently asked questions about Lablup
What does Lablup do?
Lablup develops and sells Backend.AI, a software-defined AI infrastructure operating platform that orchestrates heterogeneous GPU and NPU clusters at hyperscale with secure multi-tenancy and multi-node workload management. The company extends this enterprise platform with Backend.AI:GO, a free desktop application that runs small language models offline on personal PCs, and Backend.AI:DOL, a browser-based cloud solution for accessing open AI models, providing full-stack coverage from individual workstations to hyperscale data centers.
Is Lablup a public or private company?
Lablup is a private company. It is classified as venture growth investor backed and is currently operating.
When was Lablup founded?
Lablup was founded in 2015. It employs 11 to 50 people.
Where is Lablup based?
Lablup is headquartered in Seoul, South Korea, in the Asia region.
How does Lablup make money?
Four revenue lines are on record. Backend.AI Cloud Subscription is the primary driver. The others are on-premises Software Licenses, professional Services and GPU as a Service (GPUaaS).
Who are Lablup's main competitors?
Direct peers on record are Run:ai, Anyscale, Domino Data Lab, Modal Labs, Determined AI and Vast.ai. Broad incumbents are CoreWeave, Lambda Labs and Paperspace. Weights & Biases is listed as an emerging player.
Does Lablup have an API?
No public API is recorded for Lablup.
What industry is Lablup in?
Lablup's product category is AI Infrastructure Platform. Its primary akta.pro industry code is HDAAAAAG, AI Compute Virtualization & Scheduling (GPU virtualization, cluster schedulers), with a secondary code of HDAAACAB, Model Hosting, Serving & Inference Platforms. Its NAICS code is 54151 and its SIC code is 7372.