backend.ai

Greetings from Lablup. We're excited to be a Silver Sponsor at NVIDIA GTC 2025, which runs March. 17-21 in, San Jose, California.

Lablup (#547) is unveiling two innovative products at GTC 2025 based on our AI Infrastructure Operating Platform, Backend.AI. Backend.AI is an accelerated workload hosting platform that maximizes the performance and improves the operability of GPU infrastructure and can run in a variety of environments, including the cloud or on-premises/air-gapped environments. Backend.AI is NVIDIA DGX-Ready software certified, ensuring high compatibility and behavioral stability with the NVIDIA DGX platform.

#1. Backend.AI Continuum

Backend.AI Continuum is a solution that enables organizations using cloud API-based services to continue critical operations even in the case of network failures or service outages. The technology intelligently bridges the gap between the cloud and on-premises environments during normal operation and automatically switches to local resources in the event of a cloud outage, keeping API calls and processing uninterrupted. Backend.AI Continuum is an innovative solution that allows enterprises to have the flexibility of the cloud and the reliability of on-premises at the same time.

Given the recent increase in global organizations' reliance on cloud services, Backend.AI Continuum is expected to be a key part of our North American expansion. Lablup is currently working on pilot projects with several potential customers in the US leveraging Backend.AI Continuum.

#2. Backend.AI for Personal Supercomputer

Backend.AI for Personal Supercomputer is a lightweight version of Backend.AI that runs on personal supercomputer products such as NVIDIA DGX Spark or edge computing devices based on Jetson Orin™ Nano series™ modules. It optimizes the core functionality of the Backend.AI for use on consumer desktops, delivering high-density generative AI performance in edge AI computing environments. We believe that Backend.AI for Personal Supercomputer will make it easier for individuals interested in AI technology, or even small and medium-sized enterprises, to adopt and manage AI.

Sessions

Learn about the session announcement featuring Lablup's CEO Jeongkyu Shin and CTO Joongi Kim.

Talks & Panels | Universal NIM Acceleration With GPU-Sharing Containers (Presented by Lablup) [S74194]

https://www.nvidia.com/gtc/session-catalog/?regcode=no-ncid&ncid=no-ncid&tab.catalogallsessionstab=16566177511100015Kus&search=lablup#/session/1734356801656001mJ1r

Dive into the universal world of NIMs. NVIDIA NIMs envisions the path to achieve Multi-modal, multi-agent AI systems using optimized container templates. Lablup will explain how our GPU-native container engine further accelerates NIMs to deliver such advanced multi-agent AI systems at low cost and high performance. It exploits a novel fractional GPU-sharing technology to accommodate multiple different models having diverse performance bottlenecks in a single GPU, automate resource allocation and model combinations with memory size estimation techniques, and auto-scale the NIM containers by incorporating inference runtime metrics. All these features are implemented on both air-gapped on-premises clusters and cloud-native setups. On top of it, we've also built a streamlined UI to import, fine-tune, and serve open models in just one click, effectively hiding all those technical details from the end user.

Theater Talk | Resilient Edge-Cloud Hybrid AI Infrastructure: Orchestrating Multi-Modal Agents in Resource-Constrained Environments (Presented by Lablup) [EXS74187]

https://www.nvidia.com/gtc/session-catalog/?regcode=no-ncid&ncid=no-ncid&tab.catalogallsessionstab=16566177511100015Kus&search=lablup#/session/1734342221616001LkG3

Explore new solutions for building resilient AI infrastructure that seamlessly integrates edge and cloud computing through intelligent orchestration. Lablup will demonstrate how an eight-node NVIDIA Jetson Nano cluster serves multiple LLMs (including Gemma 2 2B and Llama 3.2 3B) and supports multi-modal AI agents while maintaining cloud GPU integration. Learn how to orchestrate distributed AI systems that process text, images, and sensor data, enabling robust edge computing with cloud failover capabilities.

See you soon!

Our team is looking forward to connect with you at GTC 2025. If you're attending GTC 2025, be sure to stop by our booth(#547) to see what's new in the AI market. In addition to live demos of the Backend.AI Continuum and Backend.AI for Personal Supercomputers, we will be presenting examples of integrations within the NVIDIA ecosystem, as well as private meetings for customers, individuals, and partners interested in learning more about Lablup.

Earlier this year, Lablup opened a regional office in Silicon Valley and is taking on the challenge of global expansion. We hope GTC 2025 will be a great opportunity to connect with our team and learn more about Lablup and Backend.AI.

About GTC 2025

NVIDIA GTC is the largest technical conference in AI field. With more than 1000 sessions, 300+ exhibits, hands-on technical training, networking events, and also a keynote from NVIDIA CEO Jensen Huang, it's the best opportunity to join thousands of developers, innovators, and business leaders to explore how AI and accelerated computing are helping solve humanity's complex problems.

Go to GTC 2025 Official Website