Lablup Successfully Hosts "lab | up > /conf/6," Showcasing the Future of Next-Generation AI Infrastructure
- Technical sessions under the theme "Make AI Composable²," covering AI infrastructure, inference, agents, and security
- Introduces Continuum Router and Continuum Hub, outlining its approach to building a "Token Factory"
- Shares AI infrastructure technologies and operational directions with NVIDIA, kt cloud, FuriosaAI, Upstage, NotaAI, and others
Lablup (CEO Jeongkyu Shin), an AI infrastructure software company, announced that it successfully concluded its sixth annual tech conference, "lab | up > /conf/6," on the 29th at COEX in Samseong-dong, Seoul.
This year's theme was "Make AI Composable²." Companies and institutions from Korea and abroad, including NVIDIA, VAST Data, kt cloud, FuriosaAI, Samsung Welstory, RLWRLD, AIM Intelligence, and KAIST, took the stage to address real-world technical and operational challenges facing the AI industry, spanning inference and serving, AI agents, platform engineering, security, data pipelines, and physical AI.
In his opening keynote, "On End-to-End in the Age of the AI Industry," CEO Jeongkyu Shin noted that AI is moving beyond its scientific and engineering eras into an industrial one, and introduced the technologies and businesses Lablup is building to fill the gaps in this new era.
The panel talk that followed, "Inference Optimization for Solving AI Bottlenecks," was moderated by Jeongseok Noh, CEO of B Factory. Panelists Joongi Kim (CTO, Lablup), Taeho Kim (CTO, NotaAI), Jinwon Lee (CTO, HyperAccel), and Hwalsuk Lee (CTO, Upstage) discussed how they are tackling inference bottlenecks in their respective fields, from AI semiconductors and model development to model optimization and infrastructure orchestration.
At the event, Lablup introduced Continuum Router, which connects multiple AI models through a single gateway, and Continuum Hub, which centrally manages token usage, policies, and billing. Lablup presented technologies for flexibly routing requests across diverse AI models and managing token usage in one place, along with its concept of the "Token Factory." Attendees also had the chance to try out key features of Continuum Router and Continuum Hub through live demos.
"As the AI industry grows rapidly, it is becoming increasingly important not only to advance individual technologies, but also for diverse companies and technologies to connect with one another and share real-world operational experience," said Jeongkyu Shin, CEO of Lablup. "This conference gave companies and experts from Korea and around the world the opportunity to discuss the technical challenges and direction of the AI industry together."
Meanwhile, Lablup recently concluded a successful global webinar and continues to expand its engagement with overseas customers and partners. The company plans to keep promoting its technologies and products through a range of global touchpoints and to broaden its foundation for international business.
▲ Lablup CEO Jeongkyu Shin delivers the keynote at lab | up > /conf/6.
▲ Four CTOs, including Lablup CTO Joongi Kim, hold a panel discussion on inference optimization for solving AI bottlenecks.