Rableup, an AI infrastructure platform company, said on July 14 it presented a case study on operating large-scale AI infrastructure at the AI for Good Global Summit 2026 hosted by the UN's International Telecommunication Union (ITU).

The AI for Good Global Summit is a UN AI event co-hosted by the ITU with the Swiss government and more than 50 UN agencies. This year's seventh edition was held from July 7 to 10 at Palexpo in Geneva, Switzerland, with more than 12,000 participants from 170 countries.

On the first day of the event, July 7, Rableup CTO Jun-gi Kim (김준기) delivered a 15-minute session at the workshop titled "The Future of AI Native Communication Networks" under ITU-T SG13 (future networks and emerging network technologies). SG13 is a study group that addresses the integration of telecommunications networks and AI technology, and the workshop discussed both the "AI for Network" perspective of advancing network operations with AI and the "Network for AI" perspective that networks must evolve as infrastructure that can accommodate large-scale AI workloads.

Kim focused his presentation on the message that as AI infrastructure scales, integrated, cross-layer optimisation across hardware, software and networks, rather than single-GPU performance, is emerging as a core challenge. He shared infrastructure operations experience from the Ministry of Science and ICT-led "independent AI foundation model" project as an empirical case.

Kim said, "In large-scale AI training environments, real performance is determined not by the performance of a single GPU, but by how the entire stack is designed and operated." He added, "Through this presentation, we were able to share infrastructure engineering experience validated in real-world operations in the context of international standardisation discussions."