Alibaba Cloud is partnering with Olympic Broadcasting Services (OBS) and the International Olympic Committee (IOC) to deploy cloud and AI technologies for Milano Cortina 2026, building on prior deployments at Tokyo 2020, Beijing 2022, and Paris 2024.

IOC CIO Ilario Corna calls AI integration a defining moment for the Olympic Movement, unlocking practical benefits for fans and helping preserve historic moments.

The AMD System will enable near-instant, AI-assisted search for OBS teams by automatically identifying athletes and moments and producing searchable descriptions.

A Real-Time 360° Replay system, AI-enhanced, will deliver immersive replays by isolating athletes from backgrounds and generating 3D reconstructions within 15–20 seconds across 17 sports.

The announcements come from Milan, Italy, on February 4, 2026.

Milano Cortina 2026 will distribute over 5,000 ready-to-use digital Olympic assets via OBS Content+, with advanced discovery tools for global teams.

OBS Live Cloud will support 39 broadcasters, deliver 428 live feeds (including 26 in 4K) and 72 audio feeds, reduce satellite reliance, and enable the OBS Olympic Video Player to stream HD through Alibaba Cloud to smaller broadcasters.

IOC’s Flex asset platform will integrate Sports AI, Alibaba Cloud’s archive solution, to tag and search more than eight petabytes of Olympic content with automatic tagging and multimodal search.

IOC launches its first large-language-model-based system, Olympic AI Assistants powered by Alibaba’s Qwen, for multilingual fan engagement on olympics.com and AI audio guides at the Olympic Museum; an internal NOC AI assistant will aid document queries.

The collaboration aims to enhance global broadcasting, streamline operations, and create scalable, cloud-based, AI-enabled solutions to capture, manage, and preserve Olympic content.

OBS’s Automatic Media Description (AMD) System, powered by Alibaba’s Qwen LLM, will automatically identify athletes and moments, generate event descriptions, and tag assets for rapid retrieval via natural-language queries.