Keynotesby GlyphComputex
Keynotes at Computex · 17 keynotes · 2026, 2025, 2024, 2023 · mostly AI infrastructure and Robotics · 13 announcements
Year by year
The top highlight of each keynote20264
NVIDIA Keynote: Extreme Co-Design: Building the AI FactoryA 28-minute keynote by Jensen Huang.
NXP CEO keynoteA 1h 11m keynote video from the official COMPUTEX TAIPEI channel.
Marvell CEO keynoteA 1h 18m keynote video from the official COMPUTEX TAIPEI channel.
Qualcomm CEO keynoteA 1h 23m keynote video from the official COMPUTEX TAIPEI channel.
20254
NXP keynoteA 1h 22m keynote video from the official COMPUTEX TAIPEI channel.
MediaTek keynoteA 1h 40m keynote video from the official COMPUTEX TAIPEI channel.
NVIDIA CEO Jensen Huang KeynoteHuang told a crowd of more than 4,000 at Taipei Music Center that AI data centers are, in fact, AI factories.Jensen Huang9 announced1h 43m
Qualcomm keynoteA 1h 46m keynote video from the official COMPUTEX TAIPEI channel.
20246
Supermicro keynoteA 1h 31m keynote video from the official COMPUTEX TAIPEI channel.
NXP keynoteA 1h 8m keynote video from the official COMPUTEX TAIPEI channel.
Intel Enables AI EverywhereA 55-minute keynote video from the official Intel Newsroom channel.
Global Press Conference and AMD Opening KeynoteA 2h 10m keynote by Lisa Su.
Snapdragon Keynote: The PC Reborn with Snapdragon X EliteA 1h 18m keynote video from the official Snapdragon channel.
NVIDIA CEO Jensen Huang KeynoteNIM, NVIDIA Inference Microservices, package models to run inside "AI factories"Jensen Huang4 announced1h 49m
20233
Acer KeynoteA 54-minute keynote video from the official COMPUTEX TAIPEI channel.
NVIDIA KeynoteA 1h 56m keynote by Jensen Huang.
Qualcomm KeynoteA 1h 16m keynote video from the official COMPUTEX TAIPEI channel.
Announcements
NVIDIA BlackwellEngine powering the AI ecosystem, one architecture from cloud to edge
NVLink FusionNew architecture letting hyperscalers build semi-custom compute with NVLink
DGX SparkPersonal AI supercomputer, full production, available in a few weeks
DGX StationUp to 20 petaflops from a wall socket, runs 1 trillion parameter models
NVIDIA RTX PRO ServersEnterprise servers for agentic AI, now in volume production
NVIDIA AI Data PlatformReference design with RTX 6000 PRO Blackwell Server Edition GPUs
Newton physics enginePhysics training engine for robotics, with Google DeepMind and Disney
Isaac GR00T-Dreams / GR00T N1.5Blueprint and foundation model to speed humanoid robot development
NVIDIA ConstellationNew NVIDIA office in Taiwan for its growing Taiwan workforce
NVIDIA NIMInference microservices, Llama 3 NIM on Hugging Face
Blackwell in productionBlackwell production boards shown
Spectrum-X roadmapSpectrum-X800 Ultra, then X1600
Rubin platformNext platform after Blackwell
About and year over year
Announcements per year
20260
20259
20244
20230
GAPreviewComingAnnounced
Companies on stage: Nvidia, ASUS, CoreWeave, DeepMind, Delta Electronics, Disney Research, Foxconn Hon Hai Technology Group, GIGABYTE, Microsoft, MSI, Oracle, Pegatron, Quanta Computer, TSMC, Wistron, SpaceXAI, Intel, AMD, Qualcomm, Hugging Face, Meta