Web AI on next generation AI PCs
summary
WebNN is a W3C web API running in Chrome/Edge Developer Preview and ONNX Runtime Web that maps models to CPU/GPU/NPU and demos show it actually works: Lunar Lake delivers 120 TOPS and absorbs GPU XMX for 2–3.5× speedups; Stable Diffusion drops from ~3s to ~700ms on integrated GPU; Segment Anything falls 40ms→20ms; object detection rises 100→150+ FPS while NPU utilization falls 36%→21%; CPU runs at near‑native 93% and on‑device Whisper is ~3× real‑time.
AI-written summary · about
chapters (11)
- 0:00Web AI and AI PCs intro
- 1:22AI PCs rise and multi-engine strategy
- 2:52Lunar Lake AI PC breakthrough 120 TOPS
- 4:00WebNN for cross-platform AI inference
- 6:53Segment Anything vs Stable Diffusion demos
- 9:03Live object detection NPU GPU
- 10:49On-device speech-to-text Khan demo
- 12:59WebNN performance near-native
- 14:26Wasm SIMD optimizations
- 15:53Emscripten AVX intrinsics future
- 17:00Call to action WebNN resources
related talks
16mThe Web is AI Ready—maximize your AI web development with WebNN
39mThe AI PC and the Era of the Modern Web
45mPrivacy-First In-Browser Generative AI Web Apps
46mThe Ultimate Guide to Browser-Based AI: Privacy-First, Offline-Ready and Future-Proof
32mPrivacy-first in-browser Generative AI web apps
17mRunning Intelligence at the Edge: Web AI in Action
30mBeyond the Web of Today
41mEscaping the Cloud: High-Performance AI in your Browser