# DeepMind Trained a Frontier Model Across Four Regions on Regular Internet Bandwidth — and Then Broke It On Purpose

- slug: deepmind-trained-a-frontier-model-across-four-regions-on-regular-internet-bandwidth-and-then-bro
- date: 2026-04-23
- category: Artificial Intelligence

DeepMind deliberately crashed its own training hardware to prove a point: its new DiLoCo system kept learning uninterrupted, delivered equivalent AI performance, and ran 20x faster than conventional methods on ordinary internet bandwidth between data centers.

---