From store outlets to methane leaks, every scrap of power is now fair game for AI compute
AI factories are outgrowing the traditional data center as models get bigger and distributed inference moves from concept to deployment. The infrastructure behind those models now has to span racks, sites and even scattered pockets of power. Dell Technologies Inc. is betting that the answer lies in building complete systems rather than components, from rack-scale […] The post From store outlets…
AI factories are expanding beyond traditional data centers as bigger models move towards distributed inference. Dell Technologies believes the answer lies in building complete systems, not just components. Senior Vice President and Chief Technology Officer for AI, Compute and Networking, Ihab Tarazi, highlighted the massive scale of modern AI clusters, which can have tens of thousands to hundreds of thousands of GPUs on a single fabric.
Tarazi discussed Dell's new 100% liquid-cooled networking system, which includes switches, optics, and cabling pretested in their factory and delivered as a single rack. This reduces assembly time and bug-fixing, allowing deployment in hours rather than weeks. Dell has also introduced liquid-cooled storage for AI racks, with unified cooling and power specifications for compute, networking, and storage.
Distributed inference software is the next step, providing a single control plane for various GPUs, CPUs, and small sites with spare power. This approach turns previously unused power sources, like methane, into AI capacity, making efficient use of available energy.
Written by urgent.news from SiliconANGLE's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.