With so much focus on huge DCs from SoftBank, Amazon, Meta, and Google, it’s easy to forget the innovation of smaller footprints that offer 8- 20 MW chunks to enterprises — where it’s sometimes the case that commodity IT infrastructure must coexist with the likes of Nvidia Superpods or Cerebras clusters. 

For enterprises that demand speed, agility, and flexibility, there is a niche demand for ultra-high-density, liquid-cooled DCs purpose-built for AI and HPC workloads. That was the focus of a quick discussion I had with Colovore’s Tomek Mackowiak before he went on stage in Virginia for a DCD >Connect panel discussion “Building the AI Ecosystem from Chip to Grid.” As a testbed for visionary AI hardware companies like Nvidia, Cerebras, Groq, SambaNova, and d-Matrix, Colovore has gained a unique perspective on the enterprise need for extreme density, high-capacity power, and cooling, as well as compliance. “We have to be mindful of ISO, SOC, HITRUST, and we want to come off the utility, as close to the workloads and the data gravity as possible,” said Mackowiak, explaining the contrast between a smaller footprint and that of bigger one pursuing commodity inference or support of heterogenous deployments. To learn more, see the full interview here.