TRUSS DATA CENTERS
Jon Eaves
CRO TRUSS
Data Centers
Each customer receives an individually engineered cooling system and different chip architectures can require separate loops.
“ We engineer new loops for that chip, and a separate loop for the normal engineering,” he notes.“ It’ s funny that I now define NVIDIA as normal engineering.”
Customers’ engineers review the single-line diagram and sign off the build, reflecting the importance of engineering at these densities.
“ If you lose the cooling for a matter of seconds, it fries,” says Jon.“ That’ s more delicate than losing a little cooling in one room.”
Jon’ s own background feeds directly into this approach. Before TRUSS, he founded and built Edge Centres and worked across telecommunications, network wholesale and hands-on data centre engineering in Australia and the US.
“ I’ ve built data centres with my own hands, and I’ ve run the networks that connect them,” he elaborates.“ That combination is the company’ s edge. When a customer’ s engineers ask about our chilled water loop or our distribution architecture, they’ re talking to the person who designed it. There’ s no translation layer. In this market, that’ s speed, and speed is everything.”
Not one‐size‐fits‐all That engineering-led philosophy also shapes TRUSS’ approach to resilience.
Facilities are engineered to Tier IIIsimilar standards where the workload justifies it, rather than treating a certification level as a universal requirement. AI training clusters have different availability economics from banking infrastructure, Jon argues.
“ An AI training cluster has different availability economics than a bank’ s core ledger,” Jon says.
14 trussdc. com