From Concept to Capability: Innovating The Future of OCP Reference Designs

Author: Greg Ratcliffe, Vertiv Collection: 2025 OCP Global Summit - Expo Hall Sessions

Summary

Vertiv's CTO office presents its innovation process for developing OCP-compatible reference designs for power and cooling infrastructure. The presentation focuses on methodology rather than products: how a technology scouting program, university partnerships, technology readiness level (TRL) advancement, and standards alignment accelerate the development of infrastructure solutions for AI data centers. The core message is that pushing the limits of physics (not just technology) at current AI power densities requires ecosystem collaboration and disciplined innovation processes.

Vertiv maintains approximately 100 "technology scouts" across the business globally -- practicing engineers who, in addition to solving current problems, scan for emerging technologies and bring forward emerging technology needs to the CTO office. These needs are advanced independently of active engineering projects, reducing time to market by developing skills and proving technologies before they enter formal product development. The TRL framework (adapted from NASA) guides readiness assessment from benchtop prototype through scaled production.

University partnerships are a key accelerator: Vertiv participates in ES2 (an industry-university collaborative research organization) with monthly check-ins on research projects. NC State and Virginia Tech are specifically cited as partners on the recently announced plus/minus 400V and 800V DC reference designs. Digital twin development running in parallel with physical prototyping is highlighted as increasingly important for testing systems beyond real-world limits. Looking forward, Vertiv sees infrastructure needing to become more compact and potentially relocatable further from compute, as AI rack values approach $37 million, making infrastructure's proximity to compute an increasingly expensive proposition.