Discover how Denvr’s vertically integrated AI infrastructure and cloud platform are enabling real-world innovation. Through customer use cases spanning public sector, industry, and research, we share how sovereign, scalable AI infrastructure is translating into meaningful outcomes.

Explore practical lessons learned from designing, deploying, and operating AI infrastructure at scale. We share how Denvr approaches efficiency, density, deployment speed, and adaptability to meet rapidly evolving technology demands, and why cross-industry collaboration is essential to innovation.

Explore how collaboration between Denvr, Unicom, and industry partners is advancing immersion-native AI hardware. We share the evolution from H200 through B300 server designs, highlighting innovations in thermal performance, compute density, and efficiency that are helping enable next-generation AI infrastructure.

In the first installment of this three-part series, Amy Short (Denvr Dataworks) and Austin Hypes (UNICOM Engineering) break down their joint effort to redesign AI servers for immersion cooling. The headline result: cutting a 6U air-cooled chassis down to 3U by reengineering the power system (AC to 48V DC), PCB layout, and thermal architecture for liquid immersion, more than tripling rack density from roughly 4 GPU servers to 14.

This episode covers UNICOM Engineering and Denvr Dataworks' redesign of the Dell PowerEdge XE-9680 (H200 HGX node) for single-phase dielectric immersion cooling, shrinking it from 6U to 3U and hitting 14 servers per rack versus 2-4 air-cooled. Topics include immersion-optimized heat sinks, an OCP-aligned power architecture, and the NVQual validation behind a production warranty, delivering 10x density, ~40% less cooling overhead, ~800W recovered per server, and zero water use, running live in Denvr's data centers since April 2024.

The final episode tackles two core engineering challenges for production-ready immersion-cooled AI servers: signal integrity (NVQual testing confirms the immersion-native H200 HGX node meets SXM specs despite being submerged in dielectric fluid) and power architecture (the shift from OCP ORV 2.2's 12V to a 48V-54V DC bus to cut I²R losses for high-density racks). It closes with the path to mainstream adoption, backed by 3.5+ years of operational data and ongoing OCP standards work.

At OCP Canada Tech Day, Denvr presents real-world data, design decisions, and the OCP collaboration behind its immersion-cooled AI NeoCloud, including PUE figures that redefine what "efficient" means.

In this Upper Bound presentation we're sharing how our vertically integrated AI solutions are accelerating real customer outcomes, including a recent deployment with the Amii.

At OCP EMEA Summit, this panel discussion was focused on scaling AI clusters, highlighting Neocloud innovations, key challenges, and collaboration pathways.

This OCP session highlights a successful collaboration between UNICOM Engineering and Denvr Dataworks to design- co-validate and deploy immersion‑native H100 & H200 SXM servers for a sovereign AI NeoCloud in a modular data center (MDC).

A three-part technical validation story — from thermal failure, to engineered solution, to production-proven performance — demonstrating what optimized AI hardware looks like under real-world stress.

Denvr & UNICOM Engineering built a production-scale immersion-cooled AI platform: 14 servers per rack, 10x density gains, zero water. Real workloads. Real results.

Denvr's keynote presentation at them 2025 OCP Global Summit in San Jose, CA.

This OCP Unicom presentation covers the real-world failure rates at the component level observed for over 3000 servers deployed in a single datacenter.

This OCP session focuses on operational insights from a multi-year scaled deployment of single-phase immersion cooling at Denvr, using fluid developed and monitored by The Lubrizol Corporation.

Liquid Cooling: Enhancing AI and Compute Density for Modern Data Centers

This OCP session explores how Denvr leverages key OCP contributions in power, cooling, networking, and systems integration to accelerate deployment, improve performance, and streamline complexity.