Retrofitting for Density in Data Centers

What started as a production challenge turned into something bigger. We just launched the first mission-critical 1200A Remote Power Panel (RPP) on the market. It started on the floor, building RPPs every day. Across the industry, panels are typically tied to specific breakers, which can limit how they’re designed based on availability, preference, or site standards. Instead of working around it, our team solved it. They built a universal panel that allows multiple breaker brands in the same system. More flexibility to design around customer needs, fewer constraints, faster builds and the ability to deliver more power in the same footprint. That matters right now, because AI is changing how data centers are built. As compute density increases, every part of the infrastructure has to keep up, including how power is delivered inside the building. That’s where our work lives. Proud of the team behind this one. 👉 Learn more about the HD-RPP: https://lnkd.in/eGRBB4aW #Accelevation #Manufacturing #Engineering #PowerInfrastructure

In next-generation liquid cooling architectures, connectors are no longer just fluid pathways. They help define pressure dynamics, flow uniformity, installation precision, contamination prevention, and service life across the cooling loop. Albert Pinto shares engineering insight into why connector selection matters as AI data centers move toward higher-density liquid cooling systems. Read the article: https://lnkd.in/eZM3CTya #DataCenterCooling #LiquidCooling #AICooling #ThermalManagement #Engineering

At the conference, Fourier and Intel jointly presented a fully integrated 20-foot modular data center container designed to combine cooling, compute, and

AI is pushing data centre power demands into new territory and it’s exposing the limits of traditional AC setups. In a new article, Zafer Cankurtaran (Product Manager Direct Current Technology at Rittal GmbH & Co. KG), explains why more operators are seriously looking at VDC and sidecar architectures to improve efficiency, reduce copper use and free up space. If you’re planning for higher loads and more efficient infrastructure, this shift is really relevant. 👉 Read more: https://smpl.is/ajmq8 #DataCentres #ITInfrastructure #AI #PowerDistribution #Rittal

High-density data centers running Artificial Intelligence workloads face power and thermal challenges that differ significantly from traditional computing environments. Large-scale GPU workloads generate rapid, concentrated power surges and localized thermal spikes that place significant stress on power and cooling infrastructure. Addressing these conditions requires an integrated approach — combining rack-level power management, advanced Uninterruptible Power Supply architectures, and liquid cooling systems designed to stabilize both electrical and thermal behavior under variable load conditions. The Vertiv white paper 'AI Workload Management: Designing Efficient Cooling and Power Architectures' examines how integrated power and cooling strategies are designed to help manage workload volatility in modern data centers. Read the white paper: http://ms.spr.ly/6048vuOTu. #Vertiv #AIInfrastructure #LiquidCooling #DataCenter #ThermalManagement

High-density data centers running Artificial Intelligence workloads face power and thermal challenges that differ significantly from traditional computing environments. Large-scale GPU workloads generate rapid, concentrated power surges and localized thermal spikes that place significant stress on power and cooling infrastructure. Addressing these conditions requires an integrated approach — combining rack-level power management, advanced Uninterruptible Power Supply architectures, and liquid cooling systems designed to stabilize both electrical and thermal behavior under variable load conditions. The Vertiv white paper 'AI Workload Management: Designing Efficient Cooling and Power Architectures' examines how integrated power and cooling strategies are designed to help manage workload volatility in modern data centers. Read the white paper: http://ms.spr.ly/6041vuOVc. #Vertiv #AIInfrastructure #LiquidCooling #DataCenter #ThermalManagement

AI workloads are not simply "more compute." They fundamentally change the physical infrastructure requirements of a data centre. Higher rack densities, heavier current loads, more aggressive cooling demands, and power distribution systems that have to handle several times what a traditional rack pulls. That is where the physical layer becomes critical. At HS Global, we fabricate the server racks, copper busbars, LV switchboards, and battery systems that data centres actually need, sized and built for the loads they will realistically run, not the loads they once did. #AIInfrastructure #DataCentre #HSGlobal #DigitalTransformation

AI and HPC are changing the rules of data center cooling. As rack densities move from traditional enterprise loads to 40 kW, 50 kW, 70 kW and beyond, engineering teams can no longer treat cooling as a late-stage equipment decision. Air cooling will continue to serve many workloads. But for high-density AI and HPC environments, the real question is: when does air cooling stop being efficient, scalable, or future-ready? In our latest blog, Shaikh Abdulla Anwar explores the shift from air cooling to liquid cooling, the engineering triggers behind that decision, and why data center owners, colocation providers, hyperscalers, and design teams in the US and Europe need to evaluate cooling strategy much earlier in the project lifecycle. The blog also looks at hybrid cooling models, rack-level density planning, CFD-linked analysis, coordinated MEP layouts, CDU placement, service access, water quality, and liquid-ready facility design. Read the full blog here https://lnkd.in/gMFhQ6Wz to understand what high-density data centers need before the rack arrives. #DataCenterDesign #DataCenterCooling #LiquidCooling #AICooling #HPC #HighDensityDataCenters #MEPEngineering

Overcoming the Infrastructure Bottleneck in High-Density Data Centres 🇬🇧 The rapid acceleration of high-density AI workloads is forcing a critical conversation around European data centre infrastructure. While traditional legacy air cooling is reaching its physical limits, transitioning to liquid thermal management has historically introduced substantial facility friction. Full immersion cooling offers pure thermal physics, but the operational realities for many UK and European operators present clear barriers: 🔹 Spatial & Structural Constraints: Complex tank installations demand extensive physical footprints and heavy floor-loading modifications — particularly challenging for urban and retrofitted facilities. 🔹 Operational Friction: Routine maintenance requires specialised lifting infrastructure to remove servers from dielectric fluids, changing standard data centre workflows. 🔹 Lack of Standardisation: Fluid types and component certifications remain highly fragmented across the industry supply chain. There is a distinct, low-friction middle path. Our latest deep dive explores how precision spray cooling is bridging the gap. By utilising 3D-printed nozzles to deliver targeted coolant directly onto CPU and GPU surfaces, our LiquidRack architecture cuts dielectric fluid volume by up to 80% compared to immersion — drastically reducing cost, structural weight, and complexity. As Tony Fischels notes, this server-level innovation is custom-built for legacy data centres, enterprise facilities, and edge deployments where footprint and additional infrastructure space are at an absolute premium. 🔗 Read the full technical analysis on the blog: https://lnkd.in/eQFty95r #DataCentres #LiquidCooling #ThermalManagement #Airsys

5. Distributed / Rack-Level BESS (The Next Evolution) This is where traditional data center design starts to break. ⸻ For decades, power has been centralized. • Utility → UPS → PDU → Rack Everything flows from the top down. ⸻ That model is now under pressure. ⸻ AI workloads are pushing: • 50kW • 100kW • 150kW+ per rack ⸻ At that level… Every conversion loss matters. Every inefficiency compounds. Every delay in response becomes risk. ⸻ So the architecture is shifting. ⸻ Instead of centralized batteries… Energy storage is moving closer to the load. ⸻ Rack-level or distributed BESS changes the game: • Faster response to load changes • Reduced conversion losses • Better alignment with DC architectures (48V / 800V) • Improved efficiency at extreme densities ⸻ This isn’t just a battery upgrade. It’s a topology shift. ⸻ From: Centralized infrastructure To: Distributed, load-aligned energy systems ⸻ And when you combine this with: • Direct-to-chip cooling • High-density AI clusters • DC power distribution ⸻ You start to see the future of the white space. ⸻ But let’s be clear: This is still early. ⸻ Standards are evolving. Designs are not fully mature. Execution risk is high. ⸻ But the direction is obvious. ⸻ Because the old model… Was never designed for what’s coming. ⸻ And the facilities that adapt first— Will define the next generation of infrastructure. ⸻ #DataCenters #BESS #AIInfrastructure #PowerArchitecture #MissionCritical #Innovation #EnergySystems #Hyperscale