Datacenter Management Practices

"How to Evaluate a Building for Data Center Conversion" Earlier this week I shared how Chicago developers turned a $12 million office building into a $40 million data center in 15 months. Today, let's talk about what to look for. The Five Critical Factors: 1. Power Infrastructure This is the dealbreaker. Can you increase capacity to 30-50 megawatts? Existing transformers? Proximity to substations? The Chicago building had substantial electrical infrastructure from its trading floor days. Without power capacity, you don't have a deal. 2. Building Structure You need: Wide, column-free floors High ceilings for cooling Floor load capacity for server weight Cavernous layouts The Cboe building was designed for trading floors—which converts perfectly to data centers. 3. Existing Connectivity "This building is very heavily wired from its time as a trading platform," said buyer Daniel English. Look for heavy wiring, fiber proximity, and urban locations near connectivity hubs. 4. Cooling Potential CRE Daily reports liquid cooling is becoming standard as power densities jump from 120 kW per rack today to 600 kW by 2027. Can the building support liquid cooling systems and upgraded HVAC? 5. Urban Location Advantage English explained why urban conversions command premiums: "Just like Amazon last-mile delivery, data centers take less time to deliver when they're close." Low-latency applications—trading, streaming, gaming—pay premiums for urban proximity. The Best Candidates: Former trading floors, financial services buildings, telecom facilities, heavy industrial with power infrastructure. My Take: The Chicago flip proves it: The biggest returns aren't in greenfield development. They're in buying assets where someone else already solved the hard problems and the market hasn't caught up. What building in your market has these five factors? Because while everyone else sees obsolete real estate, you might be looking at a 233% return in 15 months. What are you seeing that others are missing? Sources: "Flip of former Cboe Global Markets headquarters in Chicago shows soaring data storage values" by Ryan Ori, CoStar News, October 23, 2025; "Data Centers Driving Growth In AI And Real Estate" CRE Daily, PrincipalAM research

𝙋𝙤𝙬𝙚𝙧 𝙝𝙖𝙨 𝙧𝙚𝙥𝙡𝙖𝙘𝙚𝙙 𝙛𝙞𝙗𝙚𝙧 𝙖𝙣𝙙 𝙢𝙚𝙩𝙧𝙤 𝙥𝙧𝙤𝙭𝙞𝙢𝙞𝙩𝙮 𝙖𝙨 𝙩𝙝𝙚 #𝟭 𝙨𝙞𝙩𝙚 𝙨𝙚𝙡𝙚𝙘𝙩𝙞𝙤𝙣 𝙙𝙧𝙞𝙫𝙚𝙧. Most teams have not adjusted. For two decades, site selection started with fiber routes, latency, and incentives. Power was a check box. In 2026, power is the constraint. It dictates timeline, cost, and viability. Everything else is secondary. JLL is projecting average global build cost at $11.3M per MW for 2026, up from $7.7M in 2020. That delta is not general inflation. It is interconnection scarcity showing up in capex. What is actually happening in the field right now: • Sites with existing substations near retiring coal or industrial loads are commanding premiums. You are buying time, not just land. • Behind-the-meter natural gas is no longer a temporary bridge. It is the primary path when grid timelines exceed 3 to 5 years. • Operators willing to run hybrid power strategies are beating “wait-for-grid” models on speed to market. That gap is widening. If your deck still leads with fiber maps, you are optimizing the wrong variable. Lead with megawatts, queue position, and time-to-energize. Then design network, cooling, and tax strategy around that reality. Board-level translation: Power access is now the gating factor for revenue realization. Miss it and nothing else matters. What is the first question your team asks on a new site today: power or latency? #DataCenters #AIInfrastructure #SiteSelection #RoyaleStakes

Your Data Center might need this... . . What if your rack capacity isn’t limited by power… but by airflow design?... In many traditional open data halls, hot and cold air mix within the room. This thermal mixing reduces cooling efficiency and often limits practical rack loads to around 5–10 kW per rack, even if the electrical infrastructure can support higher densities. (Source: BICSI 002; ASHRAE Thermal Guidelines) With Hot or Cold Aisle Containment, airflow becomes controlled and predictable. Cold air is delivered directly to server inlets while hot exhaust air is isolated and returned to the cooling system. This simple change allows many facilities to safely support 15–25 kW per rack using air cooling, depending on airflow design and cooling capacity. (Source: ANSI/TIA-942; BICSI 002) As compute densities continue to grow with AI, HPC, and high-performance workloads, airflow management is becoming just as critical as power infrastructure. Sometimes the biggest upgrade in a data center isn’t new equipment… it’s simply controlling the air. 🗨️ For data center professionals: In your experience, which delivers better performance in your environment — Hot Aisle Containment or Cold Aisle Containment?

Most people analyze data centers like real estate. That is the mistake. After reviewing projects across the U.S., Europe, the Middle East, and Asia, one pattern keeps repeating. About 20 percent of variables explain 80 percent of outcomes. And those variables are not what most people focus on. The five that matter most are: 1. Power first. A data center is a machine that converts electricity into compute. No reliable power, no economic project. 2. Land control next. Not acreage, but control. Zoning certainty, expansion rights, water access, and fiber proximity determine whether a site becomes a long-term campus or a stranded asset. 3. Then capital. These are front-loaded projects. Impatient equity or unrealistic debt timelines can sink even strong projects. 4. Demand matters, but differently. Forecasts do not finance infrastructure. Contracts do. AI drives demand, but also volatility between training clusters and distributed inference. 5. Finally, sequencing. Discipline here separates winners from losers. The correct order is clear. Reverse it, and risk compounds. Underwrite power pathways and capital alignment before square footage. That is where the real returns come from. The winners are the ones who see the signals others ignore. Infrastructure strategy is not about buildings, It is about timing, control, and certainty. Read the article below. #datacenters

🔹 Thermal Management for 1–2MW Liquid-Cooled Data Centers — System Architecture, Value & Deployment Guidance 🔹 To meet the demands of 1–2MW+ power densities, a structured thermal management ecosystem becomes critical — one that optimizes heat removal, minimizes energy consumption, and scales without excessive footprint or operating cost. Our professional system architecture divides this solution into four engineered components: 📌 1. CDU (Cooling Distribution Unit) Industrial-grade skidded CDU with precision flow control, redundant pumps, and high-efficiency brazed plate heat exchangers. Designed for continuous operation in high-load environments. 📌 2. Dry Cooler Module Large V-coil dry cooling banks paired with EC fans enable low-water or waterless heat rejection when ambient conditions permit. This dramatically reduces water usage and lowers operating cost while maintaining design ΔT. 📌 3. Modular Containerized 1–2MW Deployments Factory-assembled, pre-tested thermal modules integrate CDU, dry coolers, and piping into service-ready enclosures. Ideal for rapid deployment and edge facility expansion. 📌 4. High-Density Liquid Cooled Rack Aisles Configured with robust manifold piping, quick-connect rack headers, and service access aisles — delivering uniform coolant distribution at high flow rates while preserving hot-aisle/cold-aisle containment. 🧠 Why This Matters Now ➡ AI and GPU clusters often exceed 40–80 kW per rack — densities air systems cannot support efficiently. ➡ Liquid cooling reduces fan energy by 40–60% and lowers total facility PUE when paired with intelligent CDU controls. ➡ Hybrid solutions (air + liquid) provide a pragmatic migration path for facilities transitioning from legacy infrastructure. 📊 What We’re Seeing in the Market ✔ Hyperscale and AI operators are rapidly adopting liquid cooling for primary compute zones. ✔ Colocation and enterprise facilities are planning liquid cooling zones within mixed-cooling halls. ✔ 1–2MW thermal modules are becoming the standard design unit for new builds and expansions in APAC, EMEA, and North America. 🛠 Deployment Guidance for Technical Teams 🔹 Start with power density targets, not cooling technology. Determine expected peak rack densities and thermal loads over a 5–10 year growth curve. 🔹 Design for redundancy and modularity. CDU N+1, dual pump trains, and scalable dry cooler banks provide operational resilience. 🔹 Optimize for water usage and energy cost. Dry cooling solutions paired with liquid systems reduce water footprint — critical in regions with water restrictions or high costs. 🔹 Plan for phased migration. Hybrid cooling eases transition from air-dominant halls into fully liquid-integrated facilities while protecting previous CapEx. Or get more details at: info@jusdon.com.cn, #LiquidCooling #DataCenter #AIInfrastructure #GreenComputing #ThermalManagement #EdgeComputing #AI #Hybridcooling #Cooler #Chiller #CDUs #AIContainer #AIFactory #HPC 

𝗡𝗲𝘄 𝗿𝗲𝘀𝗲𝗮𝗿𝗰𝗵 𝗵𝗶𝗴𝗵𝗹𝗶𝗴𝗵𝘁𝘀 𝗵𝗼𝘄 𝗱𝗮𝘁𝗮 𝗰𝗲𝗻𝘁𝗲𝗿𝘀 𝗰𝗮𝗻 𝗹𝗼𝘄𝗲𝗿 𝘁𝗵𝗲𝗶𝗿 𝗰𝗮𝗿𝗯𝗼𝗻, 𝗲𝗻𝗲𝗿𝗴𝘆, 𝗮𝗻𝗱 𝘄𝗮𝘁𝗲𝗿 𝗳𝗼𝗼𝘁𝗽𝗿𝗶𝗻𝘁𝘀 — 𝗳𝗿𝗼𝗺 𝗰𝗿𝗮𝗱𝗹𝗲 𝘁𝗼 𝗴𝗿𝗮𝘃𝗲. A new paper Nature Magazine from Microsoft researchers, (led by Husam Alissa and Teresa Nick), demonstrates the power of life cycle assessment (#LCA) to guide more sustainable data center design decisions — going beyond operational efficiency. 𝐊𝐞𝐲 𝐌𝐞𝐬𝐬𝐚𝐠𝐞:  While LCAs are often conducted after design and construction, this paper highlights the value of applying them much earlier. Integrated into early-stage design, LCAs help balance sustainability alongside feasibility and cost — leading to better trade-offs from the start. For example, the study found that switching from air cooling to cold plates that cool datacenter chips more directly – a newer technology that Microsoft is deploying in its datacenters – could: ▶️reduce GHG emissions and energy demand by ~15 % and ▶️reduce water consumption by ~30-50 % across the datacenters’ entire life spans. And this goes beyond cooling water. It includes water used in power generation, manufacturing, and across the entire value chain. As lead author Husam Alissa notes: "𝘞𝘦’𝘳𝘦 𝘢𝘥𝘷𝘰𝘤𝘢𝘵𝘪𝘯𝘨 𝘧𝘰𝘳 𝘭𝘪𝘧𝘦 𝘤𝘺𝘤𝘭𝘦 𝘢𝘴𝘴𝘦𝘴𝘴𝘮𝘦𝘯𝘵 𝘵𝘰𝘰𝘭𝘴 𝘵𝘰 𝘨𝘶𝘪𝘥𝘦 𝘦𝘯𝘨𝘪𝘯𝘦𝘦𝘳𝘪𝘯𝘨 𝘥𝘦𝘤𝘪𝘴𝘪𝘰𝘯𝘴 𝘦𝘢𝘳𝘭𝘺 𝘰𝘯 — 𝘢𝘯𝘥 𝘴𝘩𝘢𝘳𝘪𝘯𝘨 𝘵𝘩𝘦𝘮 𝘸𝘪𝘥𝘦𝘭𝘺 𝘵𝘰 𝘮𝘢𝘬𝘦 𝘢𝘥𝘰𝘱𝘵𝘪𝘰𝘯 𝘦𝘢𝘴𝘪𝘦𝘳." To support broader adoption, the team is making the methodology open and available to the industry via an open research repository: https://lnkd.in/gC5jdkMs The work builds on Microsoft’s continued efforts to construct unified life cycle assessment methods and tools for cloud providers. (read more about this here: https://lnkd.in/gq24wMrA) 𝐑𝐞𝐚𝐝 𝘁𝗵𝗲 𝗳𝘂𝗹𝗹 𝗽𝗮𝗽𝗲𝗿 𝗵𝗲𝗿𝗲: 👉https://lnkd.in/gVm25zzh #sustainability #climateaction #innovation #sciencetoaction

Datacenter sustainability starts long before the servers are switched on. ⬇️ Microsoft’s 2026 Environmental Sustainability Report maps the environmental impact of datacenters across three stages: how they are designed, how they are built, and how they are operated. Design decisions influence years of resource consumption. Microsoft highlights power and compute efficiency, lower water use for cooling, adaptation to local environmental conditions, and support for community ecosystem restoration. Construction adds embodied carbon, material demand, transport emissions, and waste. The report points to timber framing, low-carbon concrete, green steel, LEED certification, lower-emission logistics, and better management of construction waste. Operational priorities include clean power and fuels, stronger energy markets, water replenishment, waste-heat reuse, hardware recovery, and greater circularity for rare-earth elements and cloud equipment. These interventions are closely connected. More efficient computing reduces energy demand. Lower-carbon materials reduce emissions before operations begin. Longer hardware life reduces demand for new materials. Water strategies must reflect the conditions of the local watershed. For companies expanding AI capacity, environmental performance should cover concrete, steel, water, grids, hardware, logistics, ecosystems, and end-of-life management—not simply the electricity consumed once a facility is running. The environmental footprint of AI is being determined through thousands of infrastructure decisions made across the datacenter lifecycle. #sustainability #esg

Somewhere right now, a landowner is getting a call from a data center developer, and the number being floated is several times what the land was worth last year. Maybe you've read those stories. Maybe you own a large parcel near a transmission line, the highway is close, the land is flat, and you've wondered: could that be me? Maybe, but the acreage and nearby power lines are only the beginning. A developer starts with a series of gating questions. Can the site actually get enough power, and when? Proximity to transmission doesn't mean capacity is available. The utility may need a new substation, major upgrades, or years of work before the site can be energized. Is the land physically workable? Size matters, but so do shape, grading, drainage, wetlands, flood risk, soils, setbacks, and room for substations, generators, cooling, and future phases. Can the facility connect? Not ordinary broadband: access to major fiber routes, ideally multiple independent paths into the site. Can it be permitted? Zoning, environmental review, noise, air and water rules, and the local approval process can eliminate an otherwise promising property. Can it be built? Transformers, generators, and modular components have to physically reach the site. Heavy-haul routes, staging areas, workforce availability, and lodging all affect cost and schedule. Can it be cooled and operated responsibly? Air, water, reclaimed water, or a combination, the right answer depends on the project and the geography around it. Will the community support it? Jobs and tax base matter, but residents will ask about noise, traffic, water, and what the project leaves behind for the people who host it. Also, if a developer does call, what usually comes first isn't a purchase, it's an option agreement: the right to buy your land later while they spend months or years confirming power and permits. Many options are never exercised. A site can look perfect on a map and fail in diligence. Until those questions are answered, "data center land" is usually just land with data center potential. The best sites aren't the biggest or the cheapest. They're the places where power, connectivity, approvals, constructability, and community support all come together on the same schedule. Cumulus #DataCenters #SiteSelection #DataCenterDevelopment #WorkDoneRight

$2.3T data center pipeline. 68% still in pre-planning (GlobalData Q4 2025) Meaning: most of the market is still in spec-writing mode, not execution. Three gates are stalling projects: Power — With grid caps and connection timelines, power is the bottleneck. At AI training densities (often 50–120 kW/rack), parasitic load (fans, pumps) directly reduces the IT power you can monetize. Water — In many regions, WUE is becoming a permitting gate, not a KPI. Water dependency = regulator negotiation (and often schedule risk). Permitting — Noise, land use, emissions, local impact. Projects slip before procurement. Hyperscalers increasingly co-design specs early and lock long-lead supply upstream to clear these gates. So here’s the shift: architecture is the differentiator. Many specs still treat redundancy like arithmetic: N+1 as “one full extra dry cooler” is often a blunt instrument — footprint-heavy and CAPEX-intensive. But AI failures are frequently local: fan, sensor, valve, fouling. Monolithic designs turn local issues into system events. Modular changes how you engineer redundancy. You don’t improvise resilience — you design it upfront: N+1 = one extra module, not one extra unit. What that enables (when isolation and controls are designed to avoid common-mode events): Fast serviceability: isolate a module, keep the rest online. Module-level interventions can be planned around weather forecasts and load profiles, with no heavy lifts when access and spares are prepared. Granular load following: finer staging to match IT load → better part-load efficiency, tighter temperature control, lower parasitic energy. Failure containment: smaller blast radius → engineered degradation, not downtime. Second-order effect: footprint. An extra full dry cooler burns roof area and structural CAPEX. Granular redundancy protects uptime without sacrificing m² → higher heat-rejection density. The real question: which architecture clears the power/water/permitting gates and stays upgradeable as densities keep moving? At ThermoKey, we build Modular Dry Coolers and manufacture air-to-liquid aluminum microchannel heat-exchanger cores in-house, enabling geometry optimization for DC constraints: approach temperature, fan power, pressure drop, footprint, and scalability. (Patent pending) Defining a thermal concept for high-density liquid cooling? Let’s talk. #DataCenter #AICooling #Modular #LiquidCooling #ThermalManagement #WUE #PUE

𝐃𝐚𝐭𝐚 𝐂𝐞𝐧𝐭𝐞𝐫 𝐈𝐧𝐟𝐫𝐚𝐬𝐭𝐫𝐮𝐜𝐭𝐮𝐫𝐞 𝐃𝐞𝐬𝐢𝐠𝐧: The Impact of Rack Density on Facility Architecture One of the most important decisions in data center design is determining the target rack power density. A seemingly simple change from 8 kW per rack to 50 kW, 100 kW, or even 1 MW per rack fundamentally transforms the electrical, mechanical, and thermal infrastructure requirements of the entire facility. 𝐑𝐚𝐜𝐤 𝐃𝐞𝐧𝐬𝐢𝐭𝐲 𝐂𝐨𝐦𝐩𝐚𝐫𝐢𝐬𝐨𝐧 - 8 kW per Rack 200 racks = 1.6 MW IT Load Air-cooled environment Standard UPS and generator sizing Lower cooling requirements Traditional enterprise data center design - 10 kW per Rack 200 racks = 2.0 MW IT Load 25% increase in power demand Larger UPS systems Increased heat rejection requirements Enhanced electrical distribution infrastructure - 50 kW per Rack 200 racks = 10 MW IT Load Direct-to-chip liquid cooling becomes practical Large-scale CDU deployment Higher-capacity switchgear and transformers AI and HPC infrastructure class - 1 MW per Rack 200 racks = 200 MW IT Load Utility-scale power requirements Full liquid or immersion cooling Dedicated substations Massive heat rejection systems Campus-scale energy infrastructure Infrastructure Requirements That Scale with Rack Density (1) Power Systems Utility interconnection Substations Transformers Switchgear UPS / BESS Backup generation (2) Cooling Systems CRAH / CRAC units Chillers and dry coolers CDU systems Direct-to-chip cooling Immersion cooling Heat exchangers (3) Water & Thermal Management Closed-loop cooling systems Water treatment Thermal storage Heat recovery Leak detection Heat reuse opportunities (4) Controls & Monitoring Digital twins AI-driven optimization Real-time thermal monitoring Power quality management Predictive maintenance The Key Engineering Reality Every watt consumed by compute becomes heat. As rack density increases, the design challenge shifts from simply powering servers to managing massive amounts of energy and heat safely, efficiently, and reliably. The future of AI infrastructure is driving data centers toward utility-scale engineering, where power generation, cooling systems, energy storage, and digital controls become just as important as the compute itself. Rack density no longer defines only the rack. It defines the entire facility. #DataCenter #AIInfrastructure #Hyperscale #LiquidCooling #DataCenterDesign #ElectricalEngineering #MechanicalEngineering #ThermalManagement #DigitalTwin #BESS #PowerSystems #CoolingSystems #AIDataCenters #IndustrialEngineering #FutureOfAI