Importance of Liquid Cooling for AI Systems

The next wave of AI infrastructure faces a critical challenge. Cooling, not power, is the true bottleneck... With NVIDIA's Vera Rubin Ultra requiring 600kW racks by 2027, cooling infrastructure will reshape the global data center landscape. According to Uptime Institute's research, cooling systems consume up to 40% of data center energy. Yet the physics of thermal transfer create absolute limits on what air cooling can achieve. This reality is creating a three-tier market segmentation worldwide: 1. Liquid-cooled facilities (30-150kW+ per rack) capturing premium AI workloads 2. Enhanced air-cooled sites (15-20kW per rack) limited to standard enterprise computing 3. Legacy facilities facing increasing competitive disadvantages The challenge manifests differently across regions: - Tropical markets (#Singapore, #Brazil) battle 90%+ humidity that reduces cooling efficiency by 40% - Water-stressed regions face constraints with cooling towers consuming millions of liters daily - Temperate regions benefit from free cooling opportunities but still require liquid solutions for #AI densities Regional innovations demonstrate tailored approaches: 1. #Singapore's STDCT has achieved PUE values below 1.2 despite challenging humidity 2. #SouthAfrica's MTN deployed solar cooling to address energy reliability concerns 3. #Jakarta's SpaceDC uses specialized designs for both climate and power stability challenges Research from ASME shows that transitioning to 75% liquid cooling can reduce facility power use by 27% while enabling next-gen compute densities in any climate. The Global Cooling Adaptation Framework provides a strategic approach: 1. Regional Climate Assessment 2. Thermal Capacity Planning 3. Water-Energy Optimization 4. Infrastructure Evolution Timeline For investors, the implications extend beyond operations. Facilities with limited cooling capabilities may find themselves at a disadvantage when competing for higher-margin segments, regardless of location advantages. What cooling strategies is your organization implementing to prepare for the 600kW future? Read the full analysis in this week's article. #datacenters

We are not talking about it enough but data center cooling isn’t just a technical constraint anymore, it’s a wall we’re about to crash into. Air cooling is already maxed out under modern AI workloads. Liquid cooling isn’t futuristic anymore, it’s baseline. But even that won’t get us where we need to go. The next leap in computing isn’t just about silicon. It’s about materials, the ones that move heat faster, more efficiently, with less space and energy overhead. Right now, the situation is crucial: → Direct-to-chip liquid cooling is here and scaling. It’s a good step, but it doesn’t solve component-level hotspots. → Immersion cooling is being rolled out by Microsoft, Meta, and others. Great for racks, but not a silver bullet. → Liquid metals, high-end phase change materials, and engineered coolants are improving edge-level thermal interfaces, but they hit their limits fast. One material getting serious attention is synthetic diamond. Its thermal conductivity is unmatched. Some startups, like Akash Systems, are already using diamond for heat spreaders and RF devices, claiming measurable performance gains. There’s even early work on growing diamond films directly on silicon — a concept that, if scalable, could shift how we build thermal pathways in packaging. But diamond isn’t a magic solution. It’s expensive. Manufacturing is complex. Integration with standard processes is still a challenge. Still, the interest isn’t hype. The physics is real. And as compute density increases, it’s clear we’ll need new materials in the stack to handle the thermal load. If you're building AI-scale infrastructure and not exploring this layer of the problem — the materials layer — you’re not preparing for what’s coming. Because it won’t be the airflow that holds you back. It’ll be heat. #DataCenterCooling #AdvancedMaterials #LiquidCooling #FutureOfComputing

🌡️ The Heat is On in the Data Center – And It’s Time to Cool Smarter 💧 As AI workloads surge and rack densities climb, traditional air cooling is hitting its limits. The modern data center demands a new approach—one that’s efficient, sustainable, and ready for the future. That’s where liquid cooling steps in...Full Video: https://lnkd.in/gsg3ABMx Among the most innovative solutions on the market is the Enclosed Rear Door Heat Exchanger from Dell Technologies. This system brings liquid cooling directly to the rack, capturing and dissipating heat at the source—without disrupting existing airflow or requiring a full data center redesign. ✅ Why it matters: Supports higher rack densities Reduces energy consumption and operational costs Enables sustainable scaling for AI and HPC environments Dell PowerCool Enclosed RDHx is a powerful example of how we can meet the thermal challenges of tomorrow—today. 🔗 If your data center is feeling the heat, it might be time to explore liquid cooling solutions that are as forward-thinking as your workloads. Olivia Mauger #DataCenter #LiquidCooling #DellTechnologies #Sustainability #AIInfrastructure #HPC #ThermalManagement #Innovation #IWork4Dell

Got to tour my first Data Center yesterday.  Here are some takeaways from it.  Note, I’m not an expert in Data Centers and these thoughts are more from a curious investor standpoint. First off, special thank you to Allison Boen and Shell who was kind enough to host myself, Zach and Joe during the lunch and learn session with industry leaders from Green Revolution Cooling, Supermicro, Shell and AMD to name a few.. The data center we went into was a Shell Data Center. With our HVAC background, and our recent exit of Flow Service Partners, our first thought on data centers was, let’s dig into the cooling side of things. 1.  Immersion cooling is a need, not a want. The rise in kW per rack and TDP due to the growth in AI and GPU processing is resulting in rack densities that are simply too hot to be cooled through computer room air conditioning ("CRAC" - i.e., traditional AC). 2.  Companies should be excited about this! Immersion cooling has immense benefits - decreasing total cost of ownership, increase in computing densities, decreasing failure rates and increasing hardware lifespans, and increasing the number of racks able to run at once due to less space needed for cooling tanks. The dielectric fluid Shell is producing is recyclable and environmentally friendly. And don't even get me started on the noise factor of CRACs vs. immersion cooling tanks.…the sound would make you go crazy if you stayed in there there too long. 3. The current resistance and hesitancy to adopt immersion cooling techniques can be boiled down to two factors. One being the capital outlay needing to retrofit facilities (this will become less of an issue as new data centers are built to be liquid cooling friendly) and the other being the lack of warranties offered on some of the chips. Some companies currently offer product warranties, but most do not. Once more warranties are offered, I think the floodgates towards liquid cooling adoption will be opened. 4. Some hyperscalers and operators are currently utilizing immersion cooling for ~30% of total cooling efforts. If a data center is already hooked into the grid and can draw a predefined amount of power, then immersion cooling is the only way to get more density without generating a higher level of draw from the grid. This could result an additional 35% of total power usage that goes to conventional HVAC systems switching over to cool racks. 5. All this boils down to I think the industry will more and more shift towards immersion cooling.  There was even talk of combining direct to chip and immersion cooling. If you are looking for an investment in the data center space, would love to chat.