Satya Nadella’s Post

Do you think next models of #servers and other IT #equipment will be ready to handle what #AI does to it? Yes! Does that mean your #datacenter will change? Yes! Be prepared. Things are happening! ”Microsoft has successfully tested a new cooling system that removed heat up to three times better than cold plates, an advanced cooling technology commonly used today. It uses microfluidics, an approach that brings liquid coolant directly inside the silicon – where the heat is. Tiny channels are etched directly on the back of the silicon chip, creating grooves that allow cooling liquid to flow directly onto the chip and more efficiently remove heat.” #IT #Infrastructure #LiquidCooling #ChipCooling #Microfluidics

This is interesting but with some caveats: 1. Until there is an industry standard way to do this, it will be Microsoft 1st party silicon specific thus limiting the upside. 2. What is the liquid used? A lot of coolants contain PFAS and we should be using less of these types of liquids. 3. Lastly, this does very little (maybe allows for higher density in the DC but that seems a long way away and speculative at this point) to impact scope 3 emissions which represent well over 90% of Microsoft's emissions.

Steaming Hot off the press. Microfluidics, a breakthrough in cooling technology, which brings cooling directly into the electronic chips/die (Silicon in this case below). Microfluidics aka Embedded cooling offers different advantages. 1. Enables high density chips to support the ongoing AI factory explosion. 2. By bringing the coolant closer to the die, you could avoid the intermediate resistances to heat transfer which were otherwise present with conventional cooling technologies. You could derive 3X more performance using microfluidics. 3. By bringing the coolant on to the generating die, you could actually bring the coolant hotter > 70deg C in certain scenarios. (since we avoided the thermal stack). This certainly opens us an opportunity to capture this high-grade waste heat at the datacenter facility level for different heat offtake applications. 4. More importantly, Microfluidics is the only way, we could get to the 3D Heterogenous Integration (3D HI). So, what you are seeing is the fundamental building block of the compute infrastructure of the future. Dreams do get real. Not many times, you get to contribute to real impactful work, which has visibility up to the very TOP. Much gratitude to our management for the continued trust and support for advancing groundbreaking innovation and later providing the limelight and visibility 🙏. Thanks to our external partnership with Corintis, GA Tech, IEEE HIR and academic consortia for advancing this technology. Stay tuned! Link to Microsoft blog post on Microfluidics. Please do check it out. You will be amazed. https://lnkd.in/gDXCcznZ #Microfluidics #Embeddedcooling #ThroughSiliconCooling #3DHI #2.5D #Innovation @ Microsoft

🔥 AI Isn’t Just Hot It’s Reaching the Melting Point 🔥 The world talks about algorithms, models, and breakthroughs but few stop to consider the physics that underpins it all. Every new generation of AI chips runs hotter, denser, and closer to its physical breaking point. Traditional cooling methods airflow and cold plates are already straining. In five years or less, without radical innovation, the ceiling on progress is not compute, it’s heat. Microsoft Microsoft AI just announced a breakthrough: in-chip microfluidic cooling. Instead of relying on external plates, they etched microscopic channels directly into the silicon. Liquid coolant flows through these grooves, removing heat at the source—up to 3x more efficiently than today’s best cold plates. That’s not an incremental upgrade. It’s a structural shift. 💡 What it unlocks: • More power in less space → higher density datacenters without adding new buildings. • Sustainability → reduced energy demand and lower operational costs. • Overclocking without fear → running chips harder without frying them. • 3D architectures → stacking chips vertically to slash latency. • New innovation pathways → removing heat barriers opens doors to designs that were once “impossible.” The design itself was inspired by nature. Using AI, Microsoft modeled bio-inspired patterns resembling veins in a leaf or butterfly wings—optimized to direct coolant where it’s needed most. Nature has always been the best engineer; now AI is helping us decode and apply it. But here’s the larger truth: this isn’t about one company’s advantage. This is about the entire AI ecosystem facing a reckoning. Whoever masters heat management, sustainability, and system-level efficiency will hold the keys to the next era of AI. It’s not just about writing better models it’s about re-engineering the foundation they run on. ⚖️ And that matters for all of us. Because AI isn’t just software. It’s hardware, physics, and energy grids. It’s about whether communities can support next-generation datacenters without blackouts. It’s about whether the infrastructure can scale to match the dreams being pitched. Microsoft’s announcement shows one path. But the stakes are bigger than any single lab or company. The industry cannot afford to treat cooling as an afterthought. Without it, the AI revolution burns out before it reaches its promise. I watch this closely not only as an innovator, but as someone litigating in federal court with AI itself as a core tool. When the history of this era is written, it won’t just be about models. It will be about whether we solved the limits of physics in time. Because AI isn’t just hot it’s a fire testing who is ready to hold it. ⸻ 🔗 Satya Nadella Judy Priest #AI #Innovation #Sustainability #Microsoft #FutureOfWork #DataCenters #Microfluidics #PublicInterest #Leadership #AIInfrastructure

Great to see chip designers going all the way into the chip with their microfluidic channels, very inspiring. Accelsius researchers Akshith Narayanan, Qingyang Wang, Serdar Ozguc, Trevor Whitaker, and Jacob Moore will be presenting a similar concept at Interpack in a few weeks, where we use two-phase microfluidic channels (with non-electricity conducting fluids) instead of water.

Meeting the demands of next-generation AI requires innovation at every layer. This includes the chip itself. At Microsoft, we are demonstrating a new way to cool silicon chips more directly using microfluidics. By bringing liquid coolant inside the silicon, we’re addressing one of the most pressing challenges in infrastructure: heat.

This has been one of the most exciting projects I’ve had the chance to be part of at Microsoft. We successfully built the industry’s first microfluidic cooled chip. An in-chip microfluidic cooling system that can precisely and effectively cool several parts of a server. In one of our internal demos, we even managed to run a Teams meeting on the server powered by this technology! Do check out the blog post for more details.

Big milestone for our portfolio company Corintis: Microsoft just announced a breakthrough in liquid cooling, embedding Corintis’ microfluidics directly into chips to cool them up to 3x more effectively than today’s best systems. AI’s future isn’t just about faster chips, it’s about solving the physical limits that stand in the way. Heat dissipation has become a defining bottleneck for compute and Corintis is proving there’s a better path forward. Proud to back Remco van Erp, Sam Harrison and the Corintis team as they tackle one of the most mission-critical challenges in compute.

Chip-cooling breakthrough! 🧊 Exciting day as our collaboration with Microsoft goes public: Achieving staggering 3X lower chip temperatures compared to standard cold plates with microfluidic cooling, co-designed with the chip, and embedded inside the silicon. Cooling from within, making microfluidics a fundamental part of the chip as an enabler of future super-efficient AI infrastructure! Cooling is the ultimate bottleneck for the future of chips. In a time where every degree matters, a 3x reduction in temperature speaks for itself. As the article states, this result was achieved on functional chips, in real servers, running real workloads. Thank you to all the people who worked hard on this behind the scenes at Microsoft and here at Corintis and the great collaboration to make this possible, and to innovate on many fronts. We’re beyond proud to see microfluidic cooling reaching the main stage. This is just the beginning, there is so much more to come as we’re building towards 10x better cooling to enable the next 100x better chips. Stay tuned for more soon!  ❄️❄️ This great quote perfectly captures our excitement around microfluidic cooling: Jim Kleewein “Hardware is the foundation of our services. We all have a vested interest in that foundation – how reliable it is, how cost effective, how fast, how consistent the behavior we can get from it, and how sustainable, to name just a few. Microfluidics improves each of those: cost, reliability, speed, consistency of behavior, sustainability.” 🔥 Husam Alissa & Dr. Sashikanth Majety and many more!

Coolest news of the week just dropped in hot 🔥: Microsoft collaborates with Corintis to achieve 3X lower chip temperatures. Cooling is one of the major challenges for the future of chips. Just a few years ago when we started Corintis, chips were only several hundreds of Watts. Today we’re looking at an order of magnitude increase in just a few years. Cooling systems will need to evolve rapidly over the next years, to keep up with the rising power levels, and to ensure an efficient and sustainable AI buildout. These results published by Microsoft show a path to achieve just that. The secret? Microfluidic cooling. A network of thousands of small channels, optimized with microscopic precision to bring the right amount of coolant to the right location on the chip. As chips are getting bigger, more powerful, and more heterogeneous, chip-specific cooling is no longer a nice-to-have, but a must. And secondly, integrating the cooling directly inside the chip, to eliminate all barriers for heat transfer, with reliable packaging that is compliant with industry-standards of today’s cold plates.  Co-design between the chip and the cooling system, and co-packaging cooling with the chip unlocks a new dimension in cooling, minimizing temperatures in the areas where it matters most, and enabling microfluidic cooling to be backwards compatible with today’s D2C liquid cooling infrastructure. At Corintis, we’re building a bridge between the world of chips and the world of liquid cooling, to make microfluidic cooling a reality for everyone. With our platform to design, build, test and mass produce high performance liquid cooling, adapted to the chip, we help our customers bring better products to the market faster. So proud to see this collaboration go public, and share our excitement around microfluidic cooling with the world. A big thank you to all the people at Microsoft for this strong collaboration (Husam Alissa Dr. Sashikanth Majety and others), and the whole team here at Corintis for making it happen. This is just the beginning, there is so much more to come as we’re building towards 10x better cooling to enable the next 100x better chips. Stay tuned for more soon!