From Operators to Innovators: Inside Midas Immersion Cooling
The data center industry stands at an inflection point. As AI-enabled workloads drive compute densities beyond 100 kilowatts per rack, traditional air cooling approaches are reaching their limits. My recent conversation with Solidigm’s Jeniece Wnorowski and Scott Sickmiller, CEO of Midas, revealed how immersion cooling technology has evolved into a practical solution for today’s most demanding workloads.
What makes Midas’s perspective particularly valuable is their origin story. Unlike companies that developed immersion cooling as a product, Midas became a provider because they were first a user facing real cooling challenges in their Austin data center.
From Necessity to Innovation
Midas began as a data center operation in 2011, quickly becoming the go-to provider for hard-to-cool IT infrastructure. The growth trajectory forced them to look beyond traditional air cooling solutions. Between 2011 and 2012, the team iterated through multiple immersion cooling designs, ultimately developing and patenting their own solution. In 2016, they made the decision to exit the data center business and focus exclusively on providing immersion cooling infrastructure to the industry. “And the rest, as they say, is history of 4,000 tanks,” Scott said.
This user-first development approach shapes everything about Midas’s technology today. As Scott explained, having to maintain the systems themselves drove design decisions toward user-friendliness and operational efficiency that competitors who never operated the technology might overlook.
The Physics of Immersion
At its core, immersion cooling leverages a simple advantage: liquids dissipate heat approximately 1,200 times more effectively than air. By submerging IT equipment, data centers immediately gain this thermal efficiency advantage. However, as Scott emphasized, doing immersion well requires more than just dunking servers in liquid.
Early on, the team learned that success depended on computational fluid dynamics (CFDs). CFDs are critical to ensuring that the dielectric liquid reaches all heat sources, engages with them, and moves away from them with a uniform flow. CFDs are crucial to ensuring that this happens no matter the rack’s form factor. While adapting to diverse hardware designs is a challenge, Scott noted, “At the end of the day, it’s only physics. So the physics can support the workload. We just have to fit the form factor into the physics box.”
The Thermal Recovery Advantage
Beyond raw cooling efficiency, immersion cooling enables thermal energy recovery in ways air-cooled systems cannot match. The dielectric fluid not only captures heat more effectively than air, it also retains that heat longer, enabling efficient transfer to other systems.
Scott shared an example from a recent meeting with a German district heating facility. In district heating, water or another fluid is centrally heated and then pumped out into a distribution network, eventually reaching buildings where it regulates temperature through boilers. When a data center can provide water at 50° Celsius (122° Fahrenheit), this represents a significant opportunity to reuse energy already consumed for computing. The economics are compelling. “We’ve already paid for the energy once,” Scott said. “So at that point, why not use it again? And that’s where thermal recovery is really useful.”
Practical Deployment Considerations
Immersion cooling shows strong return on investment above 40 kilowatts per rack, and the technology becomes necessary at 100 kilowatts and beyond. As advancement of graphics processing units (GPUs) drives power densities higher, direct liquid cooling alone cannot solve the challenge. Peripheral components still generate heat requiring air cooling, straining facility infrastructure as power becomes the ultimate constraint.
The barrier that Midas faced for 15 years—data center operators’ resistance to liquid near equipment—has been addressed as organizations adopt rear-door heat exchangers and direct-to-chip cooling. “Many of the data centers, especially the ones that are focusing on machine learning and AI, are building water loops in the facility,” Scott said. “So that prerequisite is done. Then we need to start looking at the IT.”
The IT requirements are “quite a bit different.” One of the biggest changes? Fans are no longer needed, and the immediate benefit is significant. A one-kilowatt server that dedicates 150 to 200 watts to fans can complete the same compute at just 800 watts immersed.
The Midas Difference
What distinguishes Midas in an increasingly competitive market comes back to their operational heritage. Scott highlighted their truly concurrent maintainability and fault-tolerant design, which includes redundant cooling distribution units (CDUs) as standard. The system supports easily hot-swapping failed CDUs: with just an hour of education, the company’s global sales manager learned to hot-swap a CDU in seven minutes. The operational simplicity extends to deployment, as well. Scott described installing a system at a university in the United Kingdom in 40 minutes. “That’s an advantage of a Midas,” Scott said. “We had to maintain it ourselves, so we built it that way.”
TechArena Take
Midas’s journey from data center operator to immersion cooling provider demonstrates how real operational experience drives practical innovation. Their emphasis on user-friendly design addresses the large and small daily challenges that data center operators face. As compute densities continue climbing and power constraints tighten, immersion cooling is transitioning from alternative technology to essential infrastructure. Companies like Midas, with proven deployment experience and field-tested designs, are well-positioned to lead this transformation.
Learn more about Midas immersion cooling solutions at www.midasimmersion.com.
The data center industry stands at an inflection point. As AI-enabled workloads drive compute densities beyond 100 kilowatts per rack, traditional air cooling approaches are reaching their limits. My recent conversation with Solidigm’s Jeniece Wnorowski and Scott Sickmiller, CEO of Midas, revealed how immersion cooling technology has evolved into a practical solution for today’s most demanding workloads.
What makes Midas’s perspective particularly valuable is their origin story. Unlike companies that developed immersion cooling as a product, Midas became a provider because they were first a user facing real cooling challenges in their Austin data center.
From Necessity to Innovation
Midas began as a data center operation in 2011, quickly becoming the go-to provider for hard-to-cool IT infrastructure. The growth trajectory forced them to look beyond traditional air cooling solutions. Between 2011 and 2012, the team iterated through multiple immersion cooling designs, ultimately developing and patenting their own solution. In 2016, they made the decision to exit the data center business and focus exclusively on providing immersion cooling infrastructure to the industry. “And the rest, as they say, is history of 4,000 tanks,” Scott said.
This user-first development approach shapes everything about Midas’s technology today. As Scott explained, having to maintain the systems themselves drove design decisions toward user-friendliness and operational efficiency that competitors who never operated the technology might overlook.
The Physics of Immersion
At its core, immersion cooling leverages a simple advantage: liquids dissipate heat approximately 1,200 times more effectively than air. By submerging IT equipment, data centers immediately gain this thermal efficiency advantage. However, as Scott emphasized, doing immersion well requires more than just dunking servers in liquid.
Early on, the team learned that success depended on computational fluid dynamics (CFDs). CFDs are critical to ensuring that the dielectric liquid reaches all heat sources, engages with them, and moves away from them with a uniform flow. CFDs are crucial to ensuring that this happens no matter the rack’s form factor. While adapting to diverse hardware designs is a challenge, Scott noted, “At the end of the day, it’s only physics. So the physics can support the workload. We just have to fit the form factor into the physics box.”
The Thermal Recovery Advantage
Beyond raw cooling efficiency, immersion cooling enables thermal energy recovery in ways air-cooled systems cannot match. The dielectric fluid not only captures heat more effectively than air, it also retains that heat longer, enabling efficient transfer to other systems.
Scott shared an example from a recent meeting with a German district heating facility. In district heating, water or another fluid is centrally heated and then pumped out into a distribution network, eventually reaching buildings where it regulates temperature through boilers. When a data center can provide water at 50° Celsius (122° Fahrenheit), this represents a significant opportunity to reuse energy already consumed for computing. The economics are compelling. “We’ve already paid for the energy once,” Scott said. “So at that point, why not use it again? And that’s where thermal recovery is really useful.”
Practical Deployment Considerations
Immersion cooling shows strong return on investment above 40 kilowatts per rack, and the technology becomes necessary at 100 kilowatts and beyond. As advancement of graphics processing units (GPUs) drives power densities higher, direct liquid cooling alone cannot solve the challenge. Peripheral components still generate heat requiring air cooling, straining facility infrastructure as power becomes the ultimate constraint.
The barrier that Midas faced for 15 years—data center operators’ resistance to liquid near equipment—has been addressed as organizations adopt rear-door heat exchangers and direct-to-chip cooling. “Many of the data centers, especially the ones that are focusing on machine learning and AI, are building water loops in the facility,” Scott said. “So that prerequisite is done. Then we need to start looking at the IT.”
The IT requirements are “quite a bit different.” One of the biggest changes? Fans are no longer needed, and the immediate benefit is significant. A one-kilowatt server that dedicates 150 to 200 watts to fans can complete the same compute at just 800 watts immersed.
The Midas Difference
What distinguishes Midas in an increasingly competitive market comes back to their operational heritage. Scott highlighted their truly concurrent maintainability and fault-tolerant design, which includes redundant cooling distribution units (CDUs) as standard. The system supports easily hot-swapping failed CDUs: with just an hour of education, the company’s global sales manager learned to hot-swap a CDU in seven minutes. The operational simplicity extends to deployment, as well. Scott described installing a system at a university in the United Kingdom in 40 minutes. “That’s an advantage of a Midas,” Scott said. “We had to maintain it ourselves, so we built it that way.”
TechArena Take
Midas’s journey from data center operator to immersion cooling provider demonstrates how real operational experience drives practical innovation. Their emphasis on user-friendly design addresses the large and small daily challenges that data center operators face. As compute densities continue climbing and power constraints tighten, immersion cooling is transitioning from alternative technology to essential infrastructure. Companies like Midas, with proven deployment experience and field-tested designs, are well-positioned to lead this transformation.
Learn more about Midas immersion cooling solutions at www.midasimmersion.com.
.png)
