Everyone is talking about the AI on their screens. We debate how smart ChatGPT is or whether it will take our jobs.
But almost no one is talking about how hot the ground is beneath this intelligence.
I recently came across an analysis titled “The AI Invisibles” in the Macro Notes newsletter. The author, Paul, details an event in Northern Virginia that didn’t make the headlines but explains everything about the current moment:
A data center went dark for exactly 47 minutes. Not because of a cyberattack. Not because of a power grid failure.
Because of thermodynamics.
New clusters of H100 GPUs generated such intense heat that the cooling system couldn’t handle the thermal load. The system spiked to critical temperatures within seconds. While engineers scrambled to reroute coolant flow, millions of dollars of “intelligence” sat silent—just a pile of overheating metal.
Seeing the Invisible
This story is the untold side of the AI revolution. Everyone is buying Nvidia stock, but no one is talking about companies like Vertiv that cool those chips, or the Amphenol cables carrying that massive power.
As the Macro Notes author puts it: “The market is measuring the wrong layer. The value isn’t capturing the application layer everyone sees; it’s accumulating in the infrastructure layer nobody sees.”
In the Gold Rush, the winners weren’t the ones digging for gold. They were the ones selling shovels.
In AI, the winners will likely be: Not the ones writing the models, but the ones making the models run.
Consider what lies behind a single ChatGPT query:
Hundreds of watts of continuous power per GPU.
Precision liquid cooling with ±2°C tolerance.
Optical interconnects transferring hundreds of Gbps between racks.
Specialized power connectors rated for high amperage.
Clean energy backed up by massive generators.
ChatGPT or OpenAI produces none of this. The companies operating in this layer are quiet, little-known, but incredibly powerful.
To solve this problem, data centers are shifting to “Liquid Cooling”—expensive, but effective. They are reinforcing their infrastructure.
The lesson is simple: Don’t just watch the screen. Watch the machine behind it.
Ali YAZICI is a Senior IT Infrastructure Manager with 15+ years of enterprise experience. While a recognized expert in datacenter architecture, multi-cloud environments, storage, and advanced data protection and Commvault automation , his current focus is on next-generation datacenter technologies, including NVIDIA GPU architecture, high-performance server virtualization, and implementing AI-driven tools. He shares his practical, hands-on experience and combination of his personal field notes and “Expert-Driven AI.” he use AI tools as an assistant to structure drafts, which he then heavily edit, fact-check, and infuse with my own practical experience, original screenshots , and “in-the-trenches” insights that only a human expert can provide.
