For the last few years, the AI trade has been dominated by one question:
Who makes the chips?
$NVDA, $AMD, $MU, $ARM and the rest of the semiconductor group have captured most of the attention.
But Thursday’s move in $GNRC got me thinking about a different part of the AI equation:
Who actually powers all these data centers?
$GNRC surged after announcing a long-term agreement with Amazon to supply backup generators for its data centers. The initial deliveries are expected to total $2.4B in 2027–2028, while the agreement could ultimately support purchases of up to $8B. 
That is a pretty big signal.
AI data centers need enormous amounts of electricity — and they need that power to be reliable, not just available.
The IEA expects electricity demand to accelerate through 2030, with data centers and AI among the drivers. It also highlights the increasing power density of AI servers and the pressure this creates on electricity infrastructure. 
That creates a much bigger AI infrastructure chain:
🧠 Compute — GPUs & CPUs
💾 Memory — HBM & DRAM
🌐 Networking — moving all that data
⚡ Power — generation & backup
🏗️ Data centers — where everything comes together
And this is where I think the market gets interesting.
We may be entering a phase where power availability becomes just as important as computing capacity.
But I wouldn’t blindly chase $GNRC after the surge.
The Amazon deal is significant, but there are still questions around valuation, execution and the warrants Amazon received. Amazon received warrants for up to 1.69M GNRC shares, with vesting linked to purchases under the agreement. 
So the bigger question isn’t simply:
“Is $GNRC going higher?”
It’s:
“Are we watching the beginning of a broader AI power-infrastructure trade?”
If AI capex keeps expanding, the winners may not be limited to the companies designing the next GPU.
They could also be the companies providing the electricity, backup power, cooling, transformers and infrastructure needed to keep those GPUs running.
👀 Which part of the AI infrastructure chain do you think gets the most attention next — chips, memory, networking, power or data centers?
Comments