The physical demands of AI have hit an infrastructure wall. While Big Tech remains fixated on training larger models and securing faster chips, a primitive bottleneck is being exposed.
“AI is going to be driven by power,” former US Energy Secretary Rick Perry told Yahoo Finance’s Opening Bid.
According to the former Texas governor, energy production and the construction trades are the two critical “limiting factors” that will determine which regions and companies can actually deploy AI at scale.
Power is no longer a theoretical concern for the sector. Big Tech companies — specifically Google (GOOGL, GOOG), Meta (META), Amazon (AMZN), and Microsoft (MSFT) — are investing heavily in nuclear energy to bypass an aging national grid. Microsoft’s deal with Constellation Energy (CEG) to restart Three Mile Island and Google’s deal with Kairos Power underscore an insatiable appetite for 24/7 carbon-free electrons.
This infrastructure crunch is creating a new competitive landscape where early movers and power-rich states have a distinct advantage.
Perry pointed to Mississippi and Texas as states positioned to “get ahead” by simplifying the process of building traditional baseload — oil, gas, and coal — while championing nuclear as the long-term solution.
He predicted that unprepared regions will find themselves “scratching their heads” and forced to buy expensive power from their more industrial-friendly neighbors.
For investors, that skepticism may align more with the second limiting factor: the skilled trades. Perry noted that having people “trained up” in HVAC, electrical, and carpentry is just as vital as the energy itself.
The labor shortage is a quantifiable threat. Currently, the US faces a deficit of nearly 400,000 workers in skilled trades.
So, even if the capital for data centers is available, the labor and hardware are not. Supply chain issues have already forced companies like Perry’s venture, Fermi America, to “break the line” to secure turbines from Siemens before competitors can grab them, effectively leaving latecomers competing for limited resources.
These supply chain issues present a major, tangible, and often overlooked challenge.
The physical reality of AI directly contradicts the prevailing Silicon Valley narrative of digital-first, asset-light disruption. The AI boom is, at its core, an infrastructure revolution.
For Wall Street, the true cost of AI may be far higher, and the timeline slower, than current projections suggest, largely due to the friction of building real-world infrastructure.
