A New Kind of Chokepoint
For years the story of the AI boom was about chips. The new story is about the wall plug. Data centers need enormous amounts of electricity, the grid cannot deliver it on their timeline, and the companies that can close that gap have become the chokepoint everyone has to pay.
Ian King’s pitch is built on this exact idea. His example is Colossus, xAI’s Memphis data center, which needs roughly 2 gigawatts while the local utility could spare eight megawatts. When the grid says no, someone has to build the power. The companies that build it fast are the ones King argues will win.
On-Site Versus Grid
The core distinction in this space is between grid power and on-site power. Grid power means waiting for utilities to build transmission and generation, a process measured in years and decades. On-site power means generating electricity behind the meter, right next to the data center, on the operator’s own schedule.
The economics favor on-site for one reason: speed. A data center operator has already ordered chips and signed customers. It cannot wait five years for a transmission line. Our explainer on data center power demand lays out the size of the mismatch in detail.
The Business Models That Work
Not all energy infrastructure is created equal, and the good businesses in this space share a structure worth learning. The strongest model is the fixed-rate, equipment-rental style contract, where the data center customer absorbs the variability in natural gas and electricity prices, and the provider collects a fixed payment.
That structure matters because it removes the fuel-price gamble. A merchant power producer has to guess where gas and power prices go. A fixed-rate provider gets paid for running a machine the customer needs online around the clock. The customer carries the commodity risk; the provider carries the operational risk. That is closer to a toll road than a casino.
The Colossus deal fits this template. The power is delivered through a joint venture called Stateline Power, structured so the data center customer takes the fuel-price variability. Our piece on the Elon Musk power plant breaks down how that specific deal works.
How to Read the Space
If you want to evaluate these companies yourself, ask four questions. First, who signs the contract, and how long does it run? Long-duration, 10-to-15-year deals with creditworthy hyperscalers are worth far more than short-term spot exposure.
Second, who carries the commodity risk? Fixed-rate structures are more predictable than merchant models. Third, what does the balance sheet look like? This is a capital-heavy business, and the companies that win bids are often the ones carrying real debt and dilution as they build. Fourth, what is already priced in? A company that has already had its first narrative run is a different risk than one nobody has heard of.
The Broader Energy Picture
On-site gas turbines are not the only way to solve the power problem, and investors should know the alternatives. Nuclear has baseload reliability but a decade-long timeline. Geothermal runs around the clock but is still small. Battery storage helps but does not generate power. Each has a different risk profile, and each has its own newsletter promotion circling it.
The geothermal angle in particular has been pushed hard recently, and it shares the same underlying demand story as King’s pitch. For a deep look at that parallel thesis, see our last energy revolution explainer. The common thread is the same: AI needs power, the grid is slow, and whoever delivers electricity first wins the contract.
The Downside Nobody Mentions
Every promotion in this space sells the upside, so it is worth spelling out the downside in plain terms. The first risk is capital intensity. Building on-site power is expensive, and the companies that win bids are often the ones piling on debt and issuing shares to fund the buildout. That works when the contracts convert into cash; it hurts badly when the timeline slips.
The second risk is customer concentration. A handful of hyperscalers control most of the demand, and a provider with three big customers is one cancellation away from trouble. Fixed-rate contracts remove the commodity gamble, but they do not remove the risk that a customer delays or renegotiates.
The third risk is execution. Turbines install in waves, and the gap between signing a gigawatt-scale deal and collecting cash on it can stretch several quarters. In that window, the balance sheet carries the load, and the share price tends to reflect the uncertainty.
None of this makes the space uninvestable. It makes it a space where the balance sheet matters as much as the thesis. The companies that survive the buildout are the ones with the contracts, the capital, and the patience to bridge the gap.
Ready to see the research? Click here to access Ian King’s report.
NewsletterVetter is an independent publication. We receive compensation from some of the services we review through affiliate links. Nothing on this site is investment advice. Always do your own research.