Power(ful) Bills
The case for homeowners (and renters…) to care about subsea data centers has to start with the most important area in most, if not all, people’s lives: economics. Money profoundly influences nearly every aspect of a person's life, shaping fundamental choices about housing, food, clothing, leisure, as well as affecting mood and deeply impacting psychological well-being, relationships, and opportunities. AI is about to drive up our power bills and that’s often between 2%-15% of a person’s salary (variables are salary and electricity prices by state). Said differently, AI data centers are about to impact your household's bottom line. Your power bill is about to rise.
Famously, Warren Buffet on his daily stop and McDonald’s either orders two sausage patties OR a sausage, egg, and cheese McMuffin. His choice is influenced by the stock market's performance; if the market is down, he opts for the more economical two-sausage patty order, while a market up sees him splurge. Although, he is likely not worried about money in the more pedestrian way most of the rest of us are, there is some likeness in that we all make a multitude of choices based on what we are dealing with in a financial sense each day.
But, let’s move to assuming Warren Buffet is not reading this and isn’t too fussed about his power bill hike at the moment. What for the rest of us? Why should we care about how much electricity AI uses relative to our own bills in anything more than an abstract or fleeting manner? Well, electricity is a uniquely local commodity: though it moves across long distances, pricing is set by regional supply and demand, and is shaped by the physical limits of the grid. When a new industrial-scale consumer, like an AI data center, arrives in your grid-neighborhood, it gets supplied continuously, often at hundreds of megawatts (and that trendline is trending UP!).
How This Affects You
Your electricity bill is made up of two parts: the cost of energy generation and the cost of delivery. The first reflects what it takes to produce or procure each kilowatt-hour (kWh) of electricity. The second covers the wires, transformers, maintenance, and infrastructure needed to bring that power to your house. Utilities must recover both, and they adjust rates based on local demand, not global averages.
Your electricity bill charges you based on your usage in kilowatt-hours (kWh), which is a unit of energy representing 1,000 watts of power used for one hour. Energy companies measure the total amount of electricity consumed by a household over a billing period in kWh and then multiply that by the per-kWh rate to determine the total cost on your bill.
Now, enter an AI data center, which is an enormous electricity consumer and, as stated already, it often draws hundreds of megawatts continuously. Unlike homes, which have variable use (we don’t use much power when we are at work, etc.), these facilities run at near-full or flat-out-full capacity 24 hours a day. Their arrival changes the local grid’s load profile DRAMATICALLY. What was once a modestly balanced system must now deliver massive and constant demand. And because electricity cannot be stored cheaply (or at all) at large scale, the grid must always meet demand in real time. When a large new consumer arrives, the utility has three options:
- Increase local generation: this might mean firing up older, costlier plants or buying power on the spot market, both more expensive than baseline supply.
- Upgrade infrastructure: they might need to upgrade substations, transmission lines, transformers to handle higher loads. Those capital costs are shared across all customers in the area, regardless of whether they benefit directly.
- Import power from farther away: this option incurs transmission losses and congestion charges that again find their way into end-user rates (attenuation sucks).
The price per kWh rises because the utility’s total system costs have increased.
In other words, the cost structure of the grid is socialized, while the demand shock from a data center is very much localized. A small town or region that once required, say, 200 megawatts may now require 400. Even if households still consume the same kilowatt-hours per month, they now draw from a grid operating closer to its limits, one that must pay more for marginal electricity. So for individual consumers, their bill still depends on their consumption, but the price per kWh rises because the utility’s total system costs have increased. In energy economics, this is called the merit order effect reversed: high constant demand brings more expensive generation into play more often, setting a new, higher marginal price.
The merit order effect describes the mechanism low-cost sources are dispatched first in the merit order, which prioritizes power plants based on their increasing marginal cost of production to meet demand.
In reverse, the merit order curve becomes steeper and shorter, reducing the available capacity and leading to much higher electricity prices - which we pay, even if we aren't using more electricity.
Subsea or Far Away?
Let’s assume you have read and now agree with the above – you accept that power bills rise due to data centers being spun up in municipalities, etc. I can also probably assume you’re pretty ticked off at the thought of it. Does this mean that we should put data centers subsea or should we just move them further away from neighborhoods?
The Case for Further Away
There really isn’t a case to be made on this. Moving data centers farther away from population centers or coastal grids creates new inefficiencies that ultimately hurt the environment and the people who rely on their output (notable: all of us rely on their output). Electricity and data share a key trait: both lose efficiency with distance. When compute facilities are sited hundreds of kilometers from the users or the power sources meant to feed them, energy losses in transmission rise, latency increases, and operational costs climb and so do the environmental costs. The very thing data centers exist to provide — fast, reliable access to data and AI — becomes slower, more expensive, and much more carbon-intensive.
Moreover, the cost of building long-distance power lines and fiber routes is enormous, and utilities recover those costs through higher regional rates and, likely, taxes. Meanwhile, the data itself (AI responses, streamed content, cloud services) must travel farther to reach users, increasing network congestion and reducing performance. In essence, pushing data centers inland or more remote breaks the chain of efficiency. It isolates compute from cooling power, energy from demand, and infrastructure from people.
The Case for Subsea Data Centers
By contrast, subsea data centers relocate this heavy demand away from congested terrestrial grids. Placing compute capacity offshore means drawing the power from sources that don’t strain local residential systems. In other words, the neighborhood substation isn’t competing with a 50..100..200-MW AI cluster anymore. For the homeowner, that separation helps stabilize rates and reduce the risk of future surcharges or peak-hour penalties.
But the economic argument goes beyond lower utility bills. Subsea infrastructure can also stimulate local economies more intelligently, generating new revenue through port activity, maintenance contracts, and renewable energy integration without burdening existing grids. It shifts the cost curve from reactive (expanding inland capacity) to proactive (building efficient, low-impact offshore systems).
Case(s) Closed
So whilst AI has other important impacts on the individual (positive and negative examples abound, nothing hits the consumer harder that a hike in economics. The last U.S election, at least in part, rested on the price of eggs, lest we forget), the most important is rooted in the economics. Whilst our product cuts the costs for the businesses (the tenants that use them) through both OPEX and CAPEX, it also cuts the costs for the end users in ways they can’t begin to think about. The industry, as well as our niche part of it, is still esoteric after all. But the bottom line is that electricity prices rise for ordinary households when an AI data center moves into their region, and not because power companies are suddenly greedy, but because of how the electrical grid and the economics behind it actually work.
Ultimately, electricity is not conjured from the air. When a data center consumes as much as a small city, the system must expand, and expansion costs energy, materials, and capital. Someone must pay — and in regulated systems, that someone is everyone.
