Byte BungalowByte Bungalow

← Stories

The Boring Way to Bet on the AI Data-Center Boom (Why Pros Skip the Reactor Stocks)

Updated November 14, 2026

Someone in a stock-video comment section asked a simple question: should I buy Oklo or NuScale? The reply underneath didn't pick either one. It basically said don't, and pointed at the least exciting corner of the whole market. So I went down the rabbit hole to figure out why the people who've clearly done this before keep steering beginners toward the boring stuff.

The reactor stock everyone's chasing hasn't built a reactor yet

Walk into any video about powering AI and the comments fill up with the same few names. Oklo. NuScale, which trades under the ticker SMR for "small modular reactor." People asking "why is Oklo up more than NuScale, can anyone explain?" and "shall we buy Plug Power now?" One commenter called SMR "an absolute bargain" that'll "fly next year."

Then you scroll a little further and the mood changes. "OKLO has yet to actually produce an SMR, last I heard." "Nobody wants a nuclear reactor in their backyard. You won't be able to sell it." One person just shrugged: too many actually-profitable companies to pick from to bother with stocks that are all future.

That's the thing I kept bumping into. These flashy names are bets on a machine that mostly doesn't exist yet. The demand is real — the reactor on the ground is not. And the longer I read, the more the quiet voices were all pointing the same direction, away from the exciting stuff.

The boring version of this bet has a guaranteed return written into law

Here's what finally made it click for me. A regulated electric utility doesn't really make money by selling you more electricity. It makes money by building things and being allowed to earn a set return on what it builds — the power plants, the transmission lines, the substations. That pile of stuff is called the rate base, and a state commission decides what the utility gets to earn on it.

And the number is better than it sounds. According to Energy and Policy Institute, commissions typically let utilities earn somewhere around 9 to 10.5% on the equity they put into that rate base — while the actual market cost of that money, for a big stable utility, runs closer to 6 or 7%. Picture a landlord who's legally allowed to charge rent on every brick he lays, at a rate set a little above what the bricks cost him to finance. Now hand that landlord the biggest construction boom in a generation.

That's the whole pitch. American Electric Power has told investors it expects more than 50% load growth from data-center customers over the next five years — a staggering number for a company usually described as sleepy. S&P Global Market Intelligence put it plainly: most analysts never expected regulated utilities to end up the main winners of the AI boom. The startups have to invent something. The utility just has to build, which is the one thing it already knows how to do.

The part the bulls skip: you might be the one funding it

So I went looking for the catch, because there's always a catch. And it's this — the same rate base that makes a utility a tidy bet also means the cost of all that new building gets spread across everybody on the system. Including the person reading their own power bill at the kitchen table.

One analysis cited by Forbes pegs the amount already added to consumer electric bills from data-center demand at roughly $23 billion. And this is early. Data centers ate about 4.4% of all U.S. electricity in 2023, and a Department of Energy report built on Lawrence Berkeley National Laboratory research projects that could climb to as much as 12% by 2028. That's the turn that made the whole topic stop feeling like a spreadsheet for me. The person on the other side of this clean investment thesis is a regular customer who never asked for a data center down the road.

It's also the risk to the trade itself. Lawmakers in Texas have thrown up roadblocks on data-center projects, and Virginia created a whole new rate class for giant customers so the costs stop landing on households. The boring bet isn't risk-free. It's just that the risk is politics and public anger, not whether a company can finally ship a reactor it's been promising for years.

Where this rabbit hole actually goes

What I can't tell you is which utility to buy. That's not the point of poking at this, and it's not something I'd pretend to know from reading comment sections for an afternoon. The pattern is what stuck with me: the quiet names with signed contracts keep getting talked up by the people who've been around, and the loud names with press releases keep getting talked down.

But every answer here opened another door. Where does this power actually come from — gas turbines, nuclear, the grid itself? Why can't we just build the lines fast enough? Who's fighting over the water these things drink? That's the whole six-degrees-from-AI rabbit hole, and it's exactly what we chase on Byte Bungalow. If you want to follow the power all the way back to where it's made, watch the video and subscribe on YouTube — that's where the rest of this story lives.

Common questions

Why is Oklo's stock up more than NuScale's?
Short-term price moves between two pre-revenue companies are mostly sentiment and news flow, not fundamentals — one gets included in more ETFs, one lands a government headline, and the stock pops. As commenters on these videos point out, neither has actually produced a working small modular reactor at scale yet, so you're comparing two promises, not two track records. That's the core reason experienced investors treat both as speculation rather than a durable bet.
Should I just buy a nuclear startup like NuScale or Oklo to play the AI power boom?
You can, but understand what you're buying: a bet that a company will build and sell a product it hasn't finished building yet. The demand for power is very real. The reactors are not on the ground. That gap is why many longtime investors in the comments steer newcomers toward regulated utilities instead — companies already earning money building the grid today, rather than hoping to build something new tomorrow.
Do AI data centers actually raise my own electric bill?
Often, yes — indirectly. When a utility builds new power plants and transmission lines to serve a giant data center, those costs usually go into the shared rate base that every customer pays into, unless the state forces the data center to cover its own costs. One analysis cited by Forbes estimates about $23 billion has already been added to consumer bills from data-center demand. Some states, like Virginia, are now creating special rate classes to push those costs back onto the big users.
What's a 'regulated utility' and why does it matter for this bet?
It's a power company granted a legal monopoly over a territory in exchange for letting a state commission set its rates. The utility earns a regulated return — commonly around 9 to 10.5% on its equity — on the value of everything it builds. More building means more earnings. The data-center boom hands these companies the biggest building spree in decades, which is why analysts now call them surprise winners of the AI era.
Is it too late to invest in the data-center power story?
Nobody can promise you a return, and this isn't advice on a specific stock. But the demand curve is still early — data centers were roughly 4.4% of U.S. electricity in 2023 and could reach as much as 12% by 2028 per Department of Energy projections. The bigger near-term risk isn't that the demand fades; it's political backlash over who pays for it, which is already showing up in states like Texas and Virginia.

Keep reading

Sources

  • U.S. Department of Energy / Lawrence Berkeley National Laboratory — Data centers consumed about 4.4% of total U.S. electricity in 2023 and are projected to consume as much as 12% by 2028.
  • Energy and Policy Institute — Regulators commonly allow utilities a return on equity around 9% to 10.5%, while the actual market cost of equity for a large stable utility runs closer to 6% to 7%.
  • Forbes (Jason Kirsch) — One analysis puts the amount already added to consumer electricity bills as a result of data center demand at roughly $23 billion.
  • S&P Global Market Intelligence — Most analysts never anticipated that regulated utilities would become the main beneficiaries of the AI boom; the U.S. needs 230+ GW of new capacity over five years but utilities plan to add only ~93 GW.
  • Simply Wall St News — American Electric Power has guided to 50%+ load growth from data center customers over the next 5 years.

Byte Bungalow

I do the homework out loud.

If this is the kind of thing you want checked against the documents instead of the hype, the video's above — and there's a new one every week.

Subscribe on YouTube →

By Byte Bungalow. Home power and home tech, checked against the documents instead of the hype. Independent commentary; not affiliated with any manufacturer, utility, or builder named here. Not professional electrical advice.