“Great Scott,” Doc Brown exclaimed in Back to the Future when he found out he needed 1.21 jigawatts (a word used 40 years ago when the film was made) of power to send Marty McFly’s DeLorean back to the future.

But folks, that’s nothing compared to the 10 gigawatts of new electricity that will be needed to power new Nvidia/OpenAI data centers with a recently announced $100 billion investment.

Here are some benchmarks CNBC provided to put 10 billion gigawatts in perspective:

• FOUR TIMES the output of Hoover Dam
• Approximately equal to all of New York City’s annual power usage
• Equal to 2½ big new nuclear power plants — or 15 large natural gas plants — or about 3,000 wind turbines

Keep in mind, this is just for the one newly announced data center plan, with estimates that artificial intelligence will drive more than $1 trillion of investment over the next several years.

On CNBC, Energy Secretary Chris Wright said “we need to add 100 gigawatts of new firm capacity in the next five years.”

That would multiply all the above comparisons tenfold. In a 2024 column, I wrote about expected future demand for electricity driven by technology companies — and this new announcement amplifies the discussion.

Meeting this kind of historic demand for new power will not be simple.

In a recent CNBC interview, Thomas Fanning, former executive chairman of Atlanta-based Southern Company, which owns several electric utilities, talked about the challenges of meeting this demand.

He indicated it will take a blend of resources, including nuclear, hydro, natural gas and even coal.

He talked about issues in extending lifespans for and bringing old nuclear plants back online, permitting and licensing for new power plants, infrastructure needs (like pipe and transmission lines), and even the challenges of getting enough construction workers and engineers.

Fanning noted problems in the past in getting enough electricians and pipefitters for power projects — and demand for 100 gigawatts of new power may well leave projects begging for skilled workers.

Bottom line: This will not be easy or fast.

And who will pay for all of this? Will the $100 billion Nvidia/OpenAI plan cover just the data centers or also the power plants? Will retail consumers have to bear some of the costs?

And in the event of some disruption (natural or manmade), how will emergency power be allocated? Traditionally, hospitals come before 7-Elevens, and gas stations before movie theaters.

But how will data centers that require 24/7 stable power fit in? Will your lights go out, but you’ll still be able to use ChatGPT?

AI is clearly dominating the news — and having an incredible impact on financial markets.  As I write this, Nvidia (NVDA) is up more than 50% year to date, more than 1,300% in five years, and an incredible 410,000% since its 1999 initial public offering.

OpenAI, even though not publicly traded (yet), has a current valuation reported at $500 billion.

Microsoft (MSFT) is up almost 20% over the past year, 158% over five years, and up more than 2,800% over 20 years.

Alphabet (GOOG) is up more than 54% in the last year, 246% for five years, and up more than 5,800% since its 2004 IPO.

There are equally impressive stories for other big tech names.

But who is going to pay for the new power plants needed to meet AI data center needs? Perhaps ironically, I decided to use Perplexity AI to get some answers — hoping your lights didn’t darken due to the power usage.

Here’s what it said:

• The costs of building new power plants and upgrading grid infrastructure to meet surging data center demand are falling on a mix of tech companies, utilities and, increasingly, ratepayers (the general public). Although data centers are sometimes contractually obligated to cover the direct costs of new substations or local distribution upgrades, the broader expenses tied to transmission expansion and new generation capacity are often passed on to all utility customers — leading to higher electricity bills for residents and small businesses in many regions.

• Utilities typically spread much of the cost for new transmission lines and large power projects across all local ratepayers, even when the investment is driven mainly by new data center demand. In some states, regulators are starting to push for new rules or special rate classes to make large users — like data centers — pay more of the incremental costs they create. Examples include long-term contracts, uniquely tailored tariffs, or legislation requiring tech companies to shoulder a bigger share of grid upgrades.

Keep in mind that utilities are heavily regulated and regulators may be reluctant to pass on these new extraordinary costs to local ratepayers.

But the utilities can’t afford to eat these costs given their relatively low margins. It appears they will be looking to directly negotiate with the tech companies, not only for building costs but also locking in longer-term commitments to buy the power.

One need only remember back to the massive build-out of fiber lines that wound up sitting “dark” for years when expected demand diminished. It took years for the fiber to be fully utilized, and someone lost a lot of money in the interim.

Finally, will staid utility stocks benefit from this new technology wave?

An easy way to play this would be a utility stock exchange-traded fund like State Street’s Utilities Select Sector SPDR Fund (XLU). This index ETF seeks to provide exposure primarily to electric utilities with top 10 holdings, including NextEra, Constellation, Southern, Duke and American.

Traditionally looked at as a conservative, dividend focused investment, perhaps AI will change this. Backward looking annualized returns for ETF have been about 13% for one year, 10.7% for five years, and more than 10% for 10 years.

Baltimore-based Constellation Energy recently agreed to work on restarting the Three Mile Island nuclear plant in Pennsylvania to generate almost 1 gigawatt for Microsoft data centers.

Locally, Southern California Edison is engaged in multiple facets of power deals for data centers, ranging from new renewable power purchase agreements, infrastructure upgrades, site selection and managing utility impacts — for both current operations and planned growth in response to the AI and digital infrastructure boom.

So, stay tuned as we watch what has been argued to be the world’s most dramatic technological transformation in history.