In 1848, a 29-year-old Sacramento shop owner named Samuel Brannan was tending the cash register at his store when a pair of shoppers asked if they could pay him in gold.
Brannan was stunned when the shoppers pulled out solid gold nuggets. The gold, they said, had been found at Sutter’s Mill, about 35 miles northeast of Sacramento.
In addition to owning the only shop between San Francisco and the gold site at Sutter’s Mill, Brannan also happened to own a newspaper. And so he immediately used his paper to start spreading the word about the gold discovery.
According to legend, Brannan supposedly kept a small vial of gold to show prospective miners, and he once allegedly ran through the streets of San Francisco shouting, “Gold! Gold from the American River!”
Brannan’s newspaper constantly ran stories claiming that all you needed was a shovel, a pick, and a pan, and you too could get rich digging gold in California.
His hyping of the gold discovery is rumored to have helped spark the rush of 300,000 people to California to dig for gold— many in 1849, giving rise to the moniker “the San Francisco 49ers.”
But Samuel Brannan was NOT among them digging for gold.
Instead, during the peak of the Gold Rush, he sold about $5,000 worth of shovels, picks, pans, and other equipment to miners each DAY at his store on the way to the Mother Lode.
That would be nearly $200,000 per day in 2024 dollars. And Brannan is cited as California’s first millionaire.
“Picks and shovels,” has become a common descriptor in investments— sometimes the best opportunity when a new “gold mine” is found is to sell the “picks and shovels” needed to mine the gold.
Right now, the biggest new gold mine out there for investors is artificial intelligence.
And Nvidia is almost single-handedly responsible for producing the hardware (GPUs) for many of the most popular AI applications like ChatGPT.
But in addition to the hardware required to run AI, there is also an ongoing energy need. And it’s not trivial.
In fact, for every dollar companies spend on a GPU (which can easily run $40,000+ each), they will have to spend another dollar for the energy to run it.
AI applications like ChatGPT can generally be broken down into two stages: training, and querying. The training phase is where they feed vast amounts of data into the algorithm so that it can ‘learn’.
And this training phase is absurdly power-hungry; training a single algorithm can be the equivalent electricity of tens of thousands of homes. It’s like a small city.
Then there’s the ongoing querying phase, i.e. the power consumed when users say “Generate an image of a cat playing the harmonica”. Millions upon millions of queries in real time require substantial energy, about 10x as much as Google uses for its entire search function.
And that figure is growing.
The problem for America’s power infrastructure is that it is already strained.
With the government pushing regulations to force people to buy electric vehicles, it will only overload the grid that much more.
And exacerbating the problem is that the people in charge now keep pushing for extremely inefficient and intermittent “green energy” like wind and solar, which simply cannot produce electricity in the same quantities as conventional sources.
In short, America’s power grid is already struggling. And AI will continue to create a surge in electricity demand that utility companies may not be able to meet.
Big technology companies know this, which is why many of them are looking at building out their own power plants… just to ensure that their data centers never run out of electricity.
The best solution by far would be small scale nuclear reactors. But that technology just isn’t quite ready yet.
So the next best reliable and inexpensive way to produce electricity is through natural gas.
US natural gas is priced at around $2.50 per million BTUs. If you look at this in terms of pure energy, $2.50 natural gas is the energy equivalent of a barrel of oil selling for about $15.
In other words, US natural gas is absurdly cheap, probably one of the most underpriced commodities in the world.
Tech companies and electrical utilities know this. So there’s a good chance that natural gas becomes the favored energy commodity for new power plants that will continue to drive the AI boom.
That’s why I believe natural gas is currently the best “picks and shovels” investment to scoop up in the AI space: energy is the one thing that these power-hungry AI applications need. And natural gas is the cheapest form of energy in the world.
It also helps that the stock prices of many of the highest quality and most profitable natural gas producers are laughably cheap. So there’s clearly a lot of upside potential from here.