Snacks and Soil

My alma mater, Williams College, encouraged its students to immerse themselves in a wide range of classes. Though I majored in history, I took a diverse array of economics, statistics, and sciences to receive a well-rounded liberal arts education. One notable semester, however, I took a leap of faith and signed up for theater. I started the course expecting to learn the basics of being on stage and what it takes to become an eloquent and commanding speaker in front of crowds. Instead, I became a disciple of the comedy duo of Vince Vaughn and Owen Wilson. My professor assigned us scenes from an assortment of their movies to rehearse and perfect, encouraging us to study their movie catalogue. This seemingly tangential interlude becomes relevant to our wealth management blog, not only because any media the duo creates are a great watch, but even more importantly, their film, The Internship, helps visualize ideas regarding the AI revolution we’re in.

The film centers around Wilson and Vaughn, two 40-year-old professionals who, despite the odds against them, land an internship at Google. After arriving on campus, they find themselves stunned. Free food everywhere. Nap pods. A slide in place of a staircase. Employees playing sand volleyball between meetings. The campus looks more similar to an all-inclusive than the headquarters of one of the most valuable companies on earth. Unfortunately for tech interns today, tech companies have started to ponder the possibility of fiscal responsibility, leaving headquarters rather dissimilar to a vacation resort.

Halfway across the country outside Abilene, Texas, a farmer recently sold the worst land on his property for $100k an acre to the interns’ tech firm. The company cared nothing about yield or the quality of the soil. They only cared that it sat near a power substation and had access to water. They scrapped the employees’ free lunches on Wednesdays to pay for the land, leaving the farmer flush with cash no doubt funding a future trip to Cabo.

I say this to illustrate the incredible shift in cash flow AI is creating. Roughly six trillion dollars is projected to be spent on the buildout by 2030. That is already bigger than the interstate highway system and the Apollo program combined. The once cash flush Silicon Valley is now cash strapped, and the ‘worthless’ ground in middle America has now become a metaphorical gold mine, fetching prices previously unfathomable.

As another revered character from my theatre class, Michael Scott, famously said, “Well, well, well, how the turntables.”

What Is AI

At its most basic, artificial intelligence is the simulation of the human mind using math. Simple in idea, complicated in practice.

Every AI system, however sophisticated, is built from three moving parts. First, data: examples of the subject you want the system to understand. Second, an algorithm: a set of rules, refined through those examples or training data, for weighing what matters and what does not. Third, prediction: the system's best guess about something it has never seen before, based on the patterns it found in everything it has seen.

Your phone's face unlock works exactly this way. The data is a map of your face. The algorithm is a kind of elaborate if-then statement that weighs the different variables of your face. The prediction is the single yes-or-no your phone delivers to determine if this face is the one that unlocks the phone or not. That, in a simplified example, is how AI works.

The systems behind the AI boom run on the same three parts, only the three parts described above become vastly bigger and more complicated. A model trained to read protein structures, draft legal contracts, or recommend the next show to watch is doing the same basic thing your phone does at the lock screen: finding and recording patterns in what came before, and using them to predict what comes next.

That is the software. None of it runs without the data center, and that is where the snacks (or lack thereof) and the soil come back into the picture.

Most of us are well aware of the importance of chips in AI. Chipmakers like Nvidia and AMD make the processors that everything else depends on, and demand for these products has been extraordinary. But these companies are not the only benefactors. Caterpillar, a company most people associate with bulldozers and mining equipment, now generates its fastest growth from a different business entirely: selling the generators and gas turbines that supply backup and primary power to data centers. That single segment recently became Caterpillar's largest, and its order backlog for power equipment is booked years into the future. Constellation Energy, the largest nuclear operator in the country, has spent the past two years signing multi-decade contracts with the same tech companies. Now both of these once “boring” stocks are trading with the volatility of a tech startup.

The amount of money being spent on AI ripples through the economy. While the companies making the chips and creating the models are visible, the companies that move the soil and provide the energy are just as important.

We've Been Here Before

AI is brand new and can feel intimidating, but history rhymes and the best way for us to learn about AI, is to take ourselves back a couple hundred years to the Industrial Revolution. Where the AI Revolution seeks to simulate the human mind with math, the Industrial Revolution simulated the human body with mechanical force.

Let’s consider the American workforce. Most folks in the United States can trace their ancestry back to a farm. In fact, in 1900, roughly four in every ten working Americans farmed for a living. At that time, a single field, plowed by hand or by a team of horses, might require ten people to bring in a harvest that one man and a tractor could manage alone within a generation. By the year 2000, fewer than two workers in a hundred still farmed. The work did not disappear because food stopped being important. Rather, it decreased because the steam tractor, and the gasoline engines that followed it, made nine of those ten workers unnecessary.

For the people living through that transition, the change did not feel like progress. It felt like their very livelihood being ripped from beneath their feet. In the textile mills of England, a century earlier, skilled weavers and knitters watched as mechanized looms produced the equivalent of a week of their labor in a single afternoon. They were not ignorant. They were tradesmen watching their trade, which they had built their lives around, become worthless almost overnight. A number of them, who came to be called Luddites, broke into the mills at night to smash the machines. Their attempts to restore their livelihoods were unsuccessful. The mills kept running, the machines kept improving, and within a few decades, an entire way of life simply stopped existing.

What Does It Mean?

Ask a serf working in a field in 1820 what he thinks of the machines arriving in the cities. He will tell you, correctly, that they have made his life harder, his work less valuable. His trade is disappearing, his wages are falling, and nobody has explained to him what he is supposed to do instead.

Now ask that same man about the airline industry. He will have no idea what you are talking about, because it does not exist yet and will not for another century. He cannot picture an industry that today employs nearly sixty million people worldwide built on top of a technology nobody in his lifetime had even begun to imagine. That is not a flaw in his thinking. It is simply impossible to understand the second half of a revolution while inside the first half of it.

That is roughly where artificial intelligence stands today. The disruption is visible, yet the new industries it might create are not. And while we can pontificate and point to the charts showing GDP per capita growth from the industrial revolution, it is impossible to say for certain that the AI revolution will do the same thing. Though the possibility it might is undeniably exciting.