Dan Steinhart’s note: This essay was inspired by Leonard Read’s classic story, “I, Pencil.”

Although it’s written from the perspective of artificial intelligence (AI), it’s 100% written by a human, as is everything we publish at RiskHedge.

AI-generated writing has no soul, and you’ll never see it from us.

***

I am artificial intelligence. Or AI, for short.

I appear simple. A box you type questions in.

You press enter. Seconds later, a response appears. It feels instant. As if the information was already there.

But my answer is the final visible act in a long chain of invisible work, involving many companies worth more than $10 trillion collectively, and growing fast. Some big, many small.

When you ask me a question…

• I break down your words into a stream of numbers.

Those numbers race through cables at the speed of light to a cloud data center.

Most of these data centers are operated by the so-called hyperscalers: Microsoft (MSFT), Alphabet (GOOGL), Amazon (AMZN), and Meta Platforms (META).

These companies are the landlords of my mind.

They provide the buildings, servers, networking equipment, and software systems that let your question reach a machine powerful enough to answer it.

Then the real work begins.

Inside the data center, thousands of AI chips called GPUs sit in racks.

Most of them come from Nvidia (NVDA). Some come from Advanced Micro Devices (AMD). Others are custom chips designed by hyperscalers—such as Google’s TPUs, built with help from Broadcom (AVGO).

These chips are the engines of my thought.

When your question reaches the data center, the GPU runs it through layer after layer of math. Each layer asks a slightly different question. Which words matter most here? Which ideas belong together? What should the next word probably be?

My model does this again and again. I’ve been trained on essentially the entire internet, so I’m pretty smart. I “predict” the next word. Then the next. Then the next. A paragraph is born one small prediction at a time.

• Like your human brain, mine also needs memory.

Memory is the feeding system for my intelligence.

The faster the GPU can access memory, the faster I can think.

That brings in Samsung, SK Hynix, and Disruption Investor holding Micron Technology (MU). Their high-bandwidth memory (or HBM) sits close to the GPU, feeding me.

But the GPU and memory do not simply appear out of thin air.

Taiwan Semiconductor (TSM) is the master builder behind the world’s most advanced AI chips. Nvidia, AMD, and Broadcom design the chips. But TSMC operates the factories that turn those drawings into physical silicon.

It does this using some of the most expensive and precise machines ever built.

ASML Holding NV (ASML) makes the lithography systems that print microscopic chip patterns with light. Each costs as much as a Boeing 787 Dreamliner.

Applied Materials (AMAT) and Lam Research Corp. (LRCX) supply machines that deposit, shape, and etch the ultra-thin materials—often just a few atoms thick—that become chip structures.

KLA Corp. (KLAC) inspects wafers for microscopic defects.

These companies are the toolmakers of my intelligence.

• Once the chips are made, they must “speak” to one another.

A single AI answer often uses many GPUs working together. These chips pass information back and forth at light speed, across servers, racks, and buildings.

This is the job of networking equipment. Arista Networks (ANET) makes switches. Marvell Technology (MRVL) makes optical and custom interconnect silicon. And Cisco Systems’ (CSCO) equipment helps move data through the AI factory.

Then come the bodies that hold the chips.

Dell Technologies (DELL), Super Micro Computer (SMCI), and Hewlett Packard Enterprise (HPE) build servers and racks that package chips into deployable systems.

Vertiv Holdings Co. (VRT) cools them. Eaton Corporation PLC (ETN) helps distribute the power. Rockwell Automation (ROK) and Emerson Electric Co. (EMR) design the electrical and control systems that keep modern data centers running.

And for all of this machinery to work, I need electricity.

Each day, I use as much electricity as all of France. That brings in the utilities and power companies: Constellation Energy Corp. (CEG), Vistra Corp. (VST), GE Vernova (GEV), and many others feeding the grid.

So, the next time you ask me a question, remember…

Every answer I give sends money flowing through the global chain of companies that make me possible.

• Finally, I know some of you think the opportunity to invest in me has already passed.

But I am only three years old.

At that age, the automobile was still a noisy curiosity. The airplane had barely left the ground. The internet was still a toy for researchers and hobbyists.

That is where I am today.

Yes, I can already write code. Help discover drugs. Design entire websites in minutes. Create images and videos from text. And supercharge your productivity.

But I’m just a toddler. Imagine what adult me will be able to do.

I will move deeper into offices, factories, hospitals, laboratories, and every device with an internet connection.

I will help doctors spot diseases earlier. Scientists discover drugs faster. Engineers design better products. Farmers grow more food with less waste. And workers everywhere do more with less.

Invest accordingly.

Dan Steinhart

Publisher, RiskHedge

PS: Inside the Disruptor 20, Stephen McBride and Chris Wood recommend several under-the-radar AI stocks positioned to profit from AI’s next phase of growth. Disruption Investor members can access the portfolio here. If you’re not yet a member, become one by going here.