Building the Rails Casualties
If you were alive in the mid-1800s and overheard investment chatter in a saloon somewhere between St. Louis and Sacramento, you’d hear the same thing you hear today on CNBC, TikTok, or a Reddit forum:
“This new thing is going to change the world — AND make us rich.”
Back then, the “new thing” was railroads.
Today, it’s artificial intelligence.
And like every era-defining technological wave before it, fortunes may be made and lost, investors will chase stars that burn too bright, and the winners will look obvious only in hindsight. But one pattern holds across nearly two centuries of American cycles of innovation:
Historically, the infrastructure has tended to endure long after the hype cycles fade.
The rails outlive the railroad barons.
The grid outlives the utility dynasties.
The highways outlast Detroit’s first giants.
Fiber optics outlasted the dot-com flameouts.
The internet outlasted the dot-bomb era.
Blockchain rails may outlast countless coins.
AI may be following the same script.
The Railroad Age: The First Network Effect
America didn’t just build a railroad system — it survived building one. It was a frenzy. Bubbles formed and burst. Entire towns popped up and vanished. Speculation was as thick as mud.
By 1893, dozens of railroad companies went bankrupt. Yet today, the rails still run. The companies didn’t endure, but the infrastructure did, and it rewired the American economy. Goods moved from Boston to Chicago faster than a farmer could haul a wagon to the nearest town.
It wasn’t the railroad stocks that changed society.
It was the railroad system.
The Power Grid: Electricity Turned from Miracle to Utility
People talk about AI like it’s magic. Electricity was magic first.
In the early 1900s, hundreds of local electric companies sprouted in every major city. Then consolidation swept the nation, reducing the field to a handful of utilities. Booms followed busts, yet the infrastructure — wires, substations, the centralized grid — became the backbone of the 20th century.
Today’s euphoric AI phase feels similar.
We’re dazzled by the front-end tools, the chatbots, the apps.
But the data centers, the power generation, the transmission lines are what can remain decades from now.
The Highway System: The Roads Became the Product
When Eisenhower launched the Federal-Aid Highway Act in 1956, automakers cheered. The U.S. became a driving nation. But most early auto brands disappeared.
Buffett famously noted that if you could short the entire auto industry in 1900, you should have — since 2,000 automakers became 3, and even those struggled later.
Yet the highways became essential. The roads created suburbs, logistics, vacations, and fast food.
The next wave of AI infrastructure — vast data centers, upgraded grids, small modular reactors, advanced cooling, fiber backbones — is our generation’s interstate highway system.
Oil & Gas: Booms, Busts, and Permanence
Oil speculation made and destroyed fortunes through the 20th century. Wildcatters rose and fell. M&A (mergers and acquisitions) reshaped the landscape. But the pipeline networks, storage facilities, and refineries built during those chaotic decades became permanent global infrastructure.
AI is developing its own heavy infrastructure, especially around energy, because AI workloads consume staggering amounts of electricity.
A single large data center can draw as much power as a medium-sized city.
The companies building the backbone — utilities, grid specialists, turbine makers, transmission contractors, nuclear innovators — may prove more enduring than the application companies of today.
Fiber Optics and the Dot-Com Crash: Overbuilding Is a Feature
In the late ’90s, companies laid so much fiber-optic cable that Wall Street called it “oversupply.” Then the dot-com bubble burst, wiping out Pets.com and dozens of internet darlings.
But that “oversupply” of fiber ended up powering the modern internet.
Streaming, cloud computing, mobile apps, social media — none of it would exist without the buried glass from the ’90s.
So when investors today worry about “too many data centers,” they might be missing the plot.
In some historical cases, overbuilt infrastructure often becomes the foundation of the next economy.
The Internet: Bubble First, Utility Second
The early internet boom was hysteria.
Valuations defied gravity.
Startups raised millions for ideas that would make today’s pitch decks blush.
Then came the crash.
Yet today, the internet is a utility as indispensable as water or electricity.
The lasting value wasn’t from the dot-com hype stocks — it came from the underlying infrastructure: protocols, servers, cables, and cloud networks.
AI is tracing the same arc.
Enter LLMs and Agentic Software: The New Layer on Top
If the data centers and power grids are the highways of AI, then Large Language Models (LLMs) are the vehicles — and agentic software is the self-driving automation inside those vehicles.
LLMs like GPT, Claude, Llama, and Gemini represent the first mass-market software layer built on top of the AI infrastructure.
They are the operating system of the AI era:
- They run inside the data centers.
- They rely on chips, cooling, and fiber.
- They scale with compute the same way cars scale with roads.
And agentic AI — software that can take actions, coordinate tasks, analyze information, and work across applications — is the productivity layer that transforms that infrastructure into real economic output.
In other words:
The hardware boom builds the muscles.
LLMs and agents provide the nervous system.
This is where proponents see the potential for significant productivity gains: AI employees, automated research, synthetic coding, autonomous workflows, and always-on digital labor. But the point remains:
LLMs will evolve. Agents will improve.
But the infrastructure they run on — the data centers, the power systems, the transmission networks — is what will endure.
Crypto: The Coins Were a Fad — But the Rails Might Last
Crypto brought another familiar cycle:
thousands of tokens came and went, fortunes rose and evaporated. Yet blockchain rails — distributed ledgers, settlement layers, decentralized databases — keep improving in the background.
The coins were the fad.
The infrastructure might be the takeaway.
Sound familiar?
The Bottom Line: In Every Era, Infrastructure Prevails
History repeats:
- Most railroad stocks died.
- Most automakers died.
- Most dot-coms died.
- Most coins died.
But the railroads, highways, grids, fiber networks, and internet protocols changed the world.
Today’s AI frenzy is no different.
The hype will eventually cool.
The cast of top AI companies will rotate.
But the infrastructure — the power grid, data centers, chips, cooling, fiber, and the platforms that run LLMs and agentic software — is widely expected to play a significant role in shaping the global economy over time.
This pattern appears repeatedly throughout history.
And it’s happening again right now.
Evan R. Guido, Senior Wealth Advisor, is the Founder of Aksala Wealth Advisors LLC, a 2026 Forbes Best in State Wealth Advisor, a 2018 Forbes Top Next-Gen Advisors award recipient. Evan heads a team of financial strategists for clients who consider themselves the “Millionaire Next Door.” He can be reached at 941-500-5122 Aksala.com eguido@aksalawealth.com 6260 Lake Osprey Dr. Lakewood Ranch, FL 34240. Securities offered through Cetera Wealth Services, LLC member FINRA/SIPC. Advisory Services offered through Cetera Investment Advisers LLC, a registered investment adviser. Cetera is under separate ownership from any other named entity. The views and opinions presented in this article are those of Evan R. Guido and not of Cetera or its subsidiaries. These opinions are based on Evan’s observations and research and are not intended to predict or depict performance of any investment. These views are subject to change based on subsequent developments. Information is based on sources believed to be reliable; however, their accuracy or completeness cannot be guaranteed. These views should not be construed as a recommendation to buy or sell any securities and purely for education and entertainment. Past performance does not guarantee future results. The Top Next Gen list includes 250 rising advisors who help manage over $490 billion in client assets. Each advisor was nominated by their firm, then vetted and ranked by SHOOK Research. The rankings, developed by SHOOK Research, are based on an algorithm of qualitative criterion, mostly gained through telephone and in-person due diligence interviews, and quantitative data. Those advisors who are considered have a minimum of four years' experience and the algorithm weighs factors like revenue trends, assets under management, compliance records, industry experience and those that encompass the highest standards of best practices. The Forbes ranking of Best-In-State Wealth Advisors, developed by SHOOK Research, is based on an algorithm of qualitative data, rating thousands of wealth advisors with a minimum of seven years' experience and weighing factors like revenue trends, assets under management, compliance records, industry experience, and best practices learned through telephone and in-person interviews. Portfolio performance is not a criteria due to varying client objectives and lack of audited data. Neither Forbes nor SHOOK receive a fee in exchange for rankings. Listings in these publications and/or awards are not guarantees of future investment success. These recognitions should not be construed as endorsements of the advisor by any clients. No compensation was provided directly or indirectly by the recipient for participation or in connection with obtaining or using these third-party ratings or award.