Laying Tracks to Nowhere? What the 1893 Railroad Panic Teaches Us About AI

SAS Financial Advisors LLC | May 22, 2026

So much uncertainty surrounds the next 6 months to a year, let alone a year or more from now, that all possibilities remain open. Positive, negative, and everything in between.

The question is not just whether AI changes the economy. It already is. The better question is whether the constant controversy, focus, and discussion around AI is causing us to forget everything else happening at the same time: inflation, affordability, the war with Iran, and even the continued discussion around the Epstein files and institutional trust.

As we discussed in Headlines, Trust & Markets: Does It Change the Plan? , public confidence matters. Whenever headlines center on powerful institutions, secrecy, or accountability, it chips away at trust. Trust is foundational to financial markets.

A Little History: Railroads and AI

In the 19th and early 20th centuries, railroads were the focus of massive investment. In some years, railroad investment equaled a much larger share of the economy than today’s nearly $1 trillion investment in AI infrastructure and development.

This comparison provides perspective on technological change. Railroads did not simply create a new industry. They rewrote the operating math of the entire economy. AI may be doing something similar.

The comparison between the 19th century railroad boom and the current AI buildout is a favorite among economists because both represent “General Purpose Technologies.” These are innovations that do not just create a new industry, but fundamentally change how other industries operate.

As we discussed in Staggering Sums: Record Highs and the Rule of Unintended Consequences , large investment booms can produce both opportunity and unintended consequences.

AI vs. Railroads: The Numbers

Investment as a percentage of GDP is one way to measure how “all in” a society is on a new technology.

Metric	Railroads, Peak 1850s to 1880s	AI Infrastructure, 2025 to 2026
Annual capex as percentage of GDP	Roughly 3% to 6%	Roughly 1.2% to 1.5%
Total capital intensity	Consumed 15% to 20% of all U.S. capital	Rapidly growing, with tech capex sharply higher
Annual dollar spend	Roughly $550 billion, adjusted to 2026 dollars	Roughly $650 billion to $760 billion, Big Tech only
GDP impact	Contributed roughly 1.3% to annual growth	Estimated near 1.1% to 1.3% of GDP growth in 2025

The Deep Wallet Paradox

In raw dollars, we are spending more on AI chips and data centers today than we ever spent on tracks and locomotives. However, the U.S. economy in 2026 is so massive that even a $700 billion annual spend by Amazon, Microsoft, Meta, and Google only accounts for about 1.2% to 1.5% of GDP.

In contrast, the railroad era was pure capital intensity. In the 1850s, railroad investment often exceeded 15% of all net capital formation in the country.

We are not quite there yet with AI, but the trajectory is steep.

Speed of Buildout: A Century vs. a Decade

The most striking difference is not only the amount of money. It is the velocity.

Railroads took roughly 40 years, from 1850 to 1890, to reach something close to national saturation.

AI is attempting a similar level of infrastructure saturation, compute everywhere for everyone, in about 10 years.

Goldman Sachs and JPMorgan estimates from early 2026 suggest cumulative AI investment could reach trillions of dollars by 2031. If that holds, the AI boom may surpass the railroad boom as one of the largest industrial capital expenditure events in history.

Productive Investment or Speculative Excess?

Every boom has its “ghost tracks,” meaning infrastructure built for demand that never fully arrives.

In the railroad era, dozens of redundant lines were built. The Panic of 1893 followed, and roughly a quarter of railroads went bankrupt. But the tracks remained in the ground. The economy ultimately benefited from cheaper transportation for decades.

The AI version of this risk may be “ghost clusters,” massive data centers filled with GPUs that become obsolete or underused if the return on investment from AI software does not materialize.

The historical lesson is important. Even if railroad investors lost their shirts, the American economy won because the infrastructure was subsidized by that lost private capital. We may be seeing something similar today. Big Tech may be subsidizing a global “intelligence grid” that benefits the economy for decades, even if current stock valuations eventually face a railroad-style correction.

What About Jobs?

As far as the impact of AI on jobs, there are clear indications in 2026 that certain entry-level jobs are being negatively impacted by AI. At the same time, technological innovations have historically increased job opportunities, although often in new fields of work that were not visible at the beginning of the transition.

Importantly, even with the beginning of AI job replacement and job changes, adaptation is already beginning.

Bottom line in regard to employment: the current unemployment rate is around 4.3%. Large free market economies have proved to be very adaptable to technological change. I have learned during my career that anytime you hear “this time it is different,” it becomes a red flag. It usually means it is probably not different this time either.

AI may change the type of work people do. It may reduce some jobs. It may create others. It may do all of the above at once.

Inflation Has Not Gone Away

While concern about AI’s impact on the economy gains constant attention, inflation numbers this week are causing concern in the stock market as price changes caused by the war in the Middle East begin to filter through to the consumer.

Both the retail CPI and the Producer Price Index, which measures wholesale prices, are impacting interest rates. As we have discussed in Conflict, Shocks, and Continued Uncertainty: How to Pay for it All? , energy prices and global instability feed directly into consumer costs.

This is where the AI story can become a distraction. The economy is not only dealing with AI. It is dealing with inflation, affordability, war, interest rates, consumer debt, and market expectations all at once.

Markets Are Still Looking Ahead

The stock market, however, has so far been more impacted by the acronyms we discussed last week in TINA, FOMO, and TACO: The 3 Acronyms Defining the 2026 Market Disconnect , as well as outstanding earnings.

S&P 500 first quarter earnings have increased at a remarkable rate. It seems unfathomable how this rate of earnings growth can be sustained.

Could this mean the stock market is looking 6 months ahead, as has always been assumed?

Maybe.

But it also raises several uncomfortable questions.

How long can the top 10% of Americans, who account for a very large share of U.S. retail spending, keep going?

What are the risks of recession?

Could recession be the “black swan” awaiting us?

How long will this war last?

Currently, it is hard to see any resolution.

As we discussed in The Great Divorce: Why Wall Street Doesn’t Care About Your Gas Bill , Wall Street and Main Street are not always telling the same story. Stock prices can climb while affordability gets worse for the average household.

Apocalypse or Renaissance?

The railroad did not end the job of the teamster overnight, but it did make the horse-and-wagon business model geographically irrelevant.

Similarly, AI investment at roughly 1.3% of GDP is currently high enough to “lay the tracks” for a new economy, but not yet high enough to suggest the total replacement of human labor.

AI may be an apocalypse for some jobs. It may be a renaissance for others. More likely, it will be both.

The practical answer is to avoid the extreme headlines. The world is changing, but free market economies have adapted to technological change before. That does not mean there will be no pain. It means the story is rarely as simple as the headline.

The question remains:

Is the current $700 billion plus annual AI spend a rational buildout of a new utility, or are we building tracks to nowhere?

Source Note

By the way, the links in this newsletter are generated by AI with the query at the top of the link.

This website is informational only and does not constitute investment advice or a solicitation. Investments and investment strategies recommended in this blog may not be suitable for all investors. SAS Financial Advisors, LLC and its members may hold positions in the securities mentioned within this newsletter. SAS Financial Advisors, LLC is not responsible for any third-party content referenced.

The SAS Newsletters are posted weekly on the SAS Blog: https://www.sasadvisors.com/blog