Introduction

In the summer of 1872, Andrew Carnegie stood brimming with excitement in a British steel mill, watching a process complete in minutes what previously took two weeks. With a technological leap so vast, Carnegie knew of the fortuitous implications this would have on the economy. Henry Bessemer had patented his converter sixteen years earlier — a pear-shaped vessel that blew cold air through molten pig iron, burning off the excess carbon and other impurities in a violent, luminous reaction that took roughly ten minutes from first blow to finished steel.

The result was a product drastically stronger and more consistent than anything that had come before it, produced at a fraction of the previous cost. Carnegie had seen the future, and the future was Bessemer steel. He immediately got on a boat back to Pittsburgh and started buying land.

Iron’s Inevitable Successor: Steel

The American railroad infrastructure in the 1850s and 1860s ran on iron rails. Hastily built by speculators who mainly sought a cash grab, routes were built with no thought of their lasting durability, using cheap materials that were essentially bars of iron welded together and rolled into a T-shape. This shape worked at a disadvantage, as weld lines would weaken and rails would become warped. This became known in the industry as "snakehead" rails: dangerous strips of iron that would fling upward when a train ran over them, sometimes impaling railroad cars and passengers.

By the 1860s the US railroads started experimenting with early steel rails, finding that a steel rail would last anywhere from twice to twenty times as long as a wrought iron rail. Steel rails could also carry heavier freight loads without deforming the tracks, proving to be less brittle with heavier loads which became commonplace after the Civil War.

The industry was convinced of its value and by 1873, the United States had approximately 50 iron rail mills. Every one of them was making a product the railroads already knew they wanted to replace, but the only question was when the price of steel would become cheap enough to substitute completely.

Carnegie was poised to answer that question, since his years at the Pennsylvania Railroad gave him direct exposure to the railroad industry's wants and needs. He understood the need for better materials, and how the scale of infrastructure spending was transforming America. Not only would he take advantage of this opportunity but do so in a way that wouldn’t withhold America’s rapid growth. He decided to put all his chips in and make a massive bet, putting his name and investment funds on the line.

Regulatory Tailwinds

Let’s step back a few years to 1870 and trace our steps to the first domino to fall. Ulysses S. Grant overtakes Andrew Johnson as President of the United States, and the Republicans have full control of the White House, the House, and the Senate. As part of the post-Civil War Reconstruction period, Grant signed the Tariff Act of 1870 into law, imposing a massive $28-per-ton duty on imported British steel rails. Considering steel sold for around $31 a ton, this amounted to an astronomical 90% tax rate.

Grant’s rationale was to hedge against a national security vulnerability; with intercontinental transit via rail being a massive catalyst in driving economic growth, this cause had to be advanced through domestic steel. By establishing this self-sufficiency, the US is protecting itself against any foreign blockades or diplomatic disputes that could place this growth at risk.

With regulation cutting off Britain's competitive advantage, American steel manufacturers found an open playing field to advance their industry forward by supplying the steel needed to rebuild the post-Civil War country, create stronger material, and improve their margins through cost savings. Carnegie saw the writing on the wall, and it is through this regulation that he entered the steel industry.

Carnegie followed unconventional advice that made him unconventionally wealthy: “Put all your eggs in one basket, and then watch that basket.” Carnegie’s “egg” was the Bessemer converter, and in January 1873, he began construction on what would become the Edgar Thomson Steel Works.

But fate would have other plans for his poor egg he began watching. By September of that same year, the banking house of Jay Cooke, the primary financier of railroad expansion in America, had collapsed under the weight of Northern Pacific Railroad bonds it could not sell, triggering the first great depression in United States history. The New York Stock Exchange closed for ten days while the banks thought through a plan to prevent total collapse of the economy. During that time, one-quarter of American railroad companies declared bankruptcy, twenty thousand businesses failed, and one-sixth of the workforce lost employment.

While the economic depression would slow some competitors down who would proceed with caution, there were others that were outright ruined. While pandemonium ensued and with everything on the line, Carnegie had enough capital and nerve to buy these failing mills at a discount. While competitors froze in a turbulent market, Carnegie continued building. Materials and labor were cheap, but by riding out the economic headwinds and continuing to build, when the depression was over, he was the only producer in Pittsburgh with a Bessemer plant capable of meeting the railroads' insatiable demand for steel rails.

Henry Bessemer's Process

Why exactly was the Bessemer process so sought after? Let’s start with the problem. Steel was expensive, therefore seen as a luxury and rarely used for what it would be most beneficial for: military and industrial projects. So the goal was to develop a solution at a fraction of the cost.

In 1856, Henry Bessemer created a method that would remove the impurities from the steel, making the steel much stronger. The process would blow cold air through molten pig iron in a converter vessel. The oxygen in the air reacted with the excess carbon in the iron, burning it off and simultaneously generating enough heat to keep the metal molten throughout the process. The impurities: carbon, silicon, manganese, were removed by oxidation, resulting in stronger steel. The process was also faster, reducing the production time by a whopping 98%, from two weeks of labor and controlled heating now to a single ten-minute converter blow. Because of its production speed, cost savings were acquired by reducing the expensive fuel needed to keep the furnace running.

Carnegie was no stranger to the Bessemer process, having been part of the consortium of investors who purchased the US process rights with Holley. But seeing the process in person during his 1872 trip to England, his excitement was palpable. The scale of the operation, the ease and cheapness of production, and learning that new sources of impurity-free iron ore had been discovered that would allow high-grade steel rail production, these all contributed to the strengthening of his conviction. His vision was set and he returned to Pittsburgh intent on building his own Bessemer mill at scale.

Britain’s competitive advantage allowed them to produce over 500,000 tons of steel annually in 1873 during the launch of Carnegie’s operation. The Bessemer process had been operating in the United States since 1865, when Alexander Holley designed and built the first American plant at Troy, New York. Holley went on to design 11 of the first 13 Bessemer plants built in America, focusing on innovation improvements, while Carnegie focused on efficient, cost-effective plants at scale. Holley racked up ten patents through improvements to the Bessemer process, like including a removable converter bottom that dramatically sped up lining changes, and a heat-recovery system.

Carnegie was always one to seek and retain top talent, ensuring they are paid generously and given creative liberty. He eventually hired Alexander Holley, the world’s top Bessemer plant engineer, to be the designer and engineer of the Edgar Thomson Works. The Pittsburgh Daily Gazette wrote that Holley's improvements made Edgar Thomson “the most perfect establishment of the kind in the world."

Holley’s previous employment at Cambria Iron Company in Johnstown, Pennsylvania, sparked a fortunate encounter with Captain William "Bill" Jones, one of the most skilled Bessemer operators alive with over 16 years of experience. Jones resigned from Cambria after he was passed over for superintendent, where Holley immediately hired him as his assistant for the Edgar Thomson project.

Carnegie instantly recognized Jones's genius and appointed him General Superintendent when the plant opened, superseding what he was aiming for at Cambria. The ROI on this decision was tenfold, as Jones’s value would be higher than any machine installed. He then recruited many experienced crew from Johnstown, cutting the learning curve that would have commenced had they not joined. What Jones possessed was a rare and highly sought-after skill, setting production records that would make Edgar Thomspon the vanguard of steel development in a period of rapid industrial and civic construction development.

The Naming Gambit

Carnegie was still a young upstart at 37 years old and was aware of his reputational stock. It was good, but didn’t carry the heavy weight of some of his colleagues, like Edgar Thomson, the then president of the Pennsylvania Railroad and the most influential railroad executive in America. So Carnegie decided to name his new steel mill the Edgar Thomson Steel Works, but not just to market its brand value. Carnegie strategically chose this name to court his potentially largest, and first customer: Thomson.

Thomson and Carnegie’s mentor, Tom Scott, had backed Carnegie’s prior ventures in bridges and iron. Aware of the importance of financing, the naming of the mill added another dimension: it made Thomson a shareholder, tying themselves and their success at the hip.

Carnegie took a risk, but it was a calculated risk. He was also on the board of the Harlem, the Erie, the New Jersey Central, and the Hartford & New Haven railroads simultaneously. He understood the purchasing patterns, the capital structures, and the decision-making of the men who would buy his rails before he had produced a single one. He knew what supply the demand wanted and personally knew those customers. This was actually a sale made in advance, disguised as a construction project.

The Panic of 1873

Construction of the Edgar Thomson Works broke ground on April 13, 1873, just four months after the completion of the land purchase. Strategically chosen along the Baltimore & Ohio and Pennsylvania Railroad lines, the 107 acres at Braddock along the Monongahela River, eight miles southeast of Pittsburgh, offered direct river access for raw material delivery from upstream. Reducing that distance for transport of this raw material resulted in cost savings. Every penny counts and every ounce of profit squeezed.

Now back to September 1873 when Jay Cooke's banking house failed, spurring an economic depression. The Panic thrust 21,000 rail-mill workers out of full-time employment, after so many businesses collapsed. Yet, construction on the mill did not stop during those depression years of 1873–1875. Many outsiders looking at this decision thought it as reckless, but Carnegie knew a principle used by value investors today: when the Market panics, you buy at a discount. The depression had driven down the cost of every input to the project: land, labor, materials, contractor rates. This was part of Carnegie’s greater business doctrine known as “counter-cyclical investing,” where his correspondences over 1873 reveal the essence of his philosophy: “Hard times are the best times to enterprise. Materials are at their lowest, labor is anxious to find employment at any price, and we can position ourselves to dominate when the market returns."

Even during economic depressions one can acquire funds so long as the business is sound, the executive principled, and the grit apparent. Carnegie had raised approximately $700,000 for the venture with $250,000 of that contributed by himself, and $450,000 across a number of partners, like William Coleman (Carnegie’s mentor who pushed him to start this venture), the financially astute Henry Phipps, and Carnegie’s younger brotherThomas. These funds led to the most advanced Bessemer plant in America, and constructed at a discounted cost during an economic depression.

Commencing Steel Production

August 26, 1875 marks the inauguration day, as the first Bessemer converter blew at the Edgar Thomson Steel Works. Only six days later on September 1, the first steel rail was rolled, with ceremony as the production of steel rails began. The results immediately spoke for themselves, as the output was high quality and consistent. The plant’s efficiency can be best expressed by Captain Jones: “This is the most powerful rail mill in the country; amongst its notable productions are a 62-pound rail 120 feet long, rolled in five minutes from the time of drawing the bloom from the furnace, and 600 rails 56 pounds per yard rolled in 11½ hours." Within one year of beginning production, the mill had produced 32,228 tons of steel rail, priced at $56 per ton – a steep discount from the $100–$120 per ton when construction began. As the operational and cost efficiencies continued, the price continued to fall, enabling construction developers to grow America’s cities uninhibited.

Profits soared with each year beating the last and by 1880, iron and steel orders surpassed pre-depression levels. The price gap between iron and steel rails closed, making steel the obvious choice. After a certain point in the 1880s, not a single new iron rail was ordered for rail tracks.

By 1900, Carnegie's company produced a quarter of the nation's Bessemer steel, with annual profits of $40 million. Carnegie’s share awarded him $25 million. In that single year he made 100 times the amount of his initial investment. By 1901 wanting to pursue his writing and philanthropic pursuits, Carnegie sold his parent company Carnegie Steel to J.P. Morgan in 1901 for $480 million as part of Morgan’s aim to curb the aggressive competition in the steel industry that was hurting profits. As of this writing, Edgar Thomson Works is still operating as the longest, continuously-running steel mill in America.

Key Insights

1. Adopt breakthrough technology early on: Scientific breakthroughs backed by meaningful capital make paradigm-shift-level advancements. The Bessemer process had existed for sixteen years before Carnegie built Edgar Thomson — but he was the first in Pittsburgh to back it with the capital and human talent required to realize its full potential. Holley's plant design and Jones's operational genius were only accessible to someone who could pay for them and understood what they were buying.
2. Stay calm when the markets don’t: Three-time Formula One World Champion Ayrton Senna said: "You cannot overtake 15 cars in sunny weather... but you can when it's raining."The best time to expand is when no one else dares take the risks. When the markets react in panic, you need to stay calm and abide by a solid plan that lacks emotion. Carnegie's 25% savings on construction was the direct result of building through the Panic of 1873. While others closed shop, he went bargain hunting.
3. Set a high bar for sourcing quality talent: Carnegie was not the first to build a Bessemer mill – he was the eleventh in America. What distinguished him was who he hired to build it, who he named it after, and what he did during the depression while everyone else was waiting for it to end.

Modern Day Parallels

Nvidia’s Commercialization of AI Infrastructure

On a particularly uneventful day at a Denny’s in San Jose, CA, three electrical engineers were eating their breakfast over an energizing conversation. It was 1993, and Jensen Huang, Chris Malachowsky, and Curtis Priem were brimming with ideas. Leading up to the creation of the internet, it was clear that computing and its capabilities were growing fast. This acceleration in computing was the future and Jensen firmly believed in this direction. Most computers at the time relied almost entirely on the CPU (central processing unit), but Jensen believed specialized processors like GPUs (graphics processing unit) would eventually become essential, not just for graphics and gaming but for computation itself.

So the three engineers started discussing the creation of a company predicated on this, at the time, very controversial bet. The PC market was fragmented, many graphics companies were failing, and no one knew what the winning architecture would even be. To create Nvidia at this time seemed like a reckless business decision, much like the massive financial capital bet behind the Bessemer plant during an economic depression

The founders were often reminded of this “reckless” decision, as it's been said that in the early years Jensen was often quoted as saying: "Our company is thirty days from going out of business." The company was on the brink of failure many times, but the capacity to endure pain and unshakeable determination are a founder’s greatest assets. Nvidia’s first products were unsuccessful, cash was always tight in such a capital-intensive business, and with limited resources they were often backed into a corner, forced to throw Hail Marys.

Nvidia initially focused on graphics processing, but as technology continued to evolve, they changed their view on GPUs. While others view it as essential to gaming hardware, Jensen increased his aperture and saw them as specialized processors capable of handling massive parallel computation. This distinction matters because AI ultimately depends on performing huge numbers of calculations simultaneously.

Long before OpenAI’s release of ChatGPT, Nvidia’s watershed moment came in 2006 when they launched CUDA (Compute Unified Device Architecture). Nvidia’s proprietary parallel computing platform and programming model allowed programmers to use GPUs for general-purpose computing rather than merely graphics rendering. By creating their own programming language, they also built their moat early on among early adopters of their platform. With CUDA, developers can offload complex, compute-intensive tasks from the CPU to an Nvidia GPU, using thousands of cores simultaneously to process massive datasets. At the time this looked like an obscure engineering decision, but in hindsight it was the equivalent of Carnegie standing in a British mill watching Bessemer steel. The market didn't fully understand the implications and while most semiconductor companies competed on chips, Nvidia quietly built an ecosystem.

This is the same position Carnegie was often in with the Edgar Thomson Works: an enormous concentrated bet that took an uncomfortably long time to look rational. These were also Jensen’s concentrated bets. When a nascent technology shows promise but hasn’t been validated yet, conviction is what keeps progress going. Accelerated computing was Jensen’s “Bessemer steel”; and before AI was even a buzzword and still fairly “dumb”, he saw the future and Nvidia’s place in it. For nearly two decades, investors questioned Nvidia’s heavy research and development efforts in both hardware and software, committed to recruiting only the top chip architects, compiler engineers, AI researchers, and systems designers. Each thoughtful hire was another Alexander Holley and Captain Jones. Without elite operators skilled in a nascent, revolutionary technology, the vision wouldn’t come together.

Just as Carnegie weathered the Panic of 1873, Nvidia survived AI winters, cryptocurrency busts, and Wall Street’s continual undervaluing the company because they would classify GPUs as solely for gaming products.

The greatest fortunes are often created when someone recognizes a technological discontinuity before the market understands its implications. While the inventor sees the breakthrough, it’s the entrepreneur who sees the consequences. While Bessemer invented the converter, it was Carnegie who saw what it would do to America. And while researchers developed neural networks, Jensen saw what they would do to computing.

As of this writing, Nvidia is valued at over $5 trillion as it continues to lead the AI boom. Nvidia didn't need to know whether OpenAI, Anthropic, Google, Meta, or another company would dominate the AI space. While they were out panning for gold, Nvidia chose to sell them shovels.