170 Years of Steel

There are broadly two paths to making steel. One starts from iron ore. In a blast furnace, coal strips away the oxygen from iron ore to extract molten iron, which is then poured into a basic oxygen furnace where oxygen is blown in to turn it into steel. The other starts from scrap metal. You put scrap metal into an electric arc furnace and melt it down with electricity. The former is called the blast furnace-basic oxygen furnace route, and the latter is called the electric arc furnace route.

The two routes differ right down to the nature of their facilities. Building a single blast furnace requires a massive amount of capital, and once fired up, it is not shut down for years. An electric arc furnace can be built on a relatively smaller scale, and it can be turned on and off almost like flipping a switch. That is why, even within the same industry, these two kinds of steelmakers operate to completely different rhythms.

The share of each route varies dramatically from country to country. According to World Steel Association figures for 2025, the electric arc furnace share is 71.3% in the United States, 72.2% in Türkiye, 57.7% in India, 45.8% in the European Union, 27.0% in South Korea, 25.8% in Japan, and 10.6% in China. For the very same product, the production mix across countries is split in completely opposite directions.

It would be a mistake to read this as a generational shift in technology. The electric arc furnace is not a new invention. The first electric arc furnace in the United States began operating all the way back in 1907. And even in 2025, 69.4% of the world's crude steel still comes from the ore-based route. What drives the divide comes down to two things: how much scrap metal a country has accumulated, and how cheap its electricity is. Scrap metal does not fall from the sky. It only exists once steel made in the past reaches the end of its useful life. That is why countries with long industrial histories have plenty of it, while countries that are just building up do not.

Emissions diverge sharply, too. According to association figures based on 2024 activity, producing 1 ton of crude steel releases 2.34 tons of carbon dioxide via the blast furnace-basic oxygen furnace route, compared to 0.69 tons for an electric arc furnace melting scrap metal. Most of the pressure currently weighing on this industry stems from that gap between the two numbers.

In the mid-19th century, iron came in three forms. There was cast iron, which was hard but brittle because of its high carbon content; wrought iron, where nearly all carbon was removed, making it soft and tough; and steel, which sat somewhere in between. Steel was superior, but it had to be melted in small quantities in crucibles, making it expensive. Because of that, both railway rails and bridges were mostly made of wrought iron.

In January 1855, Henry Bessemer filed a patent titled "Improvements in the Manufacture of Iron and Steel," and on August 13, 1856, he announced his method at the British Association for the Advancement of Science in Cheltenham. His core idea was that blowing air through molten pig iron would burn the carbon, and the heat produced would make the molten metal even hotter. This meant carbon could be controlled without adding extra fuel. On September 6, twenty-four days after his presentation, two steel ingots produced in London were rolled into rails at the Dowlais Ironworks in Wales.

The difference in scale was immense. A crucible could melt only 40 to 50 pounds at a time—roughly the size of a bag of rice. Bessemer's converter could produce 5 tons in around 20 minutes, with the air-blowing phase itself lasting 15 to 20 minutes. The unit of production had scaled up a hundredfold.

This is where popular lore departs from the original text. In his paper, Bessemer explained that stopping the process at different stages yielded various grades, and he gave one of them a distinct name: semi-steel. It offered a hardness halfway between cast steel and soft wrought iron, aimed squarely at uses demanding both lightness and strength. The example he cited was railway rails, which were soft, prone to splintering along the grain, and quickly worn down. What he actually wrote about cost referred specifically to this grade: As the rolling is somewhat more difficult, the cost per ton may be fairly considered as equal to that of iron. He then added that because its tensile strength was 30 to 40% higher than bar iron, a substantially lighter weight could be used for most purposes.

In short, what the inventor pitched was not that the price had collapsed. It was that people could now use a stronger metal at roughly the same price as wrought iron. It meant building the same bridge with far less metal. The popular comparison of "£6 to £7 per ton versus £50 to £60 per ton" does appear in his autobiography. Yet the former figure was his own manufacturing cost, while the latter was the prevailing market price of crucible cast steel. It placed cost and retail price side by side, and for two entirely different products at that.

Within days of the Cheltenham announcement, British ironworks came rushing in. The Dowlais Iron Company paid £10,000 for the right to produce 20,000 tons a year at a bargain royalty of 1 farthing per ton. The Govan Iron Works in Glasgow followed with £10,000 on the same terms, Butterley in Derbyshire signed on for 10,000 tons a year, and a Welsh tinplate maker took 4,000 tons a year.

And every single trial ended in disaster. The ordinary pig iron used in Britain for bar iron contained far too much phosphorus. Steel with phosphorus becomes brittle when cold. Bessemer wrote in his autobiography that the news struck like a bolt out of the blue. Each day brought fresh reports of failure, and newspapers were filled with letters attacking him.

He chose proof over words. He imported Swedish charcoal pig iron, which had virtually no phosphorus. Delivered in Sheffield, the raw material cost between £6 10s and £7 per ton. When he fed this into his converter, it yielded high-purity steel ingots at less than £10 per ton. This was at a time when Swedish bar iron was selling for £16 to £24 per ton.

What came next is what really mattered. He built his own works in Sheffield and started making and selling it himself. By 1858, his tilting converter mounted on trunnions was running. He entered the market at £10 to £15 per ton below what others were asking, quoting high-grade tool steel at £42. Other manufacturers were asking £50 to £60 for the exact same grade. The price had certainly dropped, though it had not changed by an order of magnitude.

Another comparison from his autobiography makes the real target of his invention even clearer. When railway rails made from British coke pig iron sold for £12 per ton, he estimated the cost of his steel ingots made from Swedish pig iron at around £10. His target market was not simply undercutting crucible steel, but offering steel at a price comparable to wrought iron. Rails, bridges, and ships were applications that could never afford crucible steel.

The phosphorus that brought Bessemer down wasn't just a British problem. Continental European iron ore was also rich in phosphorus. As a result, for a time, this method only worked in places where low-phosphorus ore could be found.

The answer arrived about 20 years later. Sidney Gilchrist Thomas and his cousin Percy Gilchrist changed the material used to line the inside of the converter. When a basic lining was used instead of a silica-based acidic one, oxidized phosphorus reacted with the wall and was trapped inside the slag. According to British engineering history records, Thomas first publicly announced at a meeting of the Iron and Steel Institute in March 1878 that he had successfully eliminated phosphorus inside the converter, filed a patent in May of the same year, and their joint paper with Gilchrist was read in May 1879.

What's fascinating is Bessemer's own recollection. He wrote that the licensed ironworks lined their converter walls with silica-based materials, and that this prevented the removal of phosphorus, adding that this fact was later revealed through Thomas and Gilchrist's method. In other words, one of the culprits behind those early failures was not just the raw material, but the vessel itself.

This method was quickly adopted across Germany, Austria, France, Belgium, and Russia. Between 1884 and 1885, basic steel produced on the continent reached 748,000 tons, with Germany and Austria accounting for 617,000 tons of it. The very flaw that caused one country's failure turned another's mines into valuable assets. As an added bonus, the resulting basic slag was sold as phosphate fertilizer. Waste that used to be thrown away had turned into a commercial product.

There is an important point to note here. Steel did not become abundant because of a single breakthrough invention. It became cheap and plentiful bit by bit as steelmakers changed the raw materials, modified the vessels, and accumulated experience making the same product over and over again. From the Cheltenham announcement to mass production across the continent, it took nearly 30 years. It wasn't a single giant leap, but a series of corrections.

We are all familiar with the story that cities grew upward once steel became common. But attributing that entire cause and effect to steel alone does not match the facts. We need to distinguish three things.

First, the frame of the building often called the first skyscraper was not made of steel. In describing the Home Insurance Building, completed in 1885, the Chicago Architecture Center notes that a cast-iron skeleton frame carried the load, calling the building "one of the world's first skyscrapers." It stood 10 stories and 138 feet tall, and two more stories were added in 1891. Economist Jason Barr points out that the building lacked the technical criteria of a true skyscraper: a complete steel skeleton, separate wind bracing, and a non-load-bearing exterior envelope. Only the street-facing walls encased iron columns inside brick piers, while the rear wall was simply a load-bearing wall.

Second, there is a counterexample of a building that rose high without a steel frame. The north section of the Monadnock Building, also in Chicago, was completed in 1891, rising 16 stories just by piling brick upon brick. In exchange, the corner walls were six feet thick. When a 1.8-meter wall takes up the ground floor, usable space to rent out disappears. That was why they could not build any higher. Only with the south section built two years later was the exterior relieved of its load and hung from a metal frame.

Third, a framework alone cannot bring people up to the higher floors. Otis's official history records the safety brake in 1853, a public demonstration cutting the cable in 1854, and the first passenger elevator in New York's Haughwout Building in 1857. Without an elevator, top floors are spaces nobody wants to visit.

To put it simply: what steel resolved was not so much the physical limit of height, but the cost of wall thickness. And for that space savings to turn into profit, elevators, fireproof construction, and high land values had to arrive together.

Up until the mid-20th century, steelmaking across the world was handled by open-hearth furnaces. It was a slow process, but it allowed producers to blend in a large amount of scrap metal. The process that took its place came from an unexpected corner.

After 1945, Austria had no scrap metal. It had no energy, and no materials to make it either. The Austrian Iron and Steel Plan published in 1948 proposed introducing the oxygen converter steelmaking process, which the country had not used up to that point. The reason was spelled out in the document: Because there was not enough usable scrap to meet the annual target of 1.07 million tons using conventional methods. The goal was to reduce dependence on imported scrap and raise the share of pig iron.

On November 27, 1952, the first converter tapped molten steel at a steelworks in Linz. It officially opened on January 5 of the following year, passed 100,000 tons by June 17, and surpassed 250,000 tons in early December. A second plant began operating in Donawitz on May 22, 1953. Taking the initials of the two locations, it was named the LD converter. Refining that once took several hours was cut down to tens of minutes.

The order in which licenses went out is fascinating. The first license outside Austria went to Canada's Dofasco in August 1954, followed by Japan's Yawata in September 1957, America's Jones & Laughlin in November 1957, and Japan's Nippon Kokan in January 1958. Those that had just built open-hearth furnaces were the slowest to embrace the new method. It was not because the laggards were foolish. Scrapping equipment with years of depreciation still left on the books is never an easy calculation for any company. The brochure notes that LD steel faced severe hurdles before being accepted into national industrial standards, only gaining equal status with open-hearth and electric furnace steel in 1966. And that was despite one-fourth of the world's crude steel already coming from this process by that point.

While Austria turned toward iron ore because it lacked scrap, scrap was piling up on the other side of the world. The United States had already been using steel for more than 100 years. As automobiles, buildings, and machinery reached the end of their lifespans, scrap metal kept pouring out.

The electric arc furnace itself was not new technology. According to Nucor records, America's first electric arc furnace operated in 1907. What changed were the scale and the application. In 1962, a company that acquired a steel joist manufacturer decided to produce its own steel, and in 1969, it fired up an electric arc furnace mill in Darlington, South Carolina. Without blast furnaces, the initial investment was far lower, and they could dial production up and down according to demand. That was something blast furnace operators could rarely do.

Yet there were limitations. Because scrap steel carries impurities like copper, electric arc furnaces remained confined for decades to products with looser specifications, such as rebar and structural shapes. Flat products like automotive sheet steel were the exclusive territory of large integrated mills with blast furnaces. That barrier fell in 1989. SMS group, which built the equipment, notes that in that year it built the world's first thin-slab casting and rolling plant at Nucor's Crawfordsville, Indiana mill. By casting molten steel thin and rolling it immediately, entire steps of the process were cut out.

Today, about 70% of U.S. steel production comes from electric arc furnaces. According to World Steel Association figures, it is 71.3% in 2025. In 2024, Nucor alone recycled 20.3 million net tons of scrap across 26 steel mills. It became an industry that sources its raw materials not from mines, but from junkyards and demolition sites. In this business, the very age of a nation's industrialization becomes a resource.

It is worth considering the circumstances of those who were pushed aside. Large integrated mills with blast furnaces did not hold back from switching to electric arc furnaces simply because of slow decision-making. An existing blast furnace, along with its coke ovens, sinter plant, and harbor facilities, is tied together as a single package. Take one piece away, and the rest sits idle. Moreover, scrap alone cannot meet the volume of steel the world consumes. The reality that recycling alone cannot keep up with growing demand is an enduring constraint of this industry.

Global crude steel production stood at 189Mt in 1950. It climbed to 595Mt in 1970, 717Mt in 1980, 770Mt in 1990, 850Mt in 2000, and 1,148Mt in 2010. Most of this growth curve was driven by newcomer countries building anew.

You can see this order clearly in the case of South Korea. According to records from Woori History Net, Pohang Iron and Steel Company was established in 1968, Phase 1 construction began in April 1970, and Blast Furnace No. 1 tapped its first molten iron on June 9, 1973. Phase 1 had an annual capacity of 1.03 million tons, and through repeated expansions, it reached a cumulative 9.1 million tons. What is worth noting here is that because they built late, they could install state-of-the-art facilities right from the start. Competitors who already possessed open-hearth furnaces and aging blast furnaces did not have that choice.

China came even later, and on a far larger scale. According to statistics from the World Steel Association, in 1995 Japan produced 101.64 million tons and China produced 95.36 million tons. The very next year, in 1996, the ranking flipped as China reached 101.24 million tons while Japan produced 98.80 million tons. In the 2025 tally as well, China remains number one.

Latecomers had another advantage. The operating methods and equipment designs that pioneer nations had acquired through trial and error had already become commercial products on the market. LD converters, continuous casting, and thin-slab casting were all commodities sold by engineering firms complete with blueprints. Latecomers could pay the price to skip the tuition earlier pioneers had paid. In return, those who built first had to scrap what they had already built just to stand on the same ground.

As of 2025, production stands at 960.8Mt for China, 164.9Mt for India, 81.9Mt for the United States, 80.7Mt for Japan, 67.9Mt for Russia, and 62.2Mt for South Korea. With the global total at 1,848.9Mt, China's share is 52.0%. However, in that same year, China's production fell compared to the previous year's 1,005.1Mt. The condition where one country accounts for more than half, and the condition where that country's production is declining, are both true at the same time. To understand where this industry stands today, you have to hold both statements together.

Blast furnaces are not something you can just turn off. According to explanations from POSCO Newsroom, blast furnaces and steel mills run 24 hours a day year-round, and the molten iron that pours out exceeds 1,500 degrees. If a blast furnace is stopped, the molten metal inside solidifies into a colossal block of iron, requiring you to tear it down and rebuild it to bring it back to life. The cost is 500 billion to 1 trillion won, and it takes several months. On top of that, the downstream plants that rely on that molten iron grind to a halt as well.

That is why in regions with a high share of blast furnaces, production does not easily decline even when demand drops. There is a long stretch where running them is cheaper than shutting them down, even if prices tumble. The numbers reflect this reality clearly. According to OECD figures, global steelmaking capacity reached an all-time high of 2,445.1Mt in 2025, but actual production in the same year was only 1,848.9Mt. Excess capacity reached 640Mt, and the utilization rate dropped to 76%. The OECD projects that by 2028, excess capacity will grow to 745Mt and the utilization rate will remain in the 74% range.

The OECD itself notes that this is not a new problem. In the 1970s as well, optimistic long-term demand forecasts sparked investments across the globe, only for the first oil shock to hit right as new furnaces were being fired up. Trade relations deteriorated sharply, leading to the creation of the OECD Steel Committee in 1978.

Where capacity is currently being added also defies common perception. According to OECD tables, between 2021 and 2025, China's steelmaking capacity decreased by 11Mt (-1.0%). Over the same period, India added 41.4Mt (+28.8%), the Middle East 7.2Mt, and ASEAN 5.5Mt, while Japan cut 8.8Mt and the UK cut 4.8Mt. Meanwhile, China's steel exports reached an all-time high of 131Mt in 2025, which was greater than the entire European Union's crude steel production that same year. Cutting capacity and expanding exports are happening simultaneously within the very same country.

The response from most nations has largely taken the form of trade measures. When steel is considered to be sold below cost, countries impose antidumping duties; when it is seen as having received subsidies, they levy countervailing duties. The OECD also notes an increasing trend of affected products flowing through third countries to circumvent restrictions. Citing Section 232 of the Trade Expansion Act of 1962, the United States imposed a 25% tariff on imported steel in 2018, expanded it to derivative products in 2020, and through a February 2025 measure, eliminated country exemptions starting March 12. Retaliatory tariffs from trading partners followed, though some were suspended through negotiations.

Over the past 170 years, the demand placed on this industry was mostly "make it cheaper." The demand it faces now is of a completely different nature. It is asking the industry to change the very way steel is made.

The reason comes down to chemistry. Iron ore exists bound to oxygen. To strip that oxygen away, you need a substance that bonds with oxygen even more eagerly, and up to now, the carbon in coal has played that role. That is why carbon does two jobs in steelmaking: it acts as a reducing agent to remove oxygen, and it mixes into the iron to form steel. The carbon used as a reducing agent turns into carbon dioxide and escapes through the smokestack. No matter how much you refine the process, this chemical reaction itself remains.

Here is what the scale looks like. According to calculations by the World Steel Association, emissions per 1 tonne of steel in 2024 stood at 2.18 tonnes CO2e. Multiplied by that year's production of 1,886Mt, the entire sector generated around 4.1 billion tonnes CO2e. That represents 7–8% of the world's anthropogenic greenhouse gas emissions. However, we also have to keep in mind that this emission intensity is an aggregate of data voluntarily submitted by member companies, whose combined output accounts for just over half of global production.

Solutions are moving in two directions. One is using more scrap metal. The scrap–electric arc furnace route emits 0.69 tonnes per tonne, vastly lower than the 2.34 tonnes of the blast furnace–basic oxygen furnace route. The trouble is that scrap is finite, and it has already turned into a trade issue. The OECD notes that 42 countries restrict scrap exports. The other direction is changing the reducing agent. If hydrogen replaces coal, the byproduct is simply water. HYBRIT, launched in 2016 by Sweden's SSAB, LKAB, and Vattenfall, produced hydrogen-reduced sponge iron in 2021 and delivered the first steel made from it to Volvo Group.

Regulations have started moving too. Following a transition period from October 2023 to the end of 2025 where emissions were merely reported, the EU Carbon Border Adjustment Mechanism (CBAM) entered its definitive regime on January 1, 2026. Steel is among the initial target sectors alongside cement, aluminum, fertilizer, electricity, and hydrogen. But the price does not bite right away. Buying and surrendering certificates has been pushed to 2027, and even then it applies not to all reported emissions but only to what remains after subtracting the free allocation EU plants receive. The 2026 share is 2.5%, reaching 100% in 2034. However, the OECD also notes that with energy potentially accounting for up to 40% of steelmaking costs, several low-emission projects are being delayed.

There is no need to forecast this industry's next scene. Instead, where to look is fairly clear. There are four places.

First is where and how much scrap steel accumulates. For electric arc furnaces to expand, they need raw materials, and that feedstock only appears when previously produced steel reaches the end of its life. Because China's production exploded after the 2000s, once the steel made back then starts returning, the country's mix of production processes could change as well. We just need to watch where that current figure of 10.6% heads next.

Second is who is expanding capacity. The places that added capacity from 2021 to 2025 were India, ASEAN, and the Middle East. Within a few years, it will become clear whether demand in the countries building anew is genuinely growing by that much, or if optimism ran ahead of reality just like in the 1970s.

Third is how emissions regulations attach themselves to prices. Now that the Carbon Border Adjustment Mechanism has begun operating in earnest, it is worth watching whether prices diverge for the same steel depending on how it was made. Just because a policy is created does not mean it happens right away.

Fourth is how facilities that cannot simply be turned off are wound down. According to OECD tallies, global excess capacity now stands at 640Mt. The way to reduce this could be trade measures, national capacity policies, or simply holding out for a long time before closing down. Whichever path it takes, the decision to shut down a single blast furnace involves months and costs in the trillions of won. That is why this industry moves more slowly than others.

The story that began in Cheltenham 170 years ago has come this far. What was sold back then was not 'cheap steel,' but 'a stronger metal for the same price.' What is being sold now is 'a metal that emits less for the same price.' First came the demand to cut weight, next to cut cost, and now the demand is to cut emissions.