The Machine That Never Stopped

In 2023, electric motor systems consumed 53 per cent of the world's electricity. Seventy-two per cent of what industry uses, eighty-six per cent of what transport uses, eighty-seven per cent of what agriculture uses. The alternating-current induction motor, the specific machine patented in 1888, accounts on its own for somewhere between thirty and forty per cent of all electrical energy generated on the planet, and up to seventy per cent in some developed economies.

There is no other invention of the nineteenth century of which that can be said. The telephone was replaced. The telegraph was replaced. The steam engine was replaced. The induction motor was not improved into something else; it is the same machine, running the same way, doing more than half the work of the modern world.

This piece is about what that invention actually did, how it was directed, and what the economy it produced is now doing to itself.

What is still running

It is worth being exact about what the inventions were, because almost everything written about this man is either hagiography or myth, and the documented list is more impressive than either.

The rotating magnetic field and the polyphase system. The insight was not a device but a method: that several alternating currents, offset in phase, produce a magnetic field that rotates by itself. From that follows a motor with no commutator, no brushes and nothing to wear out, and a generator that is its mirror image. The patents date from 1887 and 1888 and were licensed to Westinghouse.

Alternating current as a complete system. Alternating current already existed; what did not exist was the whole chain. Generation, step-up transformation for transmission, long-distance lines, step-down transformation for use, and a motor at the end that could actually use it. That chain is why electricity could be generated in one place and used in another, and it is the reason a grid exists at all.

The proof at scale. In 1895 the Niagara Falls power station was built on those patents and in November 1896 sent alternating current to Buffalo, more than thirty kilometres away. Direct current could not have done it: voltage cannot be raised and lowered efficiently without transformers, and without that, losses over distance make long transmission impossible. Every grid on Earth descends from the decision made there.

Remote control. In 1898 he demonstrated a radio-controlled boat in Madison Square Garden, and patented the method. Every drone, every teleoperated machine and every remote actuator descends from that demonstration.

The radio patents. On 21 June 1943 the United States Supreme Court, in Marconi Wireless Telegraph Co. v. United States, ruled Marconi's patent invalid, finding that his claims had been anticipated by earlier work including Tesla's patents of 1900. The decision is frequently misreported as the Court declaring him the inventor of radio. It did not. It invalidated a patent, which is a narrower and more legally specific act.

Resonant coupling. The resonant transformer he built in 1891 is the ancestor of the resonant inductive coupling that charges a telephone on a pad today.

And the unit. The SI unit of magnetic flux density is the tesla, adopted in 1960.

Everything of his that is still running

The motor is the largest, and it is not the only one. These are the inventions of his that are in documented use today, with what each became.

The polyphase system and the induction motor, 1887 and 1888. More than half the electricity on the planet. Unchanged in principle.

Three-phase transmission. The standard adopted by every country on Earth. The reason a motor built in one continent works when plugged in on another is that this specific arrangement won.

The grid architecture. Generation in one place, transformation up for transmission, transformation down for use. Every electricity network in the world is this diagram.

The tuned resonant circuit. His patents of 1900 covered the method of tuning transmitter and receiver to the same frequency so that one signal can be separated from all the others in the air. That is still how a radio selects a station, how a telephone separates one network from another, and how every wireless device on Earth distinguishes its own signal from the noise. The 1943 decision of the United States Supreme Court in Marconi Wireless Telegraph Co. v. United States invalidated Marconi's patent on the ground that this work had anticipated it.

Remote control, 1898, patent 613,809. He demonstrated a radio-controlled boat in Madison Square Garden and patented the principle of controlling a machine at a distance by coded wireless signal. Every drone, every teleoperated vehicle, every remote actuator and every garage door is a descendant of that patent.

Resonant inductive coupling, from the 1891 resonant transformer. Two coils tuned to the same frequency transfer energy across a gap without contact. That is exactly what happens when a telephone charges on a pad, and it is the operating principle of the wireless charging standard used in hundreds of millions of devices.

The valvular conduit, 1920, patent 1,329,559. This is the one almost nobody knows they are using. It is a channel shaped so that fluid flows easily in one direction and is obstructed in the other, with no moving parts, no valve, no spring and nothing to wear out. For most of a century it was a curiosity. It is now in active use in microfluidics: in micropumps and micromixers, in lab-on-a-chip devices, in wearable biomedical sensors and diagnostics, and in fuel cells for managing hydrogen flow. A study published in 2021 found that the geometry works better than expected at turbulent flow and could be used to pump fluids using nothing but the vibration of the machine around it.

High-frequency gas discharge lighting. His demonstrations in the 1890s of gas-filled tubes excited at high frequency, lit without filaments, are in the ancestry of fluorescent and neon lighting.

And the unit. Magnetic flux density is measured in teslas, in the international system, since 1960. The field inside a hospital magnetic resonance scanner is quoted in that unit, which means the figure on the machine is his name.

That is eight technologies in continuous industrial use, from a body of work carried out mostly between 1887 and 1900, by one person.

What it did to the economy, and why nobody noticed for forty years

Here is the part that matters most and is almost never told, because it is an argument about productivity statistics rather than about genius.

Before the electric motor, a factory was organised around a single enormous steam engine. Power was distributed mechanically, by a central shaft running the length of the building, with belts dropping down to each machine. That architecture determined everything: machines had to be close to the shaft, the building had to be tall and narrow, the layout followed the power rather than the work, and if the engine stopped, everything stopped.

The induction motor made it possible to put a small, cheap, reliable motor on each individual machine. That is called unit drive, and it sounds like a detail. It was a reorganisation of industrial civilisation.

Once each machine had its own power, the factory could be laid out according to the sequence of the work instead of the geometry of the shaft. Buildings became single-storey and wide. Production lines became possible. Machines could be moved. Light and ventilation improved because the roof was no longer full of belts. And the energy no longer had to be bought in one enormous indivisible block.

And the measured economic effect took about four decades to arrive.

Electricity was available in American factories from the 1890s. The productivity surge associated with it does not appear clearly in the statistics until the 1920s. The reason, documented in the economic history literature, is that the gain did not come from replacing the steam engine with a motor. It came from redesigning the factory around what the motor made possible, and that required new buildings, new processes, new management and a generation of people who had grown up with the new thing.

That lag is the single most useful fact in the history of technology, and it is being argued about again right now, in this decade, about a different invention. The companies deploying artificial intelligence are discovering that buying the tool produces almost nothing, and that the gain arrives only when the organisation is rebuilt around it. That is not a new problem. It is the same problem, and the last time it took forty years.

How he directed it, and how he did not

The direction he gave this technology was a choice of system over device, and it is the reason it survived him.

He did not build a motor and sell motors. He patented a complete system and licensed it to a company with the capital to deploy it, which meant that the standard spread as an architecture rather than as a product. Had he kept it, the world might have ended up with several incompatible electrical systems, which is approximately what happened with railway gauges and is why some borders still require trains to change wheels.

He also chose the system over his own position in it. The royalty agreement with Westinghouse would have made him one of the wealthiest people of the century; he released the company from it in 1897, when its financing was in difficulty and the contract threatened the deployment itself. The arrangement survived. The architecture spread. That decision is the reason the world has one electrical standard rather than five, and it is the single most consequential thing he did after inventing the machine.

What it did to how human beings live

The economic figures are the smaller half of the story. The larger half is what changed about ordinary existence once power could be made in one place, divided into any quantity and delivered anywhere.

It removed the night as a boundary. For the whole of human history before this, activity after dark was limited by daylight and by the cost and danger of open flame. Cheap electric light moved the end of the day from sunset to bedtime. Work, study, reading, manufacturing, surgery and public life all expanded into hours that had previously belonged to nobody. One measurable consequence is that average human sleep duration fell as electrification spread, which is the first time in the record of the species that a technology changed how long people sleep.

It made the city vertical. A building without a powered lift stops at six or seven floors, because people will not climb further. The electric lift is a motor, and the modern city, with its density, its land values and its concentration of people and work, exists on the far side of that machine. Every skyline is an application of this patent.

It gave cities clean water. Delivering drinking water to millions of people and removing their sewage is, mechanically, a pumping problem, and pumps are motors. The sanitary transformation of the twentieth century, which is among the largest contributors to the rise in human life expectancy ever recorded, runs on this invention and is almost never credited to it.

It changed what humans eat and how long they live. Refrigeration is a compressor, and a compressor is a motor. The cold chain made fresh food available in winter, in cities, far from where it was grown, and it sharply reduced food-borne illness. Domestic refrigeration spread through households in the middle of the century and the incidence of stomach cancer fell across the developed world as it did, an association documented in the epidemiological literature.

It took a large share of unpaid domestic labour out of the home. The washing machine, the refrigerator, the vacuum cleaner and the rest are each a small electric motor in a box. Before them, the maintenance of a household consumed the working lives of an enormous number of people, almost all of them women. The economic history literature identifies the arrival of those appliances as one of the material enablers of women entering paid employment in the twentieth century, not as a cause on its own, but as the thing that freed the hours.

And it redefined the working day. Shift work, continuous operation, the eight-hour day and the twenty-four-hour factory all depend on lighting and powered machinery that do not tire. The modern organisation of time is an electrical artefact.

Taken together, that is a different kind of consequence from a market figure. Over the twentieth century, human life expectancy roughly doubled worldwide. No single invention caused that, and it would be absurd to claim one did. But water that is pumped, food that is kept cold, hospitals that are lit and powered, and homes that no longer consume a lifetime of labour to maintain, are all the same machine, in different housings, doing work that was previously done by human bodies or not done at all.

That is the actual answer to what this invention did to the species: it transferred an enormous quantity of physical work from people to machines, and gave back the time.

And it is still doing it. More than half the electricity generated on Earth is, at this moment, being converted into motion by that machine, which means that most of the physical effort of keeping eight billion people alive is currently being performed by the thing patented in 1888, and not by anybody's arms.

Where it directs us now

The civilisation built on those patents is now running into its own limits, and three of them are visible.

The grid he made possible has become the binding constraint on computation. Everything this newspaper has documented in the past week, the data centres measured in gigawatts, the campuses with their own generation, the orbital proposals, is a story about electricity rather than about chips. The alternating-current grid is the thing being asked to carry it, and it was designed for a world whose demand grew slowly and predictably.

And the war of the currents is being partially re-fought. The reason alternating current won in 1895 was transformers, which made voltage easy to change. But power electronics have since made it possible to do the same thing with direct current, and over very long distances direct current now loses less. High-voltage direct-current links are being built across continents and under seas. Solar panels produce direct current. Batteries store direct current. Data centres increasingly distribute direct current internally because every conversion costs a few per cent. The verdict of 1895 is being quietly revisited, not reversed: alternating current keeps the distribution network, direct current takes the long lines and the inside of the buildings.

And the motor itself is moving. The induction motor is now in cars, including in the vehicles of the company that took his name, chosen for precisely the reason he patented it: nothing to wear out. Half of the world's electricity already goes into motors, and electrifying transport and heat will raise that share further, which means the efficiency of that one machine is among the largest single variables in global energy consumption.

Who got rich on it, and how much it is worth

The question that follows is how much money this invention has made and for whom, and the answer has a shape that this newspaper has now found three times in a week.

The companies that are the direct descendants

The licensee was Westinghouse, and the competitor that lost the argument and then adopted the technology was General Electric. Between them they industrialised the electrification of the world, and their successors are still among the largest industrial companies in existence.

As of October 2026, the five largest firms whose business is electrification carry the following market values. The electrification and power business spun out of General Electric is worth about $263.3 billion at the close on 2 October 2026, on trailing revenue of $41.4 billion, and its shares have risen about 63 per cent in twelve months on demand for power infrastructure driven by artificial-intelligence data centres. The German industrial conglomerate is worth about $246.8 billion on revenue of $94.2 billion. The French specialist in low-voltage distribution, which is the leader in several data centre product categories, is worth about $187.5 billion on revenue of $48.8 billion. The American power management company is worth about $169 billion. And the Swiss-Swedish group that is among the largest makers of motors and drives is worth between $115 and $122 billion on revenue of about $32 billion.

Those five together are worth, at present prices, a little under one trillion dollars.

Behind them sits a second tier whose entire existence is the machine itself: the Japanese group that is the largest maker of small precision motors, with revenue around $17.5 billion, the Brazilian manufacturer that is among the largest producers of motors, automation and drives in the world, and a dozen others.

And the number that puts it in proportion

The entire global market for electric motors and generators is about $137 billion a year.

That figure deserves a moment. Every motor sold on Earth, in every factory, vehicle, pump, fan, compressor, lift and appliance, in a year, is worth less than a fifth of what the five companies above are worth as shares.

Now put it against what those motors do. They consume 53 per cent of the electricity generated on the planet. World electricity consumption is on the order of thirty thousand terawatt-hours a year, so motors are turning something like sixteen thousand terawatt-hours annually. At any plausible average price, that is well over a trillion dollars of electricity a year passing through machines of this design, every year, to do the physical work of civilisation.

The machine sells for a hundred and thirty-seven billion. The electricity it consumes is worth ten times that. And the output it produces, which is the movement of nearly everything that moves in industry, agriculture and transport, is not priced at all, because it is simply the economy.

What each invention is worth, one by one

The motor is the largest but not the only one with a balance sheet attached. Taken individually, these are the industries that exist on top of each patent, who buys from them and who buys from those.

The polyphase system and the induction motor. The companies that make them are worth a little under a trillion dollars between the five largest, and sell about $137 billion of motors and generators a year. Their customers are every utility, every factory, every water supply, the entire cold chain, every lift and every ventilation system. Their customers' customers are everybody. And the physical quantity underneath it all is 53 per cent of the world's electricity, which is on the order of sixteen thousand terawatt-hours a year passing through machines of this design.

The tuned resonant circuit. Selecting one signal out of all the others by tuning two circuits to the same frequency is the operating principle of every radio receiver and every wireless device. The industry that sells the equipment is worth, depending on how the category is drawn, between about $370 billion and $706 billion a year, with the wireless infrastructure segment alone at around $255 billion and projected past $600 billion by 2033. Its customers are the telephone operators. Their customers are four billion people with a telephone. There is no device in that chain that does not contain a tuned circuit.

Resonant inductive coupling. Transferring energy between two coils tuned to the same frequency, without contact, is a market of about $45.9 billion in 2026, projected to $237 billion by 2034. The devices conforming to the dominant standard account for about $11.2 billion of that, and the integrated circuits that control it about $7 billion. Its customers are the telephone and vehicle manufacturers; theirs is everyone who has ever put a phone on a pad.

Remote control. Operating a machine at a distance by coded wireless signal, patented in 1898, is now the basis of every drone, every teleoperated vehicle, every industrial remote and every garage door. It has no single market figure because it is not a product; it is a feature of thousands of them.

The valvular conduit. A channel that passes fluid one way and obstructs it the other, with no moving parts, is now used in microfluidics: in micropumps and micromixers, in lab-on-a-chip devices, in wearable diagnostics and in fuel cells. That market is about $46.9 billion in 2026 and is projected at $105.1 billion by 2033. Its customers are diagnostics and pharmaceutical companies. Theirs are hospitals and patients.

High-frequency gas discharge lighting. His demonstrations of filament-free illumination in the 1890s are in the direct ancestry of fluorescent lighting, which lit the twentieth century indoors and is only now being displaced by a different physics. The lighting industry that replaced the filament bulb is a market of over $115 billion a year.

Add the annual markets of the industries sitting directly on these patents and the figure is in the region of $715 billion to $1.05 trillion a year, every year, before counting the electricity itself and before counting anything built on top.

How much has it generated since 1888

A total figure is an estimate rather than a fact, and anyone who gives one without showing the arithmetic is guessing. What follows is the arithmetic, in three tiers, with every assumption stated so that the reader can disagree with it.

Tier one: the companies that sold the equipment.

The whole global market for electric motors and generators is about $137 billion a year today, and was far smaller for most of the last century. A generous estimate of cumulative revenue from selling the machines themselves, across the entire period and in today's money, is in the low single-digit trillions of dollars. Add the equipment that surrounds them, transformers, switchgear, cable and drives, and the figure is larger but still of that order. The five largest electrification companies are worth a little under one trillion dollars today, in total, which is the market's present valuation of all the future profits of that business.

So the direct monetisation of the invention, over 138 years, is measured in single-digit trillions. It is the smallest of the three tiers by a wide margin.

Tier two: the electricity that has passed through the machines.

This one can be approximated from physical quantities rather than from accounts.

World electricity generation is currently around thirty thousand terawatt-hours a year. It was roughly a thousand in 1950, about eight thousand in 1980 and about fifteen thousand in 2000. Integrating that growth from the beginning of electrification gives cumulative generation on the order of nine hundred thousand to one million terawatt-hours.

At an average of ten cents per kilowatt-hour, which is a conservative figure for retail electricity in current money, one terawatt-hour is worth about one hundred million dollars. So the cumulative value of all the electricity ever sold is on the order of one hundred trillion dollars in today's money.

Motor systems consume 53 per cent of it. Which puts the electricity that has passed through machines of this design, since they were patented, at something on the order of fifty trillion dollars, and over a trillion dollars a year at present.

That number is an estimate built on three assumptions, and a reader who prefers different ones will get a different answer. What the assumptions cannot change is the order of magnitude: it is tens of trillions, not hundreds of billions.

Tier three: what was built on top of it, which cannot be attributed and can be bounded.

World output is now around one hundred and ten trillion dollars a year. Essentially none of it is produced without electricity, and most of the physical part of it is produced by motors. Cumulative world output since 1890 is in the hundreds of trillions of dollars in today's money.

It would be meaningless to attribute that to one patent. Electricity would have been distributed eventually by some route, and a hundred other inventions were necessary. But the bounding statement is defensible and it is this: before the polyphase system, power had to be generated where it was consumed. After it, it did not. Every activity that depends on separating those two places, which is nearly all industrial production, every city, every cold chain, every hospital and every data centre, exists on the far side of that change.

And the figure that is actually comparable.

The companies currently spending the largest sums in the world on electricity supply are the ones building computing capacity. The data centre operator described in these pages has 4.2 gigawatts of contracted power. The accelerator manufacturer has guaranteed $108.5 billion of land, power and shell construction. The orbital proposal is denominated in a hundred gigawatts.

Every one of those numbers is a quantity of electricity, delivered to motors, transformers and cooling systems, through a grid architecture that was settled in 1895 at Niagara. The most advanced industry of this decade is buying its principal input from a system designed before the aeroplane.

And what the inventor received.

The royalty agreement he signed with Westinghouse was reported at $2.50 per horsepower of alternating current sold. Applied to a century of global motor production, that rate would have produced a fortune without precedent in private hands.

He surrendered it in 1897, when the company's financing was in difficulty and the contract threatened its survival, in exchange for the system continuing to be built. He received a lump sum that is variously reported and was, by every account, a small fraction of what the royalty would have been worth within a decade.

He died in 1943 owing money to the hotel he lived in.

Who uses it, and who uses them

The economic repercussion is easiest to see as a cascade, because each layer pays the one beneath it and almost nobody notices the bottom one.

The layer that makes the machines. The electrification companies listed above, worth a little under a trillion dollars in total, selling motors, drives, transformers, switchgear and grid equipment. The global market for the motors alone is about $137 billion a year.

The layer that cannot operate without them. This is almost all physical production. Every utility and grid operator. Aluminium smelting, which is in practice the conversion of electricity into metal. Every municipal water supply, which is pumps. The entire cold chain, because refrigeration is a compressor and a compressor is a motor, which means the world's food system runs on this invention. Every factory, every lift and escalator, every ventilation system in every building, every railway, and now every electric vehicle.

The layer that uses those. Here is where it reaches this decade. A data centre buys transformers, switchgear, uninterruptible supplies and chillers, and the chillers are motors. The operators described elsewhere in this newspaper, with gigawatts of contracted power, are customers of the companies in the first layer. That is why the electrification business spun out of General Electric has risen about 63 per cent in twelve months, and why the French distribution specialist is the leader in several data centre product categories.

And the layer above that. The artificial-intelligence companies renting that computing capacity, and then every business using their products.

Follow the money up and the chain is: a model answers a question, because a company rented a cluster, because an operator built a hall, because a utility delivered power through transformers to motors that move air and water, all of it designed around a standard settled in 1895.

Which produces a sentence that is strange and verifiable at the same time. A measurable part of the capital now flowing into artificial intelligence ends up in the accounts of companies founded in the nineteenth century to sell electrical equipment. The newest industry in the world is paying rent to the oldest one in the building.

The law that keeps appearing

Here is the finding, and it is the same one documented in these pages about two entirely different industries in the same week.

The layer that makes everything possible captures the least.

The company with one hundred per cent of the market for extreme ultraviolet lithography earns a gross margin of around half, while its customer earns 67.7 per cent and the designers above them earn 75. The foundry that manufactures essentially all advanced logic is worth a fraction of the companies that design what it builds. And the motor, which does more than half the physical work of the world, is a commodity sold in a $137 billion market by companies with thin margins, while the value accumulates in everything built on top of it.

The man who invented it is the extreme case of the same law. The royalty agreement he signed with Westinghouse would have made him, by most estimates, the richest man of his century. He surrendered it when the company's finances were threatened, in order to keep the system being deployed. He died in a hotel room in 1943, in debt, in a world in which his machine was already doing most of the work.

And the part that cannot be counted

The honest limit of the arithmetic is that the largest number is unmeasurable.

The companies above are the ones that sell electrification. The ones that exist because of it are every company that requires a factory, a cold chain, a lift, a server, a pump or a production line, which is to say almost all of them. Attributing their value to one patent would be meaningless.

But the specific, defensible version is this. Before 1888, power had to be generated where it was used, mechanically distributed, and consumed in one indivisible block. After it, power could be made in one place and used in another, divided into any quantity, by anybody. Every business model that depends on that, from the aluminium smelter to the data centre, exists inside the door that was opened then.

Which is why the companies now spending hundreds of billions of dollars on the electricity supply for artificial intelligence are, without exception, buying equipment from the direct corporate descendants of the people who built the first grid.

What this establishes

That the single most consequential invention of the industrial era is still, physically, doing more than half the work.

That the economic benefit of a general-purpose technology does not arrive when the technology arrives, but when the organisations around it are rebuilt, and that this took four decades the last time it happened at this scale.

And that the person who directed it did so by giving it away as a standard rather than keeping it as a product, which is why it is still running, and why he died without the money.

The argument now being conducted about whether a new general-purpose technology will show up in the productivity figures has been held before, with the same positions, about the machine that is at this moment turning the fan in the room where this is being read.

Sources

Electricity and motors. The share of global electricity consumed by electric motor systems, 53 per cent in 2023, with sectoral breakdowns of 72 per cent in industry, 86 per cent in transport and 87 per cent in agriculture, is from the International Energy Agency's work on electric motor systems. Estimates that induction motors alone account for 30 to 40 per cent of generated electrical energy worldwide are from the engineering literature.

Patents and the court. The polyphase system and induction motor patents date from 1887 and 1888 and were licensed to Westinghouse. The radio patents are US 645,576 and US 649,621. The decision of the United States Supreme Court in Marconi Wireless Telegraph Co. v. United States was issued on 21 June 1943 and invalidated Marconi patent claims on grounds of anticipation; it did not adjudicate the invention of radio.

Company figures. Market capitalisations and trailing revenues are as reported in October 2026. The size of the global electric motor and generator market is from published market research and is an estimate rather than a filed figure.

Economic history. The forty-year lag between the availability of electric power in factories and its appearance in productivity statistics, and its attribution to the reorganisation of production around unit drive rather than to the motor itself, is the standard account in the economic history literature on general-purpose technologies.

The valvular conduit. Patent US 1,329,559, granted 1920. Its current use in microfluidics, micropumps and micromixers, lab-on-a-chip devices, wearable diagnostics and fuel cell gas management is documented in the peer-reviewed literature, as is the 2021 finding on turbulent flow and vibration-driven pumping published in Nature Communications.

Market sizes. Figures for the markets in telecom and wireless infrastructure equipment, wireless charging, microfluidics and lighting are published market research estimates for 2026 and vary between research firms according to how each category is defined; the ranges are given in the text where they differ materially. They are estimates, not filed figures.

Estimates, identified as such. The cumulative totals in this article are constructed estimates, not reported figures. They are built from published world electricity generation data, from a stated assumption about average electricity price, and from the motor share of consumption cited above. The arithmetic is shown in the text so that different assumptions can be substituted. The royalty rate of $2.50 per horsepower and its surrender in 1897 are as reported in the historical record, where the terms of the settlement are variously described.

Scope. This article covers the inventions that are in documented industrial or scientific use today. It does not attempt to assess the later experimental work, which is outside what can be established from present-day deployment.