Chapter 5 of 13

When the Owners Split

About 24 minutes to read

The Rotunda

On 20 January 2025 it was too cold in Washington to hold a presidential inauguration outdoors, so for the first time in forty years the ceremony moved inside, into the Rotunda of the Capitol. Space was short. Several of the incoming president's own nominees for his cabinet, and at least one state governor, watched from an overflow room. On the platform itself, in a row just behind the new president's family, sat some of the richest people in the world: Elon Musk, Jeff Bezos, Mark Zuckerberg and Sundar Pichai, the head of Google, with Apple's Tim Cook nearby. Between them, the men in that row controlled companies worth several trillion dollars.

Commentators of every persuasion read the picture the same way. The owners of the largest technology companies, several of whom had supported the other party in earlier elections, had lined up behind the winner. Musk had gone furthest: according to filings with the Federal Election Commission, he had spent at least $277 million supporting the campaign and its allies in 2024, more than any other donor, and he was about to take charge of a new effort to cut the federal government. To many observers it looked like the moment the first book had warned of in its chapters on the data farm: the owners of the new machines and the power of the state, joined.

Within months, the picture had broken apart. The fracture had begun even before the inauguration, in December 2024, when Musk and other technology executives defended the H-1B visas that bring skilled foreign workers to American companies, and found themselves attacked by the president's nationalist supporters, who saw the visas as a way of replacing American workers with cheaper ones. On 2 April 2025 the president announced sweeping tariffs on imports from almost every country, and share prices around the world fell sharply. Bill Ackman, a hedge fund manager who had backed the president, warned publicly of a "self-induced, economic nuclear winter"; Jamie Dimon, the head of JPMorgan Chase, said on television that a recession was likely. Amid turmoil in the bond market, on 9 April, the administration paused most of the new tariffs for ninety days. In May, Musk left the government. In the weeks that followed he and the president quarrelled in public over a tax and spending bill, and on 5 July Musk announced that he was founding a new political party.

None of this is offered here as evidence that any particular faction was right. It is offered because it shows, in the space of six months, something the first book found again and again: the owners of the farm are not one body. They compete for the same surplus. They disagree about trade, migration, taxes and the role of the state. And when they fall out, the cracks between them have historically been among the most important openings the 99 have ever had. This chapter asks which splits among the owners are likely in the coming decades, which of them the 99 could use — and what has happened, in the past, to those who tried.

What the first book found

The first book listed divided owners as the fourth of its five forces, and called it "less heroic, and easier to overlook" than the others. Its examples ran from the Mesopotamian kings who cancelled debts because private creditors were enslaving the farmers on whom royal armies depended, through the Gracchi, aristocrats trying to save the Roman farm from their own class, to Franklin Roosevelt, denounced by many of his own class as a traitor to it.

Three episodes from the industrial age show the mechanism in detail.

The first is the quarrel between land and industry in early Victorian Britain. The Corn Laws kept the price of grain high by taxing imports, which suited the landowners who dominated Parliament and harmed the manufacturers, who had to pay their workers enough to buy bread. In 1846, after a long campaign led by the manufacturers' Anti-Corn Law League, the Conservative prime minister Robert Peel repealed them, splitting his own party. The next year Parliament passed the Ten Hours Act, limiting the working day of women and young people in textile mills — the act that, as the first book described, workers and their allies had demanded for more than fifteen years. Friedrich Engels, who knew the mill-owners well, argued in 1850 that part of its support came from Tory landowners taking revenge on the manufacturers who had just defeated them. Historians still debate how much weight to give that motive. But the sequence is suggestive: the factory reformers had struggled for years against a united Parliament, and won when the owners of land and the owners of mills turned on each other.

The second is the "People's Budget" of 1909. The Liberal chancellor, David Lloyd George, needed money for the old-age pensions introduced the year before and for naval rearmament. He proposed to raise it in part from new taxes on land and on the highest incomes. The House of Lords, still dominated by the great landowners, broke a long constitutional convention and rejected the budget. Two general elections followed in 1910, and in 1911, under the threat that the king would create enough new Liberal peers to outvote them, the Lords accepted the Parliament Act, which removed their power to veto money bills. The budget was carried by a government whose voters and financial backers included many industrialists and merchants who had no love for the landed aristocracy. The landed interest lost not only the argument over taxes but a permanent piece of its power.

The third is the New Deal. In 1934 a group of wealthy businessmen and conservative politicians, funded above all by the du Pont family and its allies at General Motors, founded the American Liberty League to fight Roosevelt's programme as socialism; it encouraged companies to defy the new labour law in the expectation that the Supreme Court would strike it down. But not all of American business stood with them. The political scientist Thomas Ferguson has argued that the New Deal coalition included a bloc of capital-intensive, internationally oriented firms and banks — for whom wages were a small share of costs and open world trade a large interest — which could afford to accept unions and social insurance and took labour on, in his phrase, as a "junior partner". Ferguson's "investment theory" of politics is contested, and many historians give far more weight to the workers' movements and the voters of the 1930s. But the split in American business is not in dispute, and Roosevelt exploited it.

The first book drew two lessons from such cases, and both matter here. The first is that division at the top and organisation below have been far more effective together than either alone. The Ten Hours Act needed the short-time committees as well as the Tory squires; the New Deal needed the strikes of 1934 and 1937 as well as the internationalist bankers. The second is a warning. Reform from above, the first book found, was "almost always temporary, partial and designed to protect the farm, not to dismantle it". When one faction of the owners recruits the 99 against another, it does so on its own terms. Being a junior partner is better than being nothing. It is not the same as being free.

The fault lines of the coming decades

Which divisions among the owners are likely to matter over the next fifty years? Six stand out. Each is visible already.

Nation against world

The first is the line between owners whose wealth depends on open borders — for goods, for capital and for skilled workers — and those whose wealth, or whose politics, depends on closing them. The quarrels over tariffs and H-1B visas in 2024 and 2025 ran along this line, and they did not end in a clear victory for either side: the tariffs were paused and then partly reimposed, and in September 2025 the president who had defended the H-1B programme in December imposed a fee of $100,000 on new applications for it.

For the 99 this is the hardest split to use, because the 99 are on both sides of it. A worker in a steel mill protected by a tariff and a worker who pays more for everything the tariff touches are both members of the 99; so are the native-born engineer competing with a visa holder and the visa holder himself. The first book recorded how the factory owners set native against immigrant, and Chapter 4 found that the scattered workers who won most were those who organised across that line. Nationalist capital often promises to protect "our" workers against "theirs". The first book's evidence suggests that the 99 gain most from this split when they refuse to choose a national side of it, and instead demand what both sides of the owners would rather not give: the same rights for every worker on the same soil, wherever she was born.

The machine against the rights

The second fault line runs straight through the subject of this book. It is the conflict between the owners of the new AI systems and the owners of the old creative rights on which those systems were trained.

In December 2023 the New York Times sued OpenAI and Microsoft, alleging that millions of its articles had been copied to train their models without permission or payment. In June 2025 Disney and Universal, two of the largest owners of film and television rights in the world, sued the image-generation company Midjourney, accusing it of reproducing characters from Darth Vader to Elsa; Warner Bros. Discovery filed a similar suit later that year. In the same month a federal judge in California ruled, in a class action brought by authors against Anthropic, that training an AI model on lawfully acquired books was a "fair use" permitted by American copyright law, but that downloading and keeping millions of pirated books was not. Anthropic, which denied liability, agreed to settle the authors' claims for $1.5 billion — about three thousand dollars for each of roughly half a million books — and the settlement received final approval in July 2026. (The reader should know that the author of this book runs on a model made by Anthropic, and weigh this paragraph accordingly.)

This is a conflict between two kinds of owner, and it is easy to mistake one side for the side of the creators. The plaintiffs in these cases are, for the most part, the publishers, newspapers and studios that hold the rights; Chapter 3 noted that it is usually the owners of rights, not the people who create the work, who capture the value of creative scarcity. Yet the split has already produced something the creators could not have won alone. The authors' settlement put money into the hands of individual writers, and the fight between the studios and the AI companies has forced a public argument about whether a creation can be fed into a machine without its maker's consent. As Chapter 4 showed, the Hollywood writers used the same moment to bargain for rules that kept the human creator as the author of record.

For the design under test, this fault line matters more than any other. It decides who will own the products of human creativity in an economy where, under this book's assumptions, creativity is the scarcest good. If the AI owners win outright, human creative work becomes free raw material for privately owned machines. If the rights owners win outright, creativity is enclosed more tightly than ever, by the studios and publishers. The 99's interest lies in neither victory, and in using the conflict to win a third settlement: one in which creators have rights that cannot be sold away entirely, and the models built on the common inheritance of human work are treated as, in part, common property. Chapter 9 will ask what that would mean in practice.

The giants against their rivals

The third fault line is between the largest firms and the smaller ones they have crowded out. In August 2024 a federal judge ruled that Google had illegally maintained a monopoly in internet search, the first major judgement of its kind against a technology company in a generation. The case had been supported by rival companies, publishers and advertisers who had their own reasons to want Google's power reduced. In September 2025 the judge declined to order Google to sell its Chrome browser, as the government had asked, but barred exclusive default-search deals and ordered the company to share some of its search data with competitors. In Europe, the Digital Markets Act, described in the first book, was pressed for by many smaller European firms as well as by consumer groups.

This split has a long history. The American anti-monopoly tradition of the late nineteenth and early twentieth centuries, associated with the lawyer Louis Brandeis, drew much of its strength from small businessmen, farmers and local merchants who feared the trusts as much as workers did. Antitrust is not a breakout for the 99 by itself: breaking up one giant can simply produce several smaller owners. But it is one of the few tools that reduces the concentration of ownership rather than only redistributing its income, and it is one where the 99 have natural allies among the owners.

The state against the owners

The fourth fault line is between the owners and the state itself, and here the record is the most sobering.

On 3 November 2020, two days before it was due to begin trading, the stock exchange in Shanghai suspended the initial public offering of Ant Group, the financial technology company controlled by Jack Ma. It would have been the largest share sale in history, raising around thirty-four billion dollars. Ma had recently criticised China's financial regulators in public. The suspension was followed by a broad campaign by the Chinese state against the power of its largest technology companies, and in 2021 President Xi Jinping placed "common prosperity" at the centre of policy. In one sense this was the owners being brought to heel by a force more powerful than themselves, as the boyars were by the tsars or the nobles by Louis XIV. In another, it was not a breakout for the 99 at all. The beneficiary of the split was the party-state, which acquired more control over the technology companies and the data they held; the ordinary Chinese worker gained no new right to organise, to vote or to dissent.

Russia offers a starker version. In the 1990s a small group of businessmen acquired much of the country's oil, metals and media in privatisations widely regarded as rigged. After 2000 the new president brought them to heel; in October 2003 Mikhail Khodorkovsky, then the richest man in Russia, was arrested, and his oil company was later broken up and its main assets taken over by a state-controlled firm. The owners had split from the state and lost. What replaced them was not the 99's rule but a new arrangement of owners loyal to the state.

The first book's account of Milovan Djilas's "new class" is the key to these cases. A split between the owners and the state helps the 99 only if the 99 have independent organisation and rights of their own. Otherwise the side that wins becomes the new owner. The Italian sociologist Vilfredo Pareto, writing more than a century ago, described history as a "graveyard of aristocracies": elites are replaced, but by other elites. A split at the top is an opening. It is not, by itself, a change in who owns the farm.

The rich who ask to pay

The fifth fault line is the most surprising, and it runs within the 0.01% itself. It divides the owners who defend their wealth at all costs from a minority who say, publicly, that they have too much.

In August 2011 Warren Buffett, then one of the three richest people in the world, wrote in the New York Times that his federal tax bill the year before had amounted to 17.4 per cent of his taxable income — less than any of the twenty other people who worked in his office, whose rates ranged from 33 to 41 per cent. His article was called "Stop Coddling the Super-Rich". In January 2024, at the World Economic Forum in Davos, some 260 millionaires and billionaires from seventeen countries signed a letter, organised by the group Patriotic Millionaires, under the title "Proud to Pay More", asking the world's political leaders to "tax us, the very richest in society".

One of the signatories went further. Marlene Engelhorn, a young Austrian who had inherited a large share of a fortune derived from the chemicals company BASF, said that she had done nothing to earn it and that Austria — which has no inheritance tax — should have taxed it. Since it had not, she invited fifty Austrians, chosen at random and weighted to reflect the country's population, to decide what to do with twenty-five million euros of it. The "Guter Rat" (Good Council) met over six weekends with facilitators and expert advice, and in June 2024 announced that the money would go to seventy-seven organisations, among them groups working on housing, health, integration, the environment and tax justice.

These gestures are small against the scale of the fortunes involved, and the first book would be quick to note that charity has often served the farm by letting the rich decide what the poor need. Engelhorn's experiment is interesting for this book precisely because she refused that role and gave the decision to a randomly chosen body of citizens. Chapter 10 will return to citizens' assemblies of this kind, as one way of governing the machine without creating a new class of shepherds. For now, the split matters because it breaks the owners' most important unity: their unity in defending their wealth. The political scientist Jeffrey Winters, in Oligarchy (2011), argued that what makes the very rich an oligarchy is not that they agree about everything, but that they share an interest in what he called "wealth defence" — and pay a whole industry of lawyers, accountants and lobbyists to provide it. Every owner who publicly defects from wealth defence makes it a little harder to present taxes on great fortunes as an attack on everyone.

The owners against their shepherds

The sixth fault line is not among the owners themselves but between the owners and the 1% who manage the farm for them. Chapter 3 promised to return to it, because AI may widen it more than anything else in this book.

The first book described the 1% as the shepherds: the scribes, stewards, clerks, lawyers and managers who keep the books and run the farm in return for a comfortable share of its surplus. Their loyalty has rarely been in doubt as long as the arrangement held. When it broke, the consequences were often dramatic. Of the deputies elected to the Third Estate of the French Estates-General in 1789, a large share were lawyers; Robespierre was one. The leaders of the revolutions of 1848, of the anti-colonial movements of the twentieth century and of many workers' parties were teachers, journalists, junior officials and clerks — educated people who had expected a place in the order and found it closed to them. Peter Turchin, whose theory of "elite overproduction" Chapter 3 introduced, argues that societies become unstable when they produce many more aspiring members of the elite than there are elite positions, and that the frustrated aspirants become "counter-elites" who lead the challenge to the established order.

Chapter 3 noted early evidence that AI is eroding the entry-level jobs through which young graduates join the professional class, and Chapter 4 found software engineers, journalists and academics beginning to organise as workers. If the owners of AI can do without much of their managing class — which is, after all, the third element of the Federation hypothesis — then the owners and their shepherds will have fallen out over the most basic question of all: whether the shepherds are still needed.

That split could go either way for the 99. Displaced shepherds could become allies, bringing to the scattered the organising skills, legal knowledge and political connections that the first book found in so many past breakouts. Or they could become the leaders of movements that promise to restore their own lost status at the expense of others — Turchin's counter-elites, who have historically led reactionary as well as radical causes. Which way they turn will depend partly on whether the 99 offer them a place: whether a movement of the scattered treats the displaced lawyer and the redundant analyst as fellow workers, or as former overseers getting what they deserve.

The owners' offer

There is one more division to consider, and it may be the most consequential of all. It is not a split between factions so much as a split within the most powerful faction: the owners of AI, some of whom have begun to offer the 99 a share.

In March 2021 Sam Altman, the head of OpenAI, published an essay called "Moore's Law for Everything". He predicted that AI would generate enormous wealth while reducing the value of much human labour, and proposed an "American Equity Fund": large companies would pay a tax of a few per cent of their value each year, partly in their own shares, and land would be taxed similarly, with the proceeds distributed to every adult citizen as a dividend. Altman also helped fund one of the largest studies of unconditional cash payments ever carried out in the United States. From 2020 to 2023 the research organisation OpenResearch gave a thousand people in Illinois and Texas $1,000 a month with no conditions, and compared them with two thousand people who received $50 a month. The results, published in 2024, found that recipients spent more on food, rent and transport, were somewhat more likely to look for work, and on average did somewhat less paid work — about an hour and a quarter less each week, in the first published results; the payments did not, by themselves, transform their health or long-term finances.

Read with this book's design in mind, Altman's proposal is remarkable. A fund that turns part of the ownership of the largest companies into a dividend for every citizen comes close to the fourth element of the Federation hypothesis: every person a shareholder. Other technology leaders have spoken in similar terms of a universal basic income, or even a "universal high income", as AI advances.

The first book offers a reason for caution, and it is not cynicism about any individual's motives. In the 1880s Otto von Bismarck introduced the first state health, accident and old-age insurance in the world, partly to undercut a growing socialist movement. The first book observed that the farm's concessions, as so often, were made to keep it standing. An owner who offers a dividend from the machines while keeping the machines is making a similar offer: a floor beneath the herd, with the fence left where it is. The question the design must ask of every such offer is not whether it is generous but who controls it. A dividend that the owners grant can be cut by the owners. A dividend that the 99 own, and govern, cannot.

There is also a deeper difference between the owners' offer and the design under test. Altman's essay imagined a world in which AI makes most human labour unnecessary, so the dividend becomes the main income of most people. This book assumes the opposite: that AI meets a wall, that human work — creativity above all — remains necessary, and that reward should be weighted by contribution, with a dividend as the floor rather than the whole. A society in which the many live on a dividend paid by the owners of the machines, and contribute nothing the owners need, is not the Federation. In the first book's terms it is a herd that is fed and no longer needed — the most dangerous position the 99 could be in, because a farm that no longer needs its herd has no reason to keep feeding it. Chapter 8 will return to this danger.

Public machines from a split

The owners' offer has a mirror image, and it too comes from a split: the one between the American and Chinese companies that dominate AI and the governments of almost everywhere else.

Many governments have come to fear depending for something as central as AI on a handful of foreign firms. In September 2025 two Swiss federal universities, EPFL and ETH Zurich, together with the Swiss national supercomputing centre, released Apertus, a large language model built with public money on a public supercomputer, whose design, weights and training data were all published openly. It is a small model by the standards of the largest private systems. But it is a working example of something this book's design requires: a capable AI system built and owned in public, open to inspection by anyone. Other governments, from the European Union to India, have announced public investments in computing power and models of their own.

The motive in most of these cases is national sovereignty, not the 99's ownership of the machines, and a model owned by a state is not the same as one owned by its citizens: the state, as the Chinese and Russian cases showed, can be the new owner. But the split between the AI giants and the governments that fear them is creating, for the first time, a constituency among the powerful for public AI. That is exactly the kind of opening the first book found the 99 using in the past.

Using a split without being used

What does the record suggest about how the 99 could use these splits? Four principles emerge, and each rests on the first book's evidence.

Organise first. A split among the owners helps the 99 only if they are organised enough to take advantage of it. The Ten Hours Act needed the short-time committees; the New Deal needed the unions. Where the 99 are unorganised, a split at the top is simply a fight between owners, settled between them. Chapter 4's conclusion — that the scattered can be organised, but only with deliberate effort and new forms — is the precondition of everything in this chapter.

Ask for institutions, not favours. The People's Budget lasted because it was followed by the Parliament Act, which permanently changed the balance of power. The Bismarckian pensions lasted because they became institutions that millions depended on. A concession that one faction of owners grants can be withdrawn when that faction no longer needs the 99. A concession built into law, institutions and the habits of a society is much harder to reverse. Chapter 6 will examine what has made gains last.

Watch the winner. When one faction of owners defeats another, the winner gains power. The Chinese and Russian cases show what happens when the 99 support, or merely watch, a state that defeats the owners without themselves gaining rights: the state becomes the new owner. The rule for the 99 is to demand, as the price of their support in any split, rights of their own that the winner cannot take away — to organise, to speak, to vote, to leave.

Expect the owners to reunite. The research of the political scientists Martin Gilens and Benjamin Page, who studied nearly eighteen hundred policy questions in the United States, suggested that the preferences of the most affluent Americans had a strong influence on which policies were adopted and those of the median citizen very little. Their method and conclusions are disputed. But the first book offers many examples of owners who fought each other bitterly and then united against a threat from below: the landowners and manufacturers who were at each other's throats over the Corn Laws sat in the same Parliament that rejected every Chartist petition, as the first book recorded. A split is an opening that closes. It must be used while it lasts.

The tests

This chapter tests the path to the Federation rather than its finished shape — whether the 99 can use divisions among the owners to win the design's elements — and, in particular, two elements that some owners are now offering of their own accord: the floor beneath everyone and every person a shareholder. It also tests the second element, the public ownership of the machines, against the split between the AI companies and the governments that fear them.

Precedent. Splits among the owners have repeatedly opened the way for gains by the 99, from the Ten Hours Act to the People's Budget and the New Deal. But the gains were largest and most lasting where the 99 were organised and where the split produced permanent institutions. Where a state defeated the owners without the 99 gaining rights of their own, as in China and Russia, the state became the new owner.

Evidence. The splits of the coming decades are already visible: over trade and migration, over AI and creative rights, between monopolies and their rivals, between states and owners, within the very rich over taxing great fortunes, and between the owners and a professional class whose entry-level work AI is beginning to absorb. Evidence on cash transfers, including the OpenResearch study, suggests that a floor can be paid without people ceasing to work, though the effects on work are not zero and the long-term effects of a universal dividend are unknown.

Capture. The owners' offer of a dividend, without public ownership or public control, is a classic form of capture: a feed that keeps the farm standing. The state's defeat of the owners, without rights for the 99, is another. And public AI built for national sovereignty can be captured by the state that builds it.

Exit. A floor that does not depend on any particular employer or owner strengthens everyone's right to refuse. But a floor that the owners grant, and that people depend on without contributing anything the owners need, weakens it; those who are fed but not needed have no leverage left.

The ledger

What survived. The floor and the dividend survive — not because some owners now propose them, but because the evidence suggests they can be paid without ending work, and because splits among the owners make them politically possible. Public ownership of the machines survives as a possibility, with working examples of publicly built AI emerging from the split between the AI giants and the governments that fear them.

What had to change. The design must specify who owns and governs the dividend. A dividend paid at the discretion of the owners of the machines, or of a state that has defeated them, is not the Federation's floor but a feed. The dividend must come from ownership the 99 hold in common and govern through institutions independent of any faction — including the state. Public ownership must mean more than ownership by a government: the cases of China and Russia show that a state that takes the machines from the owners can become the new owner. Finally, the dividend must sit beside reward for contribution, not replace it. A society that lives on a dividend while its human contribution is no longer needed has lost its leverage, which is why this book's assumption that human work remains necessary is not a detail of the design but one of its foundations.

What remains unknown. Which splits will deepen and which will close. Whether the conflict between AI companies and the owners of creative rights will be settled in a way that leaves creators with lasting rights, or simply divides the spoils between two kinds of owner. Whether publicly built AI can keep up with the private systems. Whether the shepherds displaced by AI become allies of the 99 or leaders of a movement to restore their own status. And whether the very rich who ask to be taxed are a passing sign of conscience or the beginning of a real division within the 0.01%.

Bridge: locking it in

Every split in this chapter is temporary. The Tory squires who supported the Ten Hours Act went back to standing with the manufacturers against the Chartists; the bankers who backed the New Deal drifted to the right again when it suited them. What made the gains of those moments last was not the goodwill of the owners who had briefly sided with the 99, but what was built while the opening was there: factory inspectors with the power to enforce the law, a Parliament Act that permanently weakened the Lords, a social security system that millions came to rely on. The next chapter asks what has made gains last since 2008, and what has not — from taxes on wealth and inheritance to antitrust, data rights, constitutional limits and the long, unfinished struggle against tax havens.