The Rise of Techno-feudalism

The Wars of the Roses and the looming crisis of AI politics

A couple of months ago, as everyone in DC knows and almost no one outside cares about, there was a royal dust-up between the leading AI company Anthropic and the Department of War. The debate, at least in public, concerned whether or not the Department could use Claude to engage in mass surveillance of American citizens or autonomous killing on the battlefield. Anthropic insisted on drawing redlines around these uses in any military contract, while the DoW insisted that it alone had discretion, within the constraints of the law, about how to use technologies provided by the private sector. The showdown culminated in a dramatic and draconian attempt by the government to commit “corporate murder” on Anthropic by banning any government agency or contractor from doing business with them.

I found the whole episode troubling on many levels, and had good friends on both sides of the dispute. On the one hand, Anthropic’s redlines were clearly reasonable and righteous, and the DoW’s response came across like one giant (and almost certainly illegal) tantrum. On the other hand, there was clearly something rather disturbing about the precedent: a private company, possessing technology essential for national security, able to dictate terms to the legally-accountable representatives of the people on how that technology could or should be used. Thankfully in this case, Anthropic was using its extraordinary leverage for good, but what guarantee was there that it, or its less scrupulous rivals, would continue to do so?

These misgivings have only intensified in recent weeks, as the government has been forced to effectively concede defeat, and come hat in hand to Anthropic, pleading to be allowed to use the very models that it so recently sought to ban. What changed were not court rulings (which are currently divided, with litigation ongoing), but the technology. This has continued to advance at an exponential—indeed a super-exponential—rate, making it clear that Anthropic, not the US Department of War, holds more and stronger cards. The government, in other words, now needs Anthropic more than Anthropic needs the government.

Again, thankfully the company benefiting from this remarkable power asymmetry is also a remarkably conscientious company that clearly sees itself attempting to operate in the public interest (as it understands that interest!), but this is not particularly reassuring. A couple weeks ago while I was in SF, I participated in a RAND Corporation wargame simulation, “The Day After AGI,” designed to help participants think about the power dynamics and national security implications of the emergence of advanced AI capabilities. Although so much of the discourse now is focused around “beating China,” one of the big takeaways from the game was that there was not all that much difference between China getting “AGI” capabilities, and a rogue private company, even if ostensibly US-based, getting such capabilities. Such a company might well try to dictate terms to the federal government. Faced with this situation, the government might well respond by trying to nationalize the company, and it’s anyone’s bet who would actually “win” in such a scenario. But one thing we can be sure of is that the American people would lose; indeed, we already have.

The appropriate term for our current political situation, I deem, is techno-feudalism. I have not, I must confess, read Yanis Varoufakis’s 2023 book by that title, but I have been thinking and occasionally writing about the concept for several years myself. While lots of people have thrown around the metaphor of “feudalism” for the past decade to describe our fracturing politico-economic system, few have given it the close attention or historical analysis it deserves. I attempted to do this in a 2021 essay for American Affairs, “‘Over-Mighty Subjects’: Big Tech and the Logic of Feudalism,” and these past weeks, I found my memory jogged and decided to give it another look. It is always heartening when one returns to one’s writing after several years and finds that it still holds up; even better if it feels prescient. So it was here. In a spirit of shameless self-promotion, I’d encourage you to read it and see for yourselves; but it’s a long one, so I’ll just highlight some key themes here.

The conditions I sought to diagnose in that essay have only worsened since. I opened it by highlighting the staggering and unprecedented share of the economy that the five largest tech companies—then Apple, Google, Amazon, Meta, and Microsoft—had seized: $7.7 trillion of market value, or 16% of the entire US stock market. Today the top five (now including NVIDIA; Meta has mercifully fallen to 10th) account for a staggering $21 trillion, or 27.5% of the market. Then, the top two companies by revenue (Amazon and Apple) raked in around $300 and $400 billion respectively; today, Amazon alone brings in $750 billion.

But even these numbing numbers mask the extraordinary concentration of power in the tech sector, for at least all of these are publicly traded companies, ostensibly answerable to corporate boards and major shareholders. At time of writing, there are now three $1 trillion+ privately-held companies: OpenAI, Anthropic, and SpaceX/xAI, each of which seems quite plausibly undervalued if current AI capability and adoption curves continue for much longer at all. And as sole controllers of the world’s most powerful technology, capable of mass destruction in the wrong hands, they wield extraordinary political power.

To understand feudalism—past, present, and future—we must understand three things:

1. Power flows and concentrates; if the concentration of power breaks down in one place, it will reconsolidate around new loci.

2. Economic power, political power, and juridical power cannot remain neatly separated.

3. Political stability and peace requires one clearly supreme concentration of power that can hold all lesser concentrations in check.

Each of these these phenomena can be clearly illustrated in the confluence of circumstances that culminated in England’s disastrous fifteenth-century civil war, The Wars of the Roses, which caused such a catastrophic bloodletting of the nobility that it effectively ended feudalism in England 150 years before it ended on the Continent.

“There may no greater peril grow to a prince, than to have a subject of equal power to himself”—Sir John Fortescue

As I write in the essay,

“For the purposes of this discussion, I will define “feudalism” as a polity in which political power is vested not in publicly-recognized representatives of the whole, but in large property-holders exercising effective control over territories or sectors of the polity. This was the dominant form of political order in much of medieval Europe. In England, however, from the reign of Edward I (1272–99) until Henry V (1414–22) feudalism had been largely superseded by Europe’s first effective national state. Although even the most powerful kings still relied heavily on the fiscal resources, soldiers, and police powers of the barons, Plantagenet and early Lancastrian England witnessed both a theoretical and practical consolidation of executive, legislative, and judicial authority in the institution of the Crown-in-Parliament and the King’s common law courts. But state-building is not a one-way street. Inept leadership or fiscal irresponsibility at the center can result in the rapid flow of power back to the peripheries, with devastating effects on political stability.”

The standard schoolbook story of the Wars of the Roses (assuming schoolchildren still learn about them at all!) is that they were a dynastic quarrel between the House of York and the House of Lancaster, the result of an unfortunate ambiguity in the proper genealogical line of succession to the English throne. In reality, though, such ambiguities were hardly unusual, and in fact, far more dubious royal claimants would later silence their rivals with little difficulty in the 17th and 18th centuries when the Crown was stronger. The real problem was that Henry VI was a very weak king, and the vacuum of power in Westminster allowed power to consolidate elsewhere, creating an unstable situation where powerful nobles could serve as kingmakers and effective rulers. As I write:

“The new state of affairs was epitomized by the career of Richard Neville, sixteenth Earl of Warwick, popularly known as “Warwick the King-maker.” By 1456 he had accumulated an enviable list of titles, immense landed estates, and the politically crucial appointment of constable of Calais, with England’s largest standing army at his disposal. His support of the Duke of York in the years that followed was decisive in propelling the Yorkist Edward IV to the throne in 1461, a success that still further increased his own power. By 1462 he had added the earldom of Salisbury and the duchy of Lancaster to his portfolio and was the richest man in England besides the king. His family members crowded into key positions and he came to dominate English diplomatic relations on the continent, so much so that the French governor of Abbeville, writing to King Louis XI, joked, “The English have two rulers, M. de Warwick and another whose name I have forgotten.” When Edward IV made the mistake of snubbing Warwick, the King-maker decided to throw his lot in with the Lancastrian exiles, and succeeded in briefly restoring Henry VI to the throne before dying on the battlefield of Barnet in 1471.”

Sir John Fortescue, who had a front-row seat to these troubles as Chief Justice of the King’s Bench, would write in his The Governance of England, “there may no greater peril grow to a prince, than to have a subject of equal power to himself.” The people, he saw, would flock to such a prince as their true protector and the king would soon be king in name only. The crisis would prompt Fortescue to sketch out a proposal for a vigorous centralized constitutional monarchy, which would come to provide the blueprint for the Tudor state, the emergence of Parliament, and ultimately the American republic.

In the case of Henry VI, as today, much of the power vacuum at the center was economic: the Crown managed its finances poorly, and soon found itself unable to reliably pay soldiers and officials. Thus it found itself outsourcing much of the critical work of government to reliable noblemen. These lords, like the Earl of Warwick, were thus able to amass tremendous economic power in the form of land, which enabled them to feed and employ enormous retinues. Adam Smith remarked in The Wealth of Nations that Warwick was said to have fed thirty thousand men at his tables each day. As I write in the essay, “having established effective fiscal independence and a dependent client base, the barons increasingly began to exercise sovereign powers within their own domain—they could defy the law with impunity and dispense their own rough justice, undermining the central administration of the common law built up over the previous three centuries.”

It is not hard to see the same pattern repeated today, as a fiscally irresponsible central government finds itself unable to consistently fund critical agencies or hire and retain top talent at competitive rates, and begins outsourcing more and more of the work of government to private companies. These companies—today almost entirely advanced technology companies—then amass extraordinary wealth and create huge populations of economic dependents who find themselves reliant upon the companies for their daily bread—which increasingly gives them effective political power as well as economic power. Along with this political power comes the pressure to exercise juridical power by dispensing justice within their digital domains—something I explored in my first ever project for American Compass, “We Peasants of the Metaverse.”

Finally, as I write in the American Affairs piece, “These quasi-sovereigns dared to play the role of kingmakers, bestowing titular sovereignty on whomever best advanced their own interests.” We see the same trend unfolding over the past few years, as the tech titans have come to exercise extraordinary influence over who gets elected and what laws can be passed or enforced. As power consolidates even further in the hands of the frontier AI companies, we are rapidly approaching a situation where Washington is weaker than Silicon Valley, and must come begging for scraps at its table. The humiliating spectacle of Henry VI, dragged around England in the train of the nobles he ostensibly rules over while they flatter him with titular authority, could soon be a preview of the American presidency. Does this mean that we too are headed for civil war? This feels like a melodramatic conclusion to reach, and few things are harder to imagine in contemporary America than a genuine internal military conflict. But there is no question that our situation has all the ingredients that generally lead to civil strife or at least profound political upheaval.

That said, all is not lost. For as I also write in the essay, we have been here before—or not far off at any rate. At the end of the 19th century, the age of the “robber barons” was in full swing, with unprecedented concentrations of corporate power in railroads, industry, and finance, while federal power in Washington remained weak and indecisive. Theodore Roosevelt’s extraordinary presidency managed to dramatically reduce the “feudal” power of the corporations, while dramatically strengthening the economic, moral, political, and juridical authority of a democratically accountable federal government. As I write in the essay:

“The key was to ensure a firm legal subordination of private to public power, using the power of the federal government—the only government large enough to control such titans—to bring their activities in line with the public good, rather than naïvely allowing them to self-regulate their vast domains. Crucial to this process was the legal reform that recognized railroads as common carriers—networks with public power that must serve public purposes, treating all customers fairly and equally. It also involved the consolidation of fiscal and political power at the federal level that would be sufficient to prevent a return of the cronyism that had allowed the industrial titans to bend government to their will. As Michael Lind has argued, the necessary political changes, largely delayed until the 1930s, marked a ‘third American republic’ that held sway until the deregulation craze of the 1980s and 1990s, which sowed the seeds for the new feudalism of our own day.”

Even our advanced state of techno-feudalism, then, is probably not yet irreversible, however hard it seems hard to imagine another such political renewal today. But as power continues to fracture and the storm clouds gather, we find ourselves waiting for a new, and doubtless very different, Bull Moose.