Amazon has amassed a kind of empire that most colonizers, not to mention entrepreneurs, could only dream of. It has achieved massive scale at conventional standards: The company boasts a healthy market capitalization of nearly $1.7 trillion dollars, buoyed by a recent massive increase in profit margins as the pandemic forced many into online purchases. It’s responsible for 40 percent of all US e-commerce and nearly 10 percent of online retail sales on the entire planet. Its true source of imperial majesty, though, might not be the hundreds of millions of packages it ships per month at all, but something far more ephemeral: its “vast empire of customer data.”

Maybe Amazon’s empire isn’t a group of executives that manage some important machines. Maybe Amazon’s empire is a machine. Our robot overlords are already here, and they came with articles of incorporation.

Welcome to the eighth and final entry of the series “How Much Could a Banana Republic Cost?,” where we’re trying to figure out who and what rules the world. The candidates so far have all been people or groups of people: The Big Green theories picked out solo monopolists and colluding plutocrats; the Big Guns theories chose political strongmen and outlaw mafiosos, and the first Big Graphs theory picked out nerdy technocrats.

This final Big Graphs theory is different. The “Terminator” theory holds that the world is run by a network of massive robots: corporations. This is similar to the Big Green theories, since it also targets the moneymakers and engines of global capitalism, but importantly different in that it targets the corporate entity as a whole rather than executives or investors within them. You’ll note that today’s global ruling machine is Amazon, the entity—not Jeff Bezos the plutocrat and hopeful space colonizer (as a Big Green theory might have it), or even “artificial intelligences” like Amazon’s Alexa.

Some worry about the rapid development of artificial intelligences, and rightly so: Use of AI in the automation of tasks associated with surveillance, information dissemination, and physical weapons systems poses a clear danger in the event that we create AI so sophisticated that it could “prevent us from replacing it or changing its preferences,” as philosopher Nick Bostrom warns. But philosopher Gabriele Contessa argues persuasively that the robot apocalypse is already happening, since we live in a world run by corporations… which are robots.

Contessa knows what you’re thinking: It’s not so much that corporations are people (unless you’re weird), but aren’t they at least controlled by people? But beyond the nuts and bolts, what makes something a robot is the ability to pursue goals independently of the will and supervision of a controller. And this is precisely how large corporations work. Once a corporation reaches a certain level of scale and complexity, robust top-down managerial control of the corporation’s behavior becomes literally impossible: Lower-level managers and employees and more limited artificial intelligences necessarily run without their input.

Whatever that threshold of scale is, Amazon is far beyond it. Its Amazon Web Services subsidiary alone controls more than 40 percent of global cloud infrastructure—a full third of the Internet runs on AWS. Amazon’s customers include massive private companies General Electric and Unilever, but also major government organizations like the US Central Intelligence Agency and 7,500 other government agencies. And while AWS punches above its weight in its contribution to company profits, it represents only 12 percent of Amazon’s overall revenue.

This is also to soft-pedal the problem. Even if the “managers” could control the whole show, who are they? The “managerial” level of the modern large company is itself is increasingly complex and difficult to map: Managers report to a board of directors, which reports to shareholders—which are often themselves companies, and those accordingly have the very same internal governance complications. They form together like Voltron into one large Russian nesting doll of buck-passing.

Economists Peter Nolan, Jin Zhang, and Chunhang Liu argue that we’re now seeing this on a planetary scale, that the world is increasingly reshaped by “systems integrators” consolidating power atop the global business food chain: “firms with powerful, globally recognised technologies.” Below them, smaller firms are restructured by a form of de facto “industrial policy”: The big companies pick winners and force changes to the entire supply chains that connect them to the broader global economy (much like a country’s government doing economic planning, except that nobody elected Jeff Bezos and friends).

This concern takes on an even darker hue in the context of our ongoing climate crisis. The worry about hypothetical artificial intelligences was that we might create one so powerful that it would act to prevent us from changing its preferences, all while leading us into ruin. What is the difference, really, between this hypothetical scenario and the present facts of the matter when it comes to climate crisis?

As journalists at Inside Climate News uncovered, companies like Exxon became so economically powerful that they were able to embark on largely successful planetary-scale misinformation campaigns to push naked climate denial and prevent us from changing their economic model to something less genocidal, beginning in the 1970s. Even after a sea change in public attitudes about the climate crisis a nearly half a century later, the fossil fuel industry has merely changed strategies for the prevention of public interference in their Prime Directive. They now attempting to greenwash their continued pollution instead of pushing more naked climate denial as in years past, while helping to seed draconian criminalization of climate protest in state legislatures. Royal Dutch Shell continued to fund anti-climate campaigns even after signing a pledge pinky-swearing that it would stop doing so.

This is a weird and outlandish take even by the standard of this series—should we really think we’re being ruled by robots? But as we’ve done throughout the series, we can use dollar amounts as a first-pass check on the plausibility of this way of thinking about who runs the world: It’s a way of estimating how much social resources would have to be immediately concentrated in a single cabal of people (or robots) for them to run everything else at a given scale.

If robots wanted to run the entire world, we can use Amazon’s market capitalization as a rough estimate of how much it would cost: roughly $1.7 trillion. This is a much larger aspiration, and accordingly a larger figure than the $200 million the United Fruit Company put up for the single banana republic of Guatemala or the $40 million that Petri Friedman put up to buy a Monopoly square’s worth of world domination in Honduras. But it is a much smaller figure than the titanic $130 trillion that a round table conference of the world’s biggest asset managers and corporations would represent. It’s also smaller than both of the estimates for rule by violence: Strongmen in control of formal state militaries jointly have roughly $2 trillion worth of resources at their disposal, and a Godfather in control of the entire criminal underworld would clock in at a little more with $2.2 trillion. By this measure, the Terminator theory is at least as plausible as many of the others. Of all the possibilities we’ve seen so far, I suspect that this answer has the most going for it.

However, I doubt that there’s a definitive answer to the question of “Who rules the world?” waiting in the wings. The sheer scale and scope of the various answers should humble us. As we’ve seen in this final installment, the complexity of a single large organization exceeds the managerial capacity of even the savviest groups of managers when sufficiently large and complex, and the world as a whole is far more complex than any corporation of any size. Our global power structure is likely a constantly shifting assemblage of all of the forces we’ve discussed here, dynamic equilibria emerging from the contest of the forces we’ve organized under the headings of Big Green, Big Guns, and Big Graphs. Circumstances and luck help to explain which, if any, of these forces are most prominent in a particular place and for a particular time, but none categorically lord over everything else.

Rather than thinking of Big Green, Big Guns, and Big Graphs as competing answers to a theoretical question about what the secret core of global political power is, we could instead see them as complementary descriptions of different terrains of political struggle. If there’s no “real” power structure among these, then our thinking about how to challenge our political structure can shift accordingly. Each of these are areas to vie for or concede as circumstances dictate, or to strategically play off of each other until other possibilities emerge. We may have to make peace with plodding through ground that is shifting beneath us.