We understand, almost intuitively, that geography is not an afterthought to politics, but the substrate on which political life is built. The flat, fertile land along the Hudson River enabled large estates and semifeudal tenant farming in the 18th and 19th centuries, which in turn led to hierarchical systems of local government. The stony hills of Vermont, by contrast,

discouraged such consolidation and lent themselves to smallholder farming, Congregational churches, and democratic wrangling through town meetings. The geographies themselves didn’t have meaning, but they were not side issues: They created the material conditions that made it easier for certain systems to flourish while others faltered.

The same is true for information technology. And yet too often, we treat IT as abstract and neutral, obscuring our ability to see power itself and to fight the necessary battles we need to govern ourselves.

A decade ago, most businesses stored their data on-site; today the majority of all business data is stored off-site. The same is true for personal data—remember when your e-mail was on your computer, not “out there”? The amount spent on off-site data storage could soon approach $1 trillion (we are just shy of $700 billion already).

While the word cloud suggests something ephemeral, the reality is as physical as dirt. Data stored on the cloud is physically located on servers, typically at large data centers, or server farms, which require enormous amounts of energy to maintain. Just three companies—Amazon, Microsoft, and Google—control over two-thirds of this critically important market.

A recent report, “Engineering the Cloud Commons,” by the Open Markets Institute shows how the big three have used a series of aggressive tools to protect their cloud dominance. They use opaque pricing, charge different amounts to different customers for the same service, make it unnecessarily difficult to switch providers, and exploit their power on other platforms to push users toward their own services. They are also leveraging this dominance to foreclose competition in one of the most dangerously concentrated technologies: artificial intelligence.

Current Issue

July/August 2025 Issue

These anticompetitive techniques are the kinds of abuses that anti-­monopoly law is designed to prevent. And they are particularly dangerous with the cloud because, as the report details, there are “innate characteristics of the market”—massive barriers to entry, huge capital requirements—that already make concentration very likely, though not inevitable.

By treating the cloud as neutral, we have exposed ourselves to great danger. Concentration significantly increases the risk of widespread system failure. If just one of the big three tech giants collapses, societal mayhem could follow: hospital closures, data breaches, business disasters. Even without a collapse, we have put ourselves in a position of radical dependency greater than that of society on the big banks. Not only has Big Tech become too big and too embedded to fail; the firms know they can use that power to dictate policy. Some of the democratic risks are more direct: Amazon, Google, and Micro­soft have already shown themselves willing to suppress certain speech and amplify others. Ownership of the cloud is also ownership over speech-related data.

The Open Markets report proposes an immediate remedy: Cloud providers should be required to operate in the public interest, offering fair and equal access to infrastructure, with public oversight and rate regulation. Second, since such essential services cannot coexist with monopoly ownership, the big three will have to divest their interests.

Reporting on AI often treats it as an abstract force, separate from the physical architecture that makes it possible. We talk about AI as “doing things,” with no reference to the physical infrastructure it requires or the corporate systems it serves. We erase the layers. We treat models, chips, and data centers as immaterial when, in fact, they are as structured, owned, and governed as a factory or a plantation.

We also treat technology, and by extension the architecture of information, as neutral. The novelist William Gibson famously observed, “Technologies are morally neutral until we apply them.” Barack Obama repeatedly described digital technologies as “tools,” neutral in themselves, whose effects depend on their use. Noam Chomsky made a similar argument: “Technology is basically neutral. It is like a hammer. The hammer doesn’t care whether you use it to build a house or…to crush someone’s skull.”

Both abstraction and neutrality perform a dangerous social function: They obscure power. “Established power always seeks to obscure or outright prohibit thinking about the organization that upholds their power,” the writer Joe Costello recently observed, writing about information systems. The organization of information is one of the most consequential forms of power in modern society, giving those in power a potent incentive to use the language of neutrality and abstraction to hide it.

Information lives in supply chains, fiber-­optic cables, content-moderation protocols, chip designs, algorithmic hierarchies, and server farms in Virginia and Arizona. You don’t have to understand how all of these relate technically to understand that allowing corporations to both own the infrastructure and run the services that depend on it sets up an untenable conflict of interest at the heart of our society.

As the Open Markets report documents, the big three corporations that control the infrastructure also dominate AI, advertising, retail, and government contracting. They are getting billions of dollars of privileged access to the most valuable input of the 21st-century economy—information—which they then use to control us. A functioning democracy cannot depend on Microsoft’s goodwill.

Today, the most consequential terrain of political and economic power is no longer land. It is information. And just like geography, the structures that shape and control information shape the conditions of freedom.