👤 About the Author
Jill Lepore
Harvard University
Jill Lepore (born August 27, 1966) is an American historian, author, and journalist. She serves as the David Woods Kemper '41 Professor of American History at Harvard University, a professor of law at Harvard Law School, and a long-time staff writer for The New Yorker. Lepore specializes in American political, legal, and cultural history, with research focusing on the history of technology, evidence, and democratic institutions. Her scholarship and literary nonfiction have earned major acclaim; her book The Name of War won the Bancroft Prize, Book of Ages was a National Book Award finalist, and These Truths provided an influential single-volume history of the United States. Through her books, essays, and public projects, Lepore has made essential contributions to understanding American history and liberal democracy.
Introduction
The contemporary narrative surrounding artificial intelligence often relies on the myth of dematerialization. The term Cloud computing suggests that data and computation exist in an ethereal space, independent of physicality.
This article deconstructs that illusion. It argues that AI is, in reality, a new form of heavy industry requiring vast amounts of material and energy resources. Readers will discover how the material anatomy of technology influences global power dynamics and national sovereignty.
The Cloud as Physical Industrial Infrastructure
Artificial intelligence and the cloud are not immaterial. They consist of a dense network of buildings, processors, fiber optic cables, and electrical substations. Every response from a language model is a physical flow of electrons through semiconductors and the emission of heat.
The materiality of AI directly shapes power structures. Whoever controls computational sovereignty possesses a strategic resource akin to steel or oil in the industrial age. Examples include clusters such as Meta's Prometheus or Hyperion, whose power requirements are already measured in gigawatts.
Data Centers as Modern Heavy Industry Factories
AI has ceased to be a mere digital service. It has become an industrial process of extreme power density, where the end product is computation. Data centers function like factories: they require massive capital, cooling systems, and stable energy supplies.
This shift alters the state-corporate relationship, creating a hybrid public-private complex. States are becoming dependent on private infrastructure providers, risking the privatization of state capacity. If an administration cannot independently audit AI models, it loses actual institutional sovereignty.
Conflicts of Interest in Energy and Water Resources
Big Tech investments place immense pressure on local ecosystems. Data centers compete with residents for access to electricity and the water necessary for cooling. This problem is territorial rather than global in nature.
In the name of geopolitical advantage, governments—such as in the US via the AI Action Plan—are accelerating investment procedures at the expense of environmental regulations. Physical constraints, such as access to water and energy, are becoming political issues. The decision to allocate these resources between "tomatoes and data centers" is, in essence, a dispute over the common good.
Conclusion
Every utopian vision of transcending matter and achieving digital immortality eventually collapses against the laws of physics. AI does not nullify the geopolitics of resources; rather, it adds new elements to it: chips and gigawatts.
We may imagine intelligence as an ethereal entity floating in the cloud. However, the truth is far more grounded: every fantasy of a machine ruling the world requires a transformer and access to water.
Frequently Asked Questions
Are artificial intelligence and cloud computing truly immaterial?
No, artificial intelligence and cloud computing are not immaterial; their functioning relies on a specific physical infrastructure. They require networks of buildings, processors, cables, cooling systems, and vast resources of electricity and water.
1. Is artificial intelligence still just a digital service, or is it already something more in terms of resource requirements?
2. Artificial intelligence goes beyond simple digitalization, generating demand for a new class of infrastructure characterized by extreme computational density and power, measured in gigawatts. AI data centers resemble heavy industry plants, requiring massive capital as well as advanced power and cooling systems.
3. What impact do the investments of big tech companies have on local energy and water resources?
4. Tech company investments can accelerate decarbonization and fund new generation capacities; however, they allow these firms to shape energy infrastructure according to their own needs. In terms of water resources, they lead to a significant increase in direct water consumption for cooling and indirect consumption within the energy system, creating territorial competition for this local resource.
5. How does the physical infrastructure of AI affect local communities and the political sovereignty of states?
6. AI infrastructure burdens local communities with increased demand for water and energy, creating so-called 'computational enclaves' that do not always generate a significant number of jobs. In a political dimension, the strong concentration of chip production and cloud services leads to state dependence on external providers, making computational sovereignty a key element of a state's actual capability.
7. How does control over AI infrastructure change the relationship between the state and big tech corporations?
8. Control over AI infrastructure transforms the state-corporation relationship into a symbiotic, hybrid public-private complex in which both parties are interdependent. States need corporate technology and infrastructure to achieve strategic goals, while companies rely on state resources such as energy or legal security. However, this creates a risk of losing institutional sovereignty if the state becomes epistemically dependent on providers and loses the ability to independently verify AI systems.
9. How are governments (for example, the USA) adapting laws and environmental regulations to accelerate the construction of AI infrastructure in the name of geopolitical advantage?
10. Governments, for example the USA under the AI Action Plan, are accelerating the construction of AI infrastructure by reducing regulatory burdens and applying categorical exclusions in environmental procedures (e.g., NEPA). Actions also include modifying regulations based on the Clean Air Act and Clean Water Act, as well as speeding up environmental reviews and facilitating the acquisition of water permits.
Can private AI infrastructure have political and strategic significance for a state?
Yes, private AI infrastructure has political and strategic significance because computing power is becoming an element of a state's potential, and control over semiconductors a tool of foreign policy. When private systems become essential intermediaries in public functions, their importance grows regardless of formal ownership.
Why are the physical constraints of AI (energy, water) a political issue rather than just a technical one?
The physical constraints of AI are a political issue because they concern the conflict over the right to establish the hierarchy of use for limited resources and territory. Deciding how to allocate energy or water among different social needs requires a democratic choice of values that cannot be reduced to a single technical metric.
