The Shadow Side of “AI Inside” Everything
The Shadow Side of the AI Boom: The Physical Infrastructure of Data Centers
The AI boom is upon us, reshaping industries and society in profound ways. With staggering investments like $500 billion from OpenAI, Microsoft, and Oracle into “Stargate” data centers and $65 billion from Meta to construct a facility nearly the size of Manhattan, the scale of this revolution is unprecedented. As AI models become larger, more complex, and more embedded in our daily lives, the hidden costs of supporting this digital infrastructure are mounting, revealing a shadow side we cannot afford to ignore.
At first glance, the AI revolution appears to be an ethereal shift—a leap into the virtual. But the truth is far more grounded, even gritty. Behind every text-based chatbot, every recommendation algorithm, and every autonomous system lies a vast network of physical resources. The unseen machinery of the AI era—the tech, the people, the land, the power, and the water—is growing at an alarming rate. These elements form the backbone of our digital era, but they also present significant challenges for sustainability, ethics, and resilience.
The Unseen Physicality of AI
AI factories and data centers—monolithic hubs of servers and computing power—are the physical embodiment of the digital world. Far from existing in a cloud-like vacuum, these centers are rooted in land-intensive, power-hungry, and water-dependent realities. For example, Meta’s $65 billion data center project will span nearly the size of Manhattan, a stark reminder of the physical footprint required to sustain AI’s exponential growth. Meanwhile, companies like OpenAI and Microsoft are committing half a trillion dollars to build out “Stargate” data centers, designed to handle the immense computational demands of generative AI models.
Yet, as we scale up, we encounter a critical bottleneck: power. The energy demands of AI infrastructure are immense. So immense, in fact, that areas across the United States now face power supply shortages with wait times of 5 to 7 years to connect new facilities to the grid. In response, drastic measures are being taken—including the reactivation of Three Mile Island, a nuclear power plant infamous for its near-catastrophic meltdown in 1979. Such decisions highlight the lengths we are willing to go to sustain the AI boom, but they also underscore the fragility of our energy — and digital — systems.
The Human and Environmental Costs
The shadow side of AI is not only about infrastructure; it’s also about the human and environmental toll. Data centers consume massive amounts of water to cool their servers, which can put pressure on local water supplies, particularly in drought-prone regions. The labor required to build and maintain these facilities—from construction workers to technicians—often goes unnoticed, yet their work is critical to the seamless operation of AI systems. Who will build all these data centers, as Baby Boomers age out, and a new generation takes over that lacks physical experience making, building, or fixing things? Plus, the demographic cliff where less kids are being born since the 1970s, spells even fewer workers available to build these data centers.
Moreover, the environmental costs are staggering. Internet queries utilizing artificial intelligence require about 10x the electricity of traditional internet searches, which means a need of more power. Data centers are projected to double their energy demands, to 1,065 TWh by 2030 if current trends continue. The reliance on fossil fuels in many regions means that the carbon footprint of AI is far from negligible. Even as companies make strides toward renewable energy adoption, the scale of demand often outpaces supply. Many companies committed to sustainable computing are having to leave their aspirations behind in the scramble to keep up with the AI race: A 2024 Google sustainability report admitted their greenhouse gas emissions rose last year by 13%, while electricity usage grew by 17%.
The Ethics of Growth
As the AI boom accelerates, we must grapple with the ethical implications of its physicality. Who benefits from these investments, and who bears the costs? Communities living near data centers or energy facilities often face increased pollution, higher energy costs, and limited access to water. Meanwhile, the fruits of AI—from advanced healthcare to smart cities—often disproportionately benefit wealthier, urban populations.
We also face questions of land use. The vast tracts of land required for data centers and their supporting infrastructure often compete with other critical needs, such as housing, agriculture, and green spaces. How do we balance these competing priorities in a way that ensures fairness and sustainability?
Do We Want “AI Inside” Everything?
The AI boom represents an incredible opportunity for innovation and progress, but it also demands a reckoning with its physical and social costs. Companies at the forefront of AI development—from OpenAI and Microsoft to Meta—must prioritize transparency about the resource demands of their operations. Policymakers, too, have a critical role to play in regulating and incentivizing sustainable practices, from renewable energy adoption to water conservation measures.
Finally, we as a society must ask the hard questions about how we want to live with data centers and what happens if – or when – it all stops. We may also want to ask ourselves: Do we really want, or need – “AI Inside” everything? By addressing the shadow side of AI now, we can build a future where technology serves not only profit but also people and the planet.