Data Center Frenzy: AI, Energy and Water Strain in the Global Megaproject

Summary

The B1M examines one of the fastest growing construction booms the world has ever seen: data centers. The video explains how demand is climbing with AI, where large clusters are located, and the challenge of supplying power and water. It also surveys potential responses from industry and policy makers, including new cooling approaches, grid upgrades, renewables, and even nuclear options.

• 3,4% of global electricity consumption is expected to come from data centers.

Introduction: The data center boom and AI demand

The video opens by describing data centers as a central, accelerating infrastructure in modern life, driven by AI and the cloud. The B1M treats this as a unique megaproject: not just a single building but a network of interconnected facilities that enable online shopping, streaming, and digital communication, while also placing unprecedented demands on power and water resources. As data centers expand, the infrastructure needs expand with them, creating a feedback loop between AI scale and energy use.

In this context, the data center phenomenon is framed as both indispensable and challenging. The speaker notes that a significant fraction of global electricity will be consumed by data centers in coming years, underscoring the urgency of managing electricity, water, and heat generation as AI workloads grow.

"data center industry is actually going to be responsible for 3,4% of global electricity consumption." - The B1M

Geography and clusters: where data centers live

The video highlights where these megastructures tend to cluster. The United States hosts a large share, with Virginia emerging as a hotspot, followed by Texas and California. Europe also hosts major clusters in the UK and Germany, with distribution across many other regions to support ubiquitous digital services. The location choices hinge on fiber connectivity, water availability, and, crucially, high-capacity power grids capable of supporting hundreds of megawatts. The goal is to find spaces with capacity for expansion, security against extreme weather, and access to reliable cooling and energy sources.

"By 2030, data centers are set to be consuming more electricity than the entirety of Japan." - The B1M

Energy, water, and AI: the core challenge

As AI accelerates, data centers become larger and more power-hungry. An AI data center today can be hundreds of megawatts, potentially over 1 gigawatt, concentrating large energy and water demands in a small footprint. The video uses capacity growth charts to illustrate how installed capacity has soared, and it compares the scale to major national energy installations to convey the magnitude of the load. Water use is discussed as a critical concern because cooling often requires substantial water intake from municipal supplies, stressing public water systems, especially in regions with limited surplus capacity.

The rise of AI is portrayed as the primary driver of demand growth, with AI processing constituting a sizable piece of data center electricity use. The video explains that the AI workload has fueled a shift toward much larger facilities, which in turn puts pressure on local grids and water infrastructure.

"AI systems were responsible for about 20% of total data center electricity consumption by the end of 2024." - The B1M

Technology and strategies for cooling and energy balance

The B1M discusses traditional cooling methods and the rise of liquid cooling as AI intensity grows. Traditional air cooling, often exhausted through the top or floor, is energy-intensive, while liquid cooling offers a more space-efficient and energy-efficient alternative, albeit with greater water demands and sourcing challenges. The video also mentions alternative approaches such as building data centers underground in mines or bunkers for enhanced cooling and security, and it notes that grid upgrades and on-site renewables are increasingly prioritized by hyperscalers to reduce energy costs and carbon footprints.

Mitigation, policy, and the future path

In response to these pressures, tech companies are exploring a mix of strategies, including: liquid cooling adoption, greater on-site renewable generation, and funding for grid upgrades to prevent higher energy price burdens on communities; partnerships with nuclear energy or exploration of advanced reactor concepts like small modular reactors; and unconventional approaches such as subterranean facilities to improve cooling and security. The documentary suggests that not every company can or will pursue such bold options, but many will need to adopt new technologies and processes to manage capacity, energy, and water use effectively. It also emphasizes the role of digital-twin simulations and AI-enabled design optimization to de-risk infrastructure decisions and achieve high performance with lower energy and water footprints.

IES presents itself as a tool provider and consultant, offering physics-based building performance, simulation, and digital twins to help hyperscalers design and operate AI-ready facilities efficiently while complying with grid constraints and environmental regulations. The case studies show measurable outcomes, including significantly reduced water use and cooling energy, and improved PUE metrics in retrofit projects.

Conclusion: balancing necessity with sustainability

The video concludes with a candid look at the tension between a growing need for data center capacity and the obligation to protect water resources and energy affordability for communities. It argues that megaprojects will continue growing, but the path forward must address environmental and social concerns through careful site selection, cooling innovations, on-site energy strategies, and robust grid planning. The final call is for proactive, multi-stakeholder approaches and the adoption of AI-driven design tools to make data centers future-proof and responsible.