The rapid expansion of AI-driven data centers transforms what once seemed like invisible, back-end digital infrastructure into a highly visible, very localized challenge. According to Forrester’s US Tech Market Forecast, 2025 To 2030, AI could add 1.3 to 1.7 gigatons of carbon emissions annually, equivalent to 2–3% of global emissions. Additionally, global data center demand is projected to triple by 2030, and AI’s share of US data center electricity use is expected to rise from 12% in 2024 to 70% by 2035.
As AI workloads scale, the cloud is no longer abstract. Its physical footprint is local, tangible, and, increasingly, a concern. While data centers currently account for roughly 1.5% of global electricity consumption, their impact is concentrated in certain regions. US data centers already capture 4% of the country’s electricity use, but in Virginia, this rises to 26% of electricity consumption. By 2028, US data centers could push national electricity demand to 9%, potentially increasing power costs by nearly 20%.
Power availability has become the dominant constraint on data center siting in the AI era. According to the International Energy Agency’s base case for AI adoption, China and the United States are projected to account for nearly 80% of the growth in data center electricity demand through 2030. But build patterns are bifurcating: Massive AI training clusters concentrate in power‑rich, highly scalable campuses, while AI inference and other latency‑sensitive workloads increasingly distribute closer to users across regional metros and edge locations. This dynamic is accelerating regionalization of data center builds since inference requires proximity, data sovereignty rules are expanding, and grid capacity and policy influence where growth can occur.
AI’s Local Footprint
Large data centers bring real benefits to host communities. They support the digital competitiveness of regions by attracting and lowering the barriers for further tech investments, expand local tax bases, and create construction and operations jobs. But they also concentrate electricity demand, water use, noise, and even risk of pollution, including thermal pollution, where heated or cooling water is returned to local water systems, altering temperatures and stressing ecosystems. Communities, utilities, and operators now face a shared planning challenge on how to expand AI infrastructure without shifting costs or environmental burdens onto nearby residents.
This tension is driving new approaches to location and operations. Microsoft’s Community-First AI Infrastructure initiative, announced January 13, 2026, explicitly addresses local impact. Microsoft has committed to paying its full power costs so residential electricity bills do not rise, increasing transparency and replenishment around water use, investing in local jobs and training, and, in some cases, forgoing local property tax incentives. The initiative reflects a broader recognition that community acceptance is becoming a prerequisite for continued expansion.
Understanding Local Impacts: Soft And Hard Effects
The effects of data centers on host communities are best understood by separating “soft” socioeconomic impacts from “hard” technical and environmental ones.
Soft impacts shape local economies, public finances, and perceptions of fairness. Because data centers require large up-front investments in land, buildings, and equipment, they can generate substantial property and business tax revenue. Construction phases often bring short-term job growth, and the presence of advanced digital infrastructure can attract complementary businesses.
These benefits are uneven, however. Once operational, data centers typically employ relatively few permanent jobs compared to their physical footprint. Large facilities may occupy prime land, raising questions about whether the economic return justifies the scale of development. In regions facing housing shortages or competing land use priorities, this trade-off becomes especially contentious. Locking large parcels into single-purpose infrastructure also creates long-term opportunity costs that are difficult to reverse.
Hard impacts are more visible. On the positive side, data center development frequently brings major energy infrastructure upgrades. Operators often fund or cofund new substations, transmission lines, fiber networks, roads, and utility connections. These investments modernize local infrastructure and benefit surrounding businesses and residents.
Data centers also catalyze energy innovation. Their large, predictable electricity demand can justify investments in grid upgrades, renewable generation, and energy storage. Many operators now commit to renewable energy procurement and advanced efficiency measures, including improved cooling and power management, which can strengthen grid resilience over time.
But the negative hard impacts dominate community debates. Energy consumption is the most prominent concern. Individual data centers can draw as much power as small cities, and rapid clustering can overwhelm local grids if utility planning lags. Without careful coordination, residents may face higher electricity costs, reliability issues, or delays in upgrades needed for other economic activity.
Water use is another flashpoint, particularly for facilities relying on water-intensive cooling. In water-stressed regions, large withdrawals can heighten competition with residential, agricultural, and ecological needs. Even in water-rich areas, perceptions of waste or environmental risk fuel opposition. Facilities using evaporative or hybrid cooling consume water directly, while their electricity use embeds indirect water consumption at power plants. New guidance from the American Water Works Association addresses this concern by emphasizing early, joint planning around reclaimed water, infrastructure capacity, and seasonal constraints.
Quality-of-life impacts also matter. Backup generators and cooling systems can create persistent noise, large windowless buildings may alter neighborhood character, expelled heat can affect local microclimates, and onsite fuel storage raises safety and emergency preparedness questions.
Accountability
An emerging principle is that those who create infrastructure costs should also pay for them. Traditional utility rate-making can shift grid upgrade expenses onto households, prompting growing pushbacks from regulators and communities. Microsoft’s commitment to prevent electricity cost increase directly addresses this concern and signals a broader shift toward cost causation.
Data center operators are primarily responsible for community impacts because they control siting, design, power and water procurement, and operating practices. Enterprises are secondarily responsible because their AI and cloud workloads drive the demand that determines the scale of expansion, so they must be accountable for efficient, carbon-aware, and water-aware use of compute power. Together, we must have a transparent, end‑to‑end chain of accountability that allows communities to see that both operators and users are acting responsibly.
Ultimately, minimizing the environmental and social impact of data centers requires shared responsibility. Data center operators control facility design, siting, and energy procurement. Enterprises deploying AI control workload intensity, model and vendor choices, and infrastructure decisions. Infrastructure vendors across server, storage, and networking can endeavor to implement circular economy initiatives including accurate lifecycle assessments. As AI infrastructure expands, this shared accountability may determine not just who pays but where growth is still welcome.
At Forrester, we are exploring these topics in a series of reports — beginning with a framework on the impact of AI data centers. Send me a message at [email protected] or [email protected] to talk about this research.
A version of this blog was originally written for Computer Weekly.
