Artificial intelligence has triggered a global sprint to construct sprawling data centers, with the United States leading the charge. By 2030, these facilities are expected to account for as much as 12% of American electricity consumption, while their water demands could multiply fivefold by 2028.
The scale has ignited fierce resistance. Between May 2024 and March 2026, projects worth $292 billion were shelved or delayed because of grassroots opposition—with $130 billion of cancellations happening in the first quarter of 2026 alone. Roughly three in five Americans now say they do not want a data center built in their vicinity. Even Erin Brockovich, the celebrated environmental crusader from the 1990s, has returned with a crowdsourced map documenting community grievances; the platform has already logged more than 9,000 reports.
Yet the question remains: are developers actually shifting their practices in response?
Attending the Data Center Dynamics conference in New York in March offered a window into how the sector was digesting the pushback. The mood suggested wariness—one attendee had brought personal security, though this was unrelated to any data center controversy. On the exhibition floor, where thousands gathered, sustainability appeared to be treated as a bonus rather than a requirement. While panelists on stage emphasized energy efficiency, openness, and stakeholder relations, and a handful showcased green initiatives like waste heat recovery, the dominant sentiment was frustration. Industry figures felt their sector was being unfairly targeted despite decades of operation underpinning the internet.
The hierarchy of concerns I observed, and that other professionals have since confirmed, ranks as follows: speed, speed, speed, cost, and sustainability only afterward.
Reframing sustainability as a financial advantage
For those tracking the environmental toll of this expansion, action feels overdue. A comprehensive federal policy framework to curb data center environmental damage is virtually nonexistent in the United States. Although local authorities hold some power, creating incentives for developers to embrace sustainable methods on their own volition is essential. This does not rule out broader debate about whether such rapid growth is justified, but aligning developer interests—higher profits—with environmental goals is achievable. If construction must proceed, it should be done responsibly.
Consider the business case for going green:
Faster builds and lower running expenses
Building data centers demands substantial capital investment upfront. Leading AI firms have already committed over a trillion dollars while generating only modest returns. Community resistance has inflated expenses further, as tax incentives disappear, timelines slip, and power bills climb. Geopolitical tensions, including the US-Iran conflict, are also straining AI supply chain economics.
Simultaneously, renewable and efficient technologies have become more affordable. Energy typically represents the largest operating expense for a data center, making efficiency gains a logical focus. The most economical megawatt is the one never constructed. A test case demonstrated that a 16% reduction in Power Usage Effectiveness—the measure of total energy consumed versus energy used by computing hardware—yielded $262,000 in annual savings at a 1 MW facility.
Renewable energy procurement remains the cheapest option for new power generation, even absent government subsidies. For operators concerned about solar and wind variability, battery storage costs have dropped dramatically, offering a dependable and clean complement to the energy portfolio. Gas plants, by contrast, are becoming costlier. BloombergNEF data shows construction expenses jumped 66% since 2023, with build times extending 23% longer and turbine delivery timelines now stretching past 2028.
Water consumption and sound levels rank among the most divisive issues in local disputes. Cutting-edge liquid cooling approaches, including direct-to-chip systems, minimize onsite water needs while boosting energy performance and trimming costs. They also generate less noise because they eliminate the need for loud, high-velocity cooling fans.
Circular design principles can also lower expenses. Reusing existing structural elements costs 15% to 20% less over a facility's lifespan than purchasing new materials. Decommissioned hardware can find second lives and be resold. Today, only 10% of scarce minerals from discarded data center equipment gets recovered and recycled—a squandered profit opportunity.
Converting abandoned factories or retail spaces into data centers sidesteps land clearing and often taps existing grid and water connections, accelerating permits and construction.
Strengthened resilience and regulatory readiness
Circular practices also shield against supply disruptions. Metals like copper, lithium, and aluminum lack viable alternatives and flow through narrow, concentrated channels. Rising demand paired with tightening availability has spawned startups focused on extracting these elements from waste streams.
Adopting sustainability also counters regulatory exposure. Last year saw over 200 bills introduced across all states targeting data center regulation. This legislative momentum persisted into 2026, with local freezes and statewide moratoriums in New York and Texas on large new facilities.
Facilities that embed sustainability become "future-proof" assets, meeting tomorrow's rules today. This forward-thinking stance raises approval odds, avoids expensive upgrades, prevents fines, and ensures uninterrupted operation as rules tighten.
Stronger reputation and stakeholder confidence
Developers prioritizing sustainability, openness, and authentic community partnership—from initial planning through ongoing operation—will stand apart and earn deeper stakeholder backing.
Firms that reveal their environmental footprints see stronger financial performance; energy, mining, and manufacturing sectors show the clearest link. Sustainability also strengthens business-to-business positioning, as lower-carbon data centers help clients achieve their own Scope 3 emission reductions. The sheer scale of data center expansion creates room for breakthrough sustainability innovations. Major operators can anchor demand for emerging energy and water technologies, reducing their risk.
One avenue involves partnering with city authorities to channel waste heat into district heating, agricultural greenhouses, or fish farms. Emerging studies are also encouraging AI data centers to build flexibility into electricity consumption, particularly during peak demand windows.
Government must combine incentives with guardrails
Market forces can elevate sustainability standards, but rules must establish a minimum floor to prevent damage.
Transparency should come first, enabling broader visibility into data center impacts through verifiable data. The permitting phase should mandate developer assessments of projected effects, validated by independent auditors. At a minimum, public reporting of water and energy consumption is necessary, with water figures including consumption at power generation sites.
During crises or peak periods like heat waves, municipalities should be able to reduce data center electricity draw to protect grid stability. This can pair with faster grid access for efficient facilities, as Germany does through flexible connection agreements that grant quicker hookups.
Tax breaks should go only to developers demonstrating credible sustainability commitments. The UK model, where data centers claim relief on energy-saving tech investments, illustrates this approach. Governments should also leverage purchasing power, awarding IT contracts only to firms meeting defined criteria.
Winning a different kind of race
Data center discourse fixates on construction methods. Why these facilities exist receives less scrutiny. At the DCD conference, speakers repeatedly invoked the need to "win the AI race," with vague references to "beating China."
Yet Americans harbor doubts beyond environmental concerns. Skepticism about AI itself is growing, with people questioning whether they will benefit or be left behind. Without addressing questions spanning job impacts and other AI hazards, data center opposition will persist.
Perhaps the moment has come to redefine the "AI race"—not as geopolitical competition, but as a contest between a future where centralized, large-scale models dominate, deepen inequality, and exhaust resources, versus one where distributed, smaller models emerge from communities, run on clean power, and deliver broad societal gains. Such models would also be leaner and more efficient, requiring fewer data centers overall.
Technology should serve human ends, not become the objective itself. Patience, deliberation, and environmental responsibility will ultimately prevail.
