Across the United States, the infrastructure behind AI is surfacing familiar tensions around resources, economic opportunity, and community agency, with lessons worth bringing to conversations in New York.
This week, the development sector will gather in New York for the UN General Assembly (UNGA) and Climate Week, where AI will feature prominently in conversations about the future of sustainable development. Across UNGA, much of the conversation will focus on how AI could accelerate progress toward the Sustainable Development Goals (SDGs). At Climate Week, a parallel conversation will focus more directly on the energy and infrastructure required to power it.
It is tempting to frame this as a simple tradeoff: if AI can accelerate progress across the SDGs, are the costs of building its infrastructure worth the benefits? But the costs and benefits are not fixed; they will be shaped by the choices governments, companies, and communities make along the way. What is playing out across the United States right now suggests just how much those choices matter.
The Moment
Data center construction is accelerating rapidly. Across North America’s largest data center markets, capacity under construction increased nearly 25% in the first half of 2026, reaching a record high. At the same time, public opposition to data centers in the U.S. is growing just as quickly. A recent national poll found that 75% of Americans would oppose a new data center being built near where they live, up from 42% when the same question was asked just one year earlier. As community concerns have grown, local and state governments across the country have responded with new regulations and moratoriums.
In many of those communities, those pauses have created space to reconsider the terms under which data centers should move forward. Governments are considering what conditions should govern future development, communities are pushing for a greater say, and companies are confronting growing expectations around the impacts of what they build.
But what makes this moment so interesting is not simply the scale of the boom, or the opposition to it. It is the development tensions the boom is bringing to the surface. These are classic development questions playing out around a distinctly 21st-century form of infrastructure. How do we capture the benefits of rapid development while managing its environmental and social costs? Who benefits, who bears the risks, and who gets a say? And what role can governments play in shaping the outcome?
As leaders gather in New York to debate the future of AI and sustainable development, communities across the U.S. are working through many of these same questions in real time.
The SDGs offer a useful lens for exploring them. Rather than asking whether data centers are simply “good” or “bad” for sustainable development, they allow us to look more closely at the choices and tensions emerging around resources and the planet, people and prosperity, and communities and institutions – and the questions these raise for policymakers, funders, companies, and others considering AI’s role in sustainable development.
Resources & Planet | SDGs 7 (Energy) and 13 (Climate)
Can the energy demand straining today’s grid help build a cleaner, stronger one for tomorrow?
Data centers consumed more than 4% of U.S. electricity in 2024, a figure projected to more than double by 2030. As that energy demand grows, it could increase costs for households, strain reliability, and prolong reliance on fossil fuels
But the scale of that demand also creates an opportunity. The technology companies building and operating data centers need enormous amounts of new power – and they have both the resources and incentive to invest in securing it. That creates the possibility that the same demand straining U.S. energy systems could also accelerate investment in new generation, transmission, storage, and emerging clean energy technologies.
In Nevada, for example, Amazon is working with the local utility to bring 700 megawatts of new geothermal, solar, and battery-storage capacity onto the grid to support future data centers. Amazon says it will cover the costs associated with the new energy infrastructure and generation rather than passing them on to households, businesses, and other local rate payers.
But corporate commitments alone will not determine whether this investment leaves the broader energy system better off. Governments are beginning to test how the enormous demand created by data centers can be translated into broader public benefits. Pennsylvania’s Governor’s Responsible Infrastructure Development (GRID) requirements offer one emerging model. Covered data center projects must secure enough incremental electric capacity to meet their new demand and pay the costs they impose on the grid. The requirements also phase in increasing amounts of dispatchable clean firm energy generated in Pennsylvania, reaching 32% by 2035.
The question, then, is whether the pressure created by enormous new electricity demand can become a catalyst for building a cleaner, more resilient energy system. The answer will depend on the policies, investments, and agreements being put in place around data centers, and whether they leave behind an energy system that works better for the communities around them.
Questions to bring to New York:
- How can we shape the enormous energy demand created by AI into a catalyst for investment in a cleaner, more resilient energy system?
- What mechanisms can ensure that data centers bear the energy and infrastructure costs created by their demand rather than shifting them to local households, businesses, and governments?
People & Prosperity | SDG 8 (Decent Work & Economic Growth)
What does “decent work” mean when building AI infrastructure is creating good jobs today, while AI itself may transform work tomorrow?
The jobs story around AI is complicated. One of the clearest illustrations comes from organized labor itself. In a 2026 resolution titled “We Want Good Jobs Today and Tomorrow,” the AFL-CIO called for data centers and the infrastructure supporting them to be built, operated, and maintained with union labor. In the same resolution, it called for protections against AI’s potential effects on jobs and job quality.
These positions are not necessarily contradictory. They reflect two parts of the same transformation already unfolding.
The economic development case for data centers has often centered on jobs. Yet data centers require enormous capital investment and relatively few people to operate once built. A Virginia legislative study illustrates the difference. A typical 250,000-square-foot data center may have roughly 1,500 workers on site at peak construction, compared with around 50 full-time workers once operational.
But focusing only on permanent employment can obscure another part of the story: the people building these facilities. Construction jobs are temporary, but an infrastructure boom isn’t necessarily temporary for the workers building it. In regions with a sustained pipeline of development, skilled tradespeople can move from one project to another as facilities are built, expanded, upgraded, and maintained.
For those workers, the AI boom is not an abstraction about the future of work. It can mean years of well-paid work and a pathway into the middle class. The head of the International Brotherhood of Electrical Workers Local 26 in the Washington, D.C. region recently wrote that his members worked roughly 14 million hours annually a decade ago, compared with 28 million in 2025 and a projected 33 million this year. The union added 600 apprentices last year alone. He attributes much of that growth to data center development.
But the AFL-CIO resolution points to the other half of the story. The physical infrastructure behind AI is creating significant demand for certain kinds of work today, even as the technology it enables raises profound questions about the future of other kinds of work. Organized labor is already navigating both sides of that equation.
Opportunity and disruption can happen at the same time, in different parts of the labor market. The workers benefiting from the infrastructure boom today are not necessarily the same workers whose jobs may be transformed tomorrow. The question is not simply whether AI creates or eliminates jobs, but where opportunity is being created, where disruption may occur, and how we prepare for a transformation that will not affect all workers in the same way.
Questions to bring to New York:
- What would it take to turn an infrastructure boom into durable pathways to skilled, well-paid work?
- How should workforce strategies account for the fact that AI may create opportunity in some parts of the labor market while disrupting others?
Communities & Institutions | SDGs 11 (Sustainable Communities) and 16 (Strong Institutions)
What happens when communities gain the power to stop development, but not necessarily the power to shape it?
In Coweta County, Georgia, residents concerned about a proposed data center have turned to one of the most basic tools of local democracy: gathering signatures. Hundreds showed up on a Sunday afternoon this spring to sign a petition seeking a referendum on the project and future data center development. One resident, who had never been involved in local organizing before, trained 38 volunteers and began collecting signatures from her neighbors, her veterinarian, her dentist — “everywhere I go,” she told The Guardian.
The petition is just one part of a larger community push in Coweta County. Another group, Stop Project Oak, mobilized against a proposed 255-acre campus ahead of an August public hearing; later that month, county commissioners unanimously rejected the project’s rezoning request
Coweta County is one example of something happening across the country. Communities have demonstrated remarkable power to slow or stop data center development. Increasingly, governments are responding to community concerns by adopting new regulations and moratoriums. Of the 42 local data center ordinances and ballot measures recently reviewed by the Federation of American Scientists, 33 were passed in 2026.
That is a meaningful demonstration of community agency. But the power to stop development is not necessarily the same as the power to shape it. Moratoriums can give communities and governments time to ask harder questions about where data centers should be built, who pays for new infrastructure, what environmental protections are required, and what benefits should flow back to the people hosting them. But what happens when the moratorium ends?
Community benefits agreements (CBAs) offer one potential mechanism for giving communities greater influence over the terms of development. Traditionally, CBAs have allowed community coalitions to negotiate directly with developers over major projects. But that is not yet the dominant model around data centers. The Federation of American Scientists’ review of ten executed data center agreements found that they were negotiated between developers and local governments, while the World Resources Institute finds that agreements negotiated directly between developers and community coalitions remain rare.
In practice, local governments are likely to remain the primary negotiating party in many data center agreements, and they can negotiate significant benefits on behalf of their constituents. The opportunity is to ensure those negotiations meaningfully involve the communities they represent: giving residents a voice in identifying priorities, shaping what governments negotiate for, and understanding the agreements ultimately reached.
There are early efforts to give communities more tools to influence those decisions. The NAACP, for example, has developed a community benefits agreement template and negotiation resources specifically for communities facing data center development. Whether direct community negotiation becomes a meaningful part of the data center boom remains an open question.
That makes the data center boom an important test of what meaningful community agency looks like during rapid technological change: not just the power to say no, but the power to shape what gets built, how it gets built, and who benefits.
Questions to bring to New York:
- How can communities translate the leverage created through opposition into meaningful influence over the terms of development?
- As AI development accelerates, what can the data center debate teach us about creating meaningful avenues for public participation in technological change?
What Can We Learn From This Moment?
The U.S. data center boom is neither a model for the rest of the world nor simply a cautionary tale. It is a live experiment in what happens when technological transformation meets the realities of development.
What makes it worth watching is that the outcomes are not predetermined. Enormous energy demand could strain already-stretched systems or help drive investment in cleaner, more resilient ones. An infrastructure boom can create meaningful economic opportunity while the technology it supports transforms the future of work. And communities can exercise real power to stop development while still searching for ways to shape what happens next.
How these tensions play out will depend on choices being made now, by governments, companies, workers, and communities. And those choices will inevitably look different in different places.
As conversations in New York turn toward what AI could mean for sustainable development, it is worth also asking what its infrastructure is already teaching us. The opportunity is to bring those conversations closer together: pairing questions about what AI could do for sustainable development with questions about what it will take to build the infrastructure behind it. The U.S. data center boom does not offer easy answers. But it gives us a fascinating place to start asking better questions.
