Who Builds the AI Future – and Who Pays for It?
The AI Data Center Boom, Redlining’s Long Shadow, and the Urgent Fight for Community Power
Midway through a March 2026 public meeting in Imperial County, California, chants of “No data center!” grew so loud that county officials had to pause the proceedings at times, and the developer slipped out during recess. The proposed nearly one-million-square-foot data center in Imperial County would be the largest in California, if completed. Despite massive community backlash, plans to move forward were approved last month by Imperial County. The City of Imperial is suing in response.
Imperial is not alone. Across the country, in towns that have weathered deindustrialization, disinvestment, and decades of broken promises from decision-makers, a new industry is arriving with extraordinary speed, massive capital, and very little accountability to the people already living there. Data centers, a critical piece of artificial intelligence (AI) infrastructure, are being built at a historic pace. For many communities, especially communities of color, the deal on offer looks suspiciously familiar: take on the costs, absorb the risk, and wait to be told what benefits, if any, trickle down.
We have seen this before. We know how it ends.
The Infrastructure of Tomorrow Is Being Built on the Inequities of Yesterday: A New Frontier of Redlining Built on Old Foundations
Data centers are the physical backbone of AI – the warehouses of servers that power everything from large language models to predictive algorithms used in healthcare, housing, hiring, criminal justice, and others. As AI investment has exploded in recent years, so has the demand for this infrastructure. Billions of dollars are flowing into new data center construction, landing predominantly in economically distressed communities where land is cheaper and regulatory pushback has historically been weaker. These are, overwhelmingly, communities of color.
The significant environmental and economic impacts of data center development are well-documented. Data centers are incredibly energy-intensive – a single large facility can consume as much electricity as a small city, straining local power grids and driving up utility rates for residents – and energy demand is projected to double or even triple by 2028. The facilities draw millions of gallons of water for cooling, often utilizing diesel-powered generators that also cause adverse health impacts like asthma for residents in nearby communities.
Job creation is often cited as the primary benefit of data center development, but the new jobs created are often temporary and specialized, with too few permanent jobs to justify the millions in public subsidies. Without adequate intervention, too often data centers provide little long-term benefit to communities targeted for development. The economic gains largely flow outward, while the costs stay local.
Redlining’s Legacy Is Being Rewritten Into the AI Economy
For decades, The Greenlining Institute has fought the compounding consequences of redlining – the now illegal practice by financial institutions of systematically excluding Black and Brown communities from access to credit, homeownership, and wealth-building opportunities. Decades of redlining concentrated environmental burdens and economic precarity in communities of color, triggering cycles of neglect and chronic disinvestment that persist today. While the redlining of the past has been outlawed, its consequences are deeply embedded in these communities. At the same time, new forms of redlining are emerging, as these same communities continue to be targeted for infrastructure development – absorbing environmental impacts, without sharing in the economic benefits.
This is now becoming the case with AI infrastructure – communities redlined out of prosperity then, are being asked to shoulder the costs of a technological revolution whose benefits will flow largely elsewhere. As it stands, it serves less as economic development, and more as extraction dressed up in the language of innovation.
This is environmental injustice and economic inequity, and it is happening right now, at scale, with little public accountability. And the communities bearing the greatest burden of this reckless expansion are not the communities making decisions about it. It is a story of exclusion and disenfranchisement we have seen before.
Community Power Is the Only Adequate Response
Greenlining has spent years fighting for communities navigating an increasingly automated world – challenging discriminatory decision-making systems, advocating for equal access to opportunity free from algorithmic bias, and demanding accountability from institutions that are deploying technologies in our communities. We have seen what happens when technology is built on the backs of communities rather than with them. We have also seen what becomes possible when communities have genuine power to set terms.
When communities have a say in technological advancements that directly impact them, it is possible to mitigate environmental and economic harms that reinforce the legacies of redlining. Community involvement and decision-making power can also help create new pathways to prosperity and community wealth-building.
But right now, meaningful community governance around AI infrastructure is largely nonexistent – especially at scale. Decisions about where data centers are built, how large they grow, their water and electricity usage, and what safeguards apply are determined through processes that are public but often inaccessible – and frequently obscured by nondisclosure agreements. State and local policies requiring transparency and disclosure are severely limited. Recent efforts to require data centers to report on water usage in California for instance, were thwarted last year.
Not to mention, local governments, often under-resourced and eager for investments promised by new development, can lack the capacity to advocate on their residents’ behalf fully, while developers are happy to circumvent existing processes to avoid environmental and community reviews. For example, the developers behind the $10B Imperial County data center intentionally proposed the project on unincorporated county land, deliberately designing the project to circumvent lengthy environmental analysis under California Environmental Quality Act. These practices are common, and often by the time communities understand the full scope of what has been approved, the terms are already set.
Communities Must Set the Terms
This is not how consequential, long-term infrastructure decisions should be made. Communities – particularly those directly impacted by the development of AI infrastructure – deserve a genuine seat at the table before decisions are finalized, not after. They need clear and honest information about proposed developments and their projected impacts. They need organized capacity to participate meaningfully, and they need binding commitments, not just aspirational language, that guarantee local economic inclusion, environmental protection, and real accountability.
There are tools available to make this happen. Tools like community benefits agreements could offer a promising, underutilized path for communities to negotiate development terms. For decades, Greenlining has worked with formerly redlined communities to negotiate CBAs on their behalf with banks, leading to billions being invested back into disinvested communities of color. Similar agreements for AI data centers could mean developments must include local hiring pipelines, pathways to technical training, community investment funds, meaningful environmental monitoring, and accountability mechanisms if development commitments are not kept. It means communities would not just host the infrastructure of the AI economy, but have a meaningful stake in it.
Getting It Right
The decisions being made right now about AI and its infrastructure will shape communities for generations. The industry is moving fast; policy is moving slowly; and in that gap, communities are losing leverage every day.
Without urgent intervention, the current trajectory threatens to deepen the impacts of redlining. Formerly redlined communities will bear greater environmental burdens, and already disinvested communities will be denied any meaningful economic benefits from these developments. Especially as labor displacement from AI automation will also fall disproportionately on workers of color, without intervention, AI and data center development will ultimately widen the racial wealth gap.
At Greenlining, we are fighting for community self-determination in the development of AI infrastructure. Communities of color who have been systematically excluded from decisions that shape their lives must be able to set the terms under which this infrastructure enters their neighborhoods. To ensure their needs are met, their concerns are genuinely addressed, harms are remediated, and the economic benefits of the AI economy reach them, not just pass through them. It requires those of us working at the intersection of racial equity and technology policy to name clearly what’s at stake, and to fight accordingly.
Working with our partners at James Irvine Foundation, and others in our coalition, we are building a portfolio of strategies to meet this moment. Data centers are one urgent piece of a larger, interconnected fight over how AI reshapes our collective futures: our shared economy and opportunities within it, and environmental and wellbeing impacts to our communities. If you are organizing in communities affected by data center development, and working on community engagement in AI infrastructure governance, or building coalitions at this intersection – we want to connect.
This post is the first in a series on community-centered AI infrastructure and the fight for economic equity in the age of artificial intelligence.
