By Mark T. Holcombe

What happens when a city regulates AI infrastructure but fails to give affected communities a meaningful role in deciding where its burdens fall?

A recurring mistake in public debates about large infrastructure projects is the assumption that legality settles morality. If a project satisfies zoning rules, receives permits, and promises economic development, then the ethical work appears finished. But that assumption is precisely what AI infrastructure forces us to question. Data centers are not weightless “cloud” facilities. They are industrial sites that consume land, water, electricity, public attention, and political trust.

In my earlier posts on AI infrastructure in Alabama, I argued that the ethical question is not simply whether a data center is legal or economically useful. The question is who bears the burdens, who receives the benefits, and whether affected communities are treated as genuine parties of interest rather than obstacles to development (Holcombe, 2026a, 2026b). The Birmingham City Council’s recent data center ordinance and the Birmingham Industrial Development Board’s tax abatement package for Nebius confirm why that framework matters.

Birmingham did not get everything wrong. That needs to be said clearly. But Birmingham also did not get this right. The result is a partial regulatory success attached to a serious moral failure of democratic process and public accountability.

 

What Birmingham City Council Got Right

The Birmingham City Council was right to recognize that hyperscale data centers require specific regulation. Generic industrial zoning is inadequate for facilities whose impacts may include high electrical demand, water use, noise, backup power systems, heat, stormwater issues, and long-term neighborhood disruption. The City’s own materials state that Birmingham’s electrical grid, industrial infrastructure, and land affordability make it attractive to data center developers, which creates a need for clear protective standards (City of Birmingham, 2026a).

The ordinance also includes several meaningful safeguards. The City’s 20-point framework requires a 500-foot setback from residential or urban neighborhood districts, a 1,000-foot separation from high-capacity transit stations, a minimum five-acre lot size, disclosure of peak electrical demand, water-use estimates, proof of adequate water and wastewater capacity, backup generator disclosures, pre- and post-construction noise studies, closed-loop cooling, restrictions on routine fossil-fuel power generation, lighting limits, and certified-mail notification to property owners within 500 feet (City of Birmingham, 2026b). These are not trivial requirements.

The closed-loop cooling requirement is especially important. In my earlier post on Alabama utility reform, I argued that AI infrastructure ethics must account for the physical burdens of computation, especially water, power, and utility costs (Holcombe, 2026b). Birmingham’s ordinance at least

 

recognizes that water demand is not a side issue. It requires applicants to describe water sources, cooling systems, efficiency plans, and wastewater capacity, and it regulates cooling-water discharge to reduce contamination risks (City of Birmingham, 2026b).

The ordinance also correctly treats full build-out plans as ethically relevant. Mayor Randall Woodfin emphasized that data center companies should not be able to “phase” their way around regulation by disclosing only part of a project at a time (Bazinaw, 2026). That is a legitimate concern. A project’s moral and civic significance cannot be evaluated if the public sees only a partial footprint.

The Council also had a defensible reason for acting quickly. Councilor Josh Vasa argued that having some regulatory framework in place, even one that could be improved, was better than leaving existing gaps unaddressed. He also expressed concern that delay could allow already permitted projects to seek expansion under weaker rules (Denham, 2026). That argument should not be dismissed. In a fast-moving infrastructure environment, regulatory delay can function as deregulation.

So what did Birmingham get right? It recognized that data centers are not ordinary office buildings. It created real environmental, infrastructure, and siting requirements. It attempted to close the loophole of incremental expansion. It acknowledged that AI infrastructure has local material costs.

That is the strongest case for the Council’s action.

 

Where City Council Morally Failed

The Council’s central failure was procedural. The ordinance removed the special exception requirement for hyperscale data centers if they satisfy the 20 conditions. That matters because a special exception process would have required zoning-board approval and public hearings. Without it, a project that satisfies the checklist can move through an administrative process without the same public forum for affected residents (Bowser, 2026; Brooks & Harksen, 2026).

This is not a minor technical dispute. It goes to the heart of moral legitimacy.

In my earlier post on digital redlining and the Bessemer data center controversy, I argued that the ethical issue is not merely whether land can legally be rezoned or developed. The issue is whether affected communities are treated as full participants in decisions that shape their homes, health, environment, and future (Holcombe, 2026a). The removal of the special exception requirement weakens that participation.

Supporters of administrative review may argue that clear standards produce predictability and prevent politicized permitting. That argument has some force. Government should not create arbitrary, endless uncertainty for lawful development. But predictability for developers cannot come at the cost of voice for residents. The Southern Environmental Law Center argued before passage that special exceptions are not “anti-growth” but a key protection when a land use may significantly affect neighboring properties or public welfare (Terrell, 2026a). That is the better argument.

The public response shows the legitimacy problem. Roughly 300 people crowded City Hall for the hearing, and many speakers argued that the ordinance left too much uncertainty and too little public oversight (Terrell, 2026b). WBHM reported that residents objected to the lack of clear quantitative standards for sound, water use, heat, and emissions (Terrell, 2026b). WBRC similarly reported concerns about enforcement, proximity, public input, and the absence of specific noise limits or monitoring requirements (Brooks & Harksen, 2026).

 

This is where “compliance-as-ethics” fails. A checklist can be useful. But a checklist is not a democratic process. A checklist does not ask whether the people most affected have had meaningful input. A checklist does not resolve whether burdens and benefits are fairly distributed. A checklist does not answer whether affected residents have a credible enforcement mechanism when something goes wrong.

The Council’s error was not that it regulated. The error was that it treated substantive safeguards as a substitute for procedural justice. They are not substitutes. They are different moral requirements.

 

The Nebius Problem

The Nebius project exposes the deepest tension in the ordinance. The City’s new rules apply to future data center projects and expansions, but the already permitted Nebius project in Oxmoor Valley is not fully subject to the new ordinance unless it expands or materially changes its plans (Denham, 2026). Mayor Woodfin acknowledged that residents feel something major is being built near their homes without their full consent, and he stated that the City could not undo projects already in the pipeline (Bazinaw, 2026).

That may be legally true. But legal truth does not dissolve ethical responsibility.

If the Nebius project was already far enough along to avoid the strongest version of public review, then the City’s process was structurally inadequate before the controversy reached full public attention. Council President Wardine Alexander stated that many decisions can be made administratively before the Council has an opportunity to weigh in, and that she learned about the project’s scope after key steps were already underway (Brooks & Harksen, 2026). That admission matters. It suggests that Birmingham’s decision structure was not designed for projects of this scale.

The result is ethically backward. The project that generated much of the public controversy is also the project least governed by the newly adopted protections. The City passed stronger regulations that does not apply to an already permitted controversial data center.

 

What the IDB Got Right

The Birmingham Industrial Development Board also has a defensible argument. According to WBHM, IDB Chair David Perry defended the Nebius incentives by pointing to projected tax revenue. He stated that the site currently generates about $200,000 per year in property taxes and that, with the project, direct tax revenues over 30 years are projected to rise substantially (Gelderman, 2026). The Birmingham Times reported that Nebius is projected to invest nearly $35.9 billion at the site over 30 years and create 78 jobs with an average salary above $90,000 (Birmingham Times, 2026).

The IDB also did not approve a simple blank check. The Birmingham Times reported that the agreement includes no cash and no upfront incentives. The abatements are contingent on actual investment, and the agreement contains clawback provisions if Nebius fails to meet projected payroll commitments. If Nebius ceases operations during the abatement period, it must restore the property to no worse condition than when it purchased it (Birmingham Times, 2026).

Those are meaningful protections. A clawback provision is better than an unconditional giveaway. Excluding education taxes from the abatement is also important, because public schools should not be treated as expendable collateral in economic development deals (Birmingham Times, 2026).

 

So the IDB’s strongest defense is this: the project may generate far more public revenue than the vacant site currently produces, the incentives are conditional, education taxes are not abated, and the agreement contains clawbacks. That is the best version of the IDB’s argument.

 

Where the IDB Failed

The IDB’s failure is that it approved a major incentive package for a controversial AI infrastructure project without satisfying the higher transparency burden such a project should demand.

On May 29, 2026, the Birmingham Industrial Development Board voted 6-0 to approve a package that included a 65% abatement on non-education property taxes for up to 30 years and an 80% abatement on construction-related sales and use taxes for up to 30 years (Birmingham Times, 2026). WBHM reported that the package amounted to $3.2 billion in tax abatements over 30 years (Gelderman, 2026). A project of that magnitude is not a routine business recruitment matter. It is a public governance decision.

The IDB’s institutional status creates an accountability problem. The board operates independently under state law, but its members are appointed by the Birmingham City Council (Birmingham Times, 2026). Mayor Woodfin therefore can say, in a technically narrow institutional sense, that the City of Birmingham did not itself give Nebius tax incentives. The IDB, not the mayor or City Council, approved them. But that answer is too narrow to satisfy public accountability.

Residents do not experience government in compartmentalized legal categories. The public experiences one civic ecosystem. If the City appoints the board members, if the project depends on municipal zoning and permitting, if the public bears local infrastructure and environmental risks, and if the mayor publicly distances himself from the incentive decision after the fact, then the formal independence of the IDB does not settle the ethical question. It may explain authority. It does not justify the process.

The IDB also appears to have relied heavily on projected public benefit. Projections are important and relevant, but projections are not moral conclusions. They are empirical predictions that require scrutiny. How reliable are the revenue forecasts? What assumptions underlie the $35.9 billion investment projection? What are the net benefits after infrastructure demands, utility impacts, environmental monitoring, emergency preparedness, and administrative enforcement costs? Who pays if those projections fail? These are not anti-development questions. They are due-diligence questions.

The IDB’s reasoning also raises a proportionality problem. Seventy-eight jobs with an average salary above $90,000 may be good jobs, but the job count is small relative to a projected $35.9 billion investment and a multi-decade abatement structure (Birmingham Times, 2026). That does not prove the deal is bad. It does mean that the public-benefit argument cannot rest primarily on job creation. It must rest on verified, enforceable, transparent fiscal and community benefits.

 

The Deeper Ethical Failure: Burden Shifting Without Full Consent

The deeper issue is moral displacement. In my earlier Alabama utility reform post, I argued that data center governance must prevent private profit from being subsidized by public utility customers and vulnerable communities (Holcombe, 2026b). The same principle applies here. If AI infrastructure uses public land-use authority, public utilities, public environmental capacity, and public tax policy, then it

 

owes more than private compliance. It owes public justification. That justification requires at least four things:

First, residents need meaningful participation before major decisions are effectively settled. Notification after a checklist has been satisfied is not the same as deliberation or public input.

Second, public benefits must be specific, measurable, and enforceable. Vague claims about innovation, growth, or future prosperity are not enough.

Third, environmental and quality-of-life protections must include clear quantitative standards. Noise, heat, emissions, water use, and generator operation should not be governed only by broad commitments and after-the-fact studies.

Fourth, enforcement must be transparent. Residents need to know who investigates complaints, what measurements trigger violations, what penalties apply, and whether those penalties are strong enough to deter noncompliance.

Without those elements, Birmingham risks reproducing digital redlining: not necessarily redlining in the narrow historical mortgage sense, but a contemporary infrastructure pattern in which communities with less power carry disproportionate burdens for technologies whose benefits are captured elsewhere (Holcombe, 2026a).

A responsible digital redlining argument requires evidence: demographic patterns, comparative siting practices, differences in procedural protections, environmental burdens, property impacts, and public-benefit distribution. But the Birmingham and Bessemer cases show why the question is legitimate.

Data infrastructure is being routed through local governments that may not yet have the institutional capacity, transparency norms, or environmental standards required for projects of this scale.

 

Birmingham City Council and the Missing Question of Siting Equity

Birmingham City Council did regulate data center siting, but it did not adequately address siting equity. That distinction matters. The ordinance asks whether a hyperscale data center is far enough from residential districts, whether it satisfies technical infrastructure requirements, and whether it provides disclosures about water, power, generators, lighting, stormwater, and noise (City of Birmingham, 2026a, 2026b). Those are legitimate land-use concerns. But they do not answer the deeper justice question: why are these facilities being placed where they are, and are vulnerable communities more likely to bear the burdens?

This is where the digital redlining concern remains largely unanswered. In my earlier analysis of AI infrastructure and redlining, I argued that the ethical issue is not simply whether a data center is legal, permitted, or economically beneficial. The ethical issue is whether communities with less political power are being asked to absorb the physical burdens of AI infrastructure, including land use, power demand, water use, noise, backup generators, environmental risk, and reduced neighborhood control (Holcombe, 2026a). Birmingham’s ordinance addresses some of those burdens individually. It does not appear to address their distribution across race, income, neighborhood vulnerability, or historical patterns of industrial siting.

A 500-foot residential setback is not the same as an environmental justice analysis. A noise study is not the same as a cumulative-burden analysis. A certified-mail notice to nearby property owners is not the same as meaningful community consent. These protections may reduce some harms, but they

 

do not determine whether those harms are being routed toward communities already carrying disproportionate infrastructure burdens.

The removal of the special exception requirement makes this problem worse. If a hyperscale data center satisfies the ordinance’s conditions, the project can proceed through a more administrative process rather than requiring the same level of public hearing and discretionary review (Bowser, 2026; Walter, 2026). That matters because public hearings are one of the few formal mechanisms through which residents can raise concerns about unequal siting, cumulative burdens, and neighborhood-specific vulnerabilities before approval. A checklist can determine compliance. It cannot, by itself, determine justice.

To address digital redlining, Birmingham’s ordinance would need at least six additional protections.

1.    A demographic-impact analysis.

Developers should be required to analyze the racial, economic, age, health, and housing characteristics of communities near proposed hyperscale data centers. The question is not merely whether the land is legally available. The question is whether the burdens are falling disproportionately on Black, low-income, elderly, medically vulnerable, or politically underrepresented communities.

2.    A cumulative-impact analysis.

The city should require an assessment of existing environmental and infrastructure burdens in the affected neighborhood. A community already located near industrial sites, heavy traffic, utility infrastructure, flooding risk, or degraded air quality should not be evaluated as if the data center were its first burden. Ethical siting must consider cumulative impact, not isolated impact.

3.    A comparative siting report.

Developers should be required to explain why a proposed site was selected over feasible alternatives. This report should include land cost, utility access, zoning, neighborhood demographics, environmental risks, and proximity to vulnerable populations. Without comparative analysis, the public cannot know whether a site was chosen because it is genuinely appropriate or because it offers the path of least political resistance.

4.    A restored special exception or equivalent public hearing process.

Hyperscale data centers should require a public hearing process with real procedural consequences. Residents should be able to ask questions, submit evidence, challenge assumptions, and require responses before approval. Public comment after major decisions are functionally settled is not meaningful participation.

5.    A community benefits agreement.

When a data center is placed near a vulnerable or historically burdened community, the developer should be required to negotiate enforceable community benefits. These may include infrastructure improvements, utility protections, local hiring commitments, emergency response funding, environmental monitoring, neighborhood investment, or direct mitigation funds. If the public bears risk, the public should receive concrete, enforceable benefit.

6.    A public infrastructure and incentive map.

Birmingham should publish a regularly updated map showing the location of data centers, substations, major utility upgrades, industrial abatements, and related infrastructure burdens. This would allow residents, journalists, researchers, and policymakers to see whether AI

 

infrastructure is clustering in particular neighborhoods. Without public mapping, inequitable siting patterns can remain invisible until they become entrenched.

These requirements would not prohibit data centers. They would force Birmingham to ask the correct ethical questions before approving them. A city can support technological development while still refusing to let vulnerable communities become the default host sites for the physical machinery of artificial intelligence.

The City Council’s ordinance is therefore incomplete. It treats hyperscale data centers as technical land-use projects, but not as potential environmental justice problems. That is the core failure.

Birmingham may have regulated the footprint of AI infrastructure, but it has not yet regulated the fairness of where that footprint lands.

 

What Should Happen Next?

Birmingham should amend the ordinance post haste. The City should restore a public hearing requirement for hyperscale data centers, whether through a special exception process or another process with equivalent legal force. A community meeting is not enough unless it has procedural consequences. Residents must have more than an opportunity to speak. They must have a meaningful role in the decision structure.

The City should also adopt quantitative standards for noise, heat, water use, generator testing, emissions, and complaint response. A rule without measurable thresholds is difficult to enforce. A safeguard that cannot be enforced is public relations not public protection.

The IDB should adopt a higher transparency standard for large AI infrastructure incentives. For projects above a certain capital investment threshold, the board should publish the proposed incentive terms, projected public benefits, assumptions behind those projections, clawback provisions, infrastructure-cost estimates, and community-impact analysis before approval. It should also require a community benefits agreement for projects that receive large abatements.

Finally, Birmingham should create a standing AI infrastructure review process that brings together planning, environmental review, utility capacity, emergency response, public health, neighborhood representatives, and economic development. AI infrastructure is not merely a technology issue. It is land-use policy, environmental policy, utility policy, fiscal policy, and democratic governance policy.

 

Conclusion

Birmingham City Council got one major thing right: hyperscale data centers require real regulation. The 20-point ordinance is better than the regulatory vacuum that existed before. The IDB also got one thing right: economic development incentives can be ethically defensible when they are conditional, tied to measurable investment, and protected by clawbacks.

But both institutions failed at the same deeper point. They treated public benefit as something that could be asserted from above rather than justified with the affected communities. City Council weakened public participation by removing the special exception requirement. The IDB approved a major tax abatement package through an institutional structure that is legally independent but politically entangled and void of public awareness who receive the benefits of public taxes in the form of social and capital services. Both decisions may be legal. Neither fully satisfies the ethical burden

 

created by AI infrastructure.

The lesson is not that Birmingham should reject all data centers. The lesson is that Birmingham must stop treating AI infrastructure as ordinary industrial development. The “cloud” has a tangible footprint. That footprint is imprinted somewhere. The moral question is whether the people living under it are treated as full participants in deciding what gets built, what gets subsidized, and what burdens they are expected to carry.

 

FAQ Section

What is the ethical issue in Birmingham’s data center debate?

The ethical issue is not only whether Birmingham’s data center projects are legal or economically beneficial. The deeper question is whether affected communities have meaningful participation in decisions that shape land use, public infrastructure, environmental risk, utility demand, and long-term neighborhood burdens.

 

What did Birmingham City Council get right?

Birmingham City Council correctly recognized that hyperscale data centers require specific regulation. The ordinance includes meaningful safeguards, including setback rules, water-use disclosures, cooling-system requirements, backup generator disclosures, lighting limits, noise studies, and infrastructure-capacity requirements.

 

Where did Birmingham City Council fail ethically?

The central failure was procedural. By removing the special exception requirement for hyperscale data centers that satisfy the ordinance’s conditions, the Council weakened public hearing requirements and reduced the formal role of affected residents in decisions with major local consequences.

 

What is the Nebius problem?

The Nebius project exposes a timing and accountability problem. The data center project that generated much of the public controversy is not fully governed by the new ordinance unless it expands or materially changes its plans. This creates an ethical mismatch between public concern and regulatory protection.

 

What did the Birmingham Industrial Development Board get right?

The IDB’s strongest argument is that the Nebius project may generate substantial public revenue, that the incentives are tied to actual investment, that education taxes are not abated, and that clawback provisions are included if promised conditions are not met.

 

Where did the IDB fail?

The IDB failed to meet the higher transparency burden appropriate for a major AI infrastructure incentive package. A multi-decade tax abatement tied to a controversial data center should require

 

public disclosure of assumptions, projected public benefits, infrastructure costs, community impacts, and enforceable safeguards before approval.

 

Why does the article connect data centers to digital redlining?

The article uses digital redlining to describe a possible infrastructure pattern in which communities with less political power carry disproportionate physical burdens for technologies whose benefits are captured elsewhere. The argument does not claim this pattern is proven in every case, but it argues that Birmingham should adopt siting-equity protections to prevent it.

 

What should Birmingham do next?

Birmingham should restore a public hearing requirement, adopt quantitative standards for noise, heat, emissions, water use, generator operation, and complaint response, require community benefits agreements for large projects, and create a standing AI infrastructure review process that integrates planning, utilities, public health, environmental review, emergency response, and neighborhood representation.

 

 

References

Bazinaw, H. (2026, June 11). Birmingham mayor makes statement on data center ordinance passing.

ABC 33/40.

Birmingham Times. (2026, June 26). Birmingham Mayor says the city didn’t give Nebius data center tax breaks. The Birmingham Times.

Bowser, J. H. (2026, June 9). Birmingham City Council votes 6-3 to pass new regulations for data centers. The Birmingham Times.

Brooks, G., & Harksen, L. (2026, June 10). Birmingham City Council approves data center zoning rules after nearly 3 hours of public comment. WBRC.

City of Birmingham. (2026a, June 9). The City of Birmingham approves Alabama’s toughest zoning ordinance, establishing 20-point community protections for data center development. City of Birmingham, Alabama.

City of Birmingham. (2026b). Data centers in Birmingham. City of Birmingham, Alabama.

Denham, H. (2026, June 11). Birmingham, Ala., City Council addresses data center vote. Government Technology.

Gelderman, A. (2026, July 1). “I would make the same decision”: Board chair defends incentives for Birmingham data center project. WBHM.

Holcombe, M. T. (2026a). Is AI infrastructure the new redlining? Inside Bessemer’s $14.5 billion data center dilemma? Mark T. Holcombe.

Holcombe, M. T. (2026b). When AI infrastructure meets utility reform, what Alabama gets right, and what it still misses. Mark T. Holcombe.

Terrell, F. (2026a, May 7). SELC calls on City to maintain data center provisions. BirminghamWatch.

 

Terrell, F. (2026b, June 10). Crowds throng Birmingham City Hall as council passes data center rules.

BirminghamWatch.

Walter, W. (2026, June 10). Birmingham approves data center ordinance, facing public’s criticism. Alabama Political Reporter.