AI data center construction faces local backlash, with power, costs, and timelines becoming the key constraints

Overview

This report discusses the conflict between AI data center expansion and the local political backlash in the United States. Morgan Stanley points out that community opposition to data center construction accelerated significantly in 2026, mainly through local legislation, moratoriums, zoning adjustments, cost-sharing requirements, ratepayer protections, and environmental disclosure rules. The report believes this opposition will not simply halt US AI infrastructure construction, but it will raise costs, extend timelines, and push projects to disperse toward regions with land, power, and more supportive policies.

Core views

The core views include: first, the political risk of data center construction has become an important variable in whether AI capex can be realized; second, most local- and state-level actions are pauses rather than permanent bans, with the focus on slowing the pace, demanding concessions, and reallocating costs; third, residents are most concerned about electricity prices, grid stress, water resources, waste heat, noise, dust, and traffic; fourth, if opposition continues to spread, the 2026 AI capex estimate of $877B may face downside revision risk; fifth, power bottlenecks and permitting delays will increase the importance of on-site generation, demand response, energy storage, and grid support services; sixth, the macro impact is more likely to appear as regional, sticky electricity inflation rather than a sudden nationwide power price shock.

Analysis framework

The report uses a combination of thematic research, policy tracking, consumer surveys, geographic distribution analysis, and cross-asset mapping. The research team tracks data center moratoriums and regulatory actions at the local, state, and federal levels, and combines Morning Consult consumer survey data, Data Center Watch project delay data, Data Center Watch and Interconnected Capital moratorium tracking, and 451 Research data center location data to assess the impact on equities, credit, securitized credit, municipal bonds, inflation, and US AI competitiveness.

Methodology notes

• Policy and regulatory trackingThree-level policy action framework across local, state, and federal levels

  Data center moratoriums and regulatory guardrails

  The report breaks opposition actions down by local government, state government, and federal level, and distinguishes among policy tools such as temporary pauses, cost-sharing, ratepayer protections, environmental disclosure, and zoning adjustments.

• Market sentiment analysisConsumer survey dimensional analysis

  Residents' net attitude toward the impact of AI data centers

  The report cites Morning Consult survey data, comparing consumer positive and negative views on dimensions such as electricity prices, the power grid, water prices, and the environment, while also tracking changes in preferences between halting construction and continuing construction.

• Cross-asset impact mappingFramework linking equities, credit, securitized credit, municipal bonds, and macro inflation

  Transmission of construction slowdowns across asset classes

  The report maps data center construction delays separately to on-site generation, utilities, data center REITs, corporate bond supply, ABS/CMBS refinancing risk, municipal bond supply, and regional electricity inflation.

• Operational efficiency frameworkThree levers: utilization, power, and water resources

  Improving data center efficiency and easing community concerns

  The report proposes improving utilization of existing capacity, adopting on-site generation and grid support services, and increasing water efficiency to reduce pressure for new construction and improve community acceptance.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• SEI, INIO, BE

  Beneficiaries of on-site generation and data center power bottlenecks

  Strengths

  The report believes data centers are turning to on-site generation because of grid interconnection and power shortages, which benefits these OW names.

  Weaknesses

  The benefit depends on data center customer capex and the pace of deployment for on-site power solutions.

  Comparison

  Compared with traditional grid expansion, on-site generation can shorten waiting times of more than 5 years in some regions.

  Risks

  If policy restrictions spread to on-site generation, fuel supply, or emissions permits, the magnitude of the benefit may decline.

• Publicly traded colocation data center REITs such as EQIX and DLR

  Relatively defensive data center assets

  Strengths

  The report believes they are smaller than hyperscale AI training data centers and are viewed as critical infrastructure, so they are relatively less vulnerable to local political backlash.

  Weaknesses

  They may still be affected by electricity prices, permitting, tax incentive adjustments, and regional supply-demand changes.

  Comparison

  Compared with large self-built training centers by hyperscalers or Neoclouds, colocation REITs have lower political sensitivity.

  Risks

  If local regulation expands from large training centers to all data center projects, the defensive characteristics will weaken.

• US power and utilities

  Core absorbers of data center load growth and cost allocation

  Strengths

  Large-load customers may bear more infrastructure construction and upgrade costs, giving utilities investment expansion opportunities.

  Weaknesses

  Regulatory uncertainty, residential ratepayer protection, and political risk may compress returns.

  Comparison

  The degree of benefit depends on whether the regulatory framework allows cost recovery, rather than simply on load growth.

  Risks

  Rising electricity prices could trigger residential backlash, leading to stricter ratepayer protections and project restrictions.

• Corporate credit bonds

  Financing channel for AI infrastructure capex

  Strengths

  If construction timelines are extended, forward supply may decline, helping ease long-term issuance pressure.

  Weaknesses

  Front-loaded near-term capex may continue to increase corporate bond issuance pressure.

  Comparison

  The AI ecosystem has contributed nearly the main part of incremental corporate bond supply this year, so policy disruptions have a significant impact on the supply narrative.

  Risks

  If projects are delayed but financing has already been front-loaded, this may increase fundamental exposure and refinancing pressure.

• ABS and CMBS

  Existing securitized transactions are relatively insulated from early-stage data center political risk

  Strengths

  Slower construction can reduce future oversupply risk, especially benefiting existing deals ahead of key refinancing windows.

  Weaknesses

  If the underlying assets are directly exposed to early-stage projects lacking power access or permits, risk may still rise.

  Comparison

  Compared with financing for new projects, existing transactions are less affected by local moratoriums.

  Risks

  If data center leasing demand or valuations weaken due to downward revisions in AI capex, securitized assets may still come under pressure.

• Municipal bonds

  Financing tool for local infrastructure, tax incentives, and energy prepayment

  Strengths

  More balanced tax incentive arrangements may increase local property tax revenue, which is beneficial for credit quality.

  Weaknesses

  Higher project costs may raise municipal financing demand and supply.

  Comparison

  The report believes local backlash is mildly positive for municipal credit, but it could also bring about $100B of incremental supply.

  Risks

  If local fiscal concessions are too large or projects are canceled, expected tax revenues may fail to materialize.

• Private credit, BDCs, and loans

  Relatively benefiting sectors under a slowdown in AI disruption

  Strengths

  If compute buildout slows, the AI disruption timeline for some companies shifts back, giving weaker financing entities more time to refinance.

  Weaknesses

  This logic depends on persistent compute constraints and cannot offset all fundamental pressure.

  Comparison

  The report says private credit, BDCs, and loans that have lagged since the start of the year may become an anti-AI trade.

  Risks

  If AI application diffusion outpaces compute construction, disruption pressure may still emerge earlier.

• US macroeconomy and inflation

  Data center load growth affects regional electricity prices and household purchasing power

  Strengths

  The impact is more likely to be regional and gradual, not a sudden nationwide shock.

  Weaknesses

  Electricity CPI may remain at 4% to 5% YoY in the medium term, and the household burden is regressive.

  Comparison

  Evidence is stronger in data center-dense areas such as South Atlantic/Northern Virginia.

  Risks

  If grid expansion lags load growth, regional inflation and political backlash may reinforce each other.

• US AI competitiveness

  Data centers are the physical infrastructure foundation of AI competition

  Strengths

  Geopolitical and national security realities reduce the probability of major construction obstruction at the federal level.

  Weaknesses

  Continued scrutiny at the local and state levels will delay projects, add conditions, and alter construction pathways.

  Comparison

  The report links the pace of US construction with its competitive position against China in AI.

  Risks

  If local opposition becomes permanent and expands to the federal level, the US may struggle to maintain the pace of AI infrastructure construction.

Key data

• Report date2026-07-14The report cover shows July 14, 2026 04:01 AM GMT.
• Scale of canceled or delayed projects in 2025About $156BEstimated using third-party data such as Data Center Watch.
• Scale of canceled or delayed projects in 1Q26$130BThe report states that $130B of projects had already been canceled or delayed in 1Q26.
• 2026 AI capex estimate$877BMorgan Stanley estimate; expanding local opposition could cause actual spending to fall below this level.
• Hyperscaler and Neocloud capex forecastMore than $870B in 2026 and more than $1.3T in 2027The report says analysts continue to see upside bias in capex expectations.
• Consumer support for halting constructionAbout 45%In the May 2026 Morning Consult survey, the share supporting halting construction exceeded the share supporting continued construction and expanded energy supply for the first time.
• Consumer support for continuing construction and expanding capacityAbout 38%Under the same survey methodology, this was below the share supporting halting construction.
• Medium-term US electricity CPI expectation4% to 5% YoYThe report believes data center demand will bring stickier, more regionalized electricity inflation pressure.
• Potential power shortfallAbout 38GW by 2028Based on this, the report concludes data centers will be more inclined toward on-site generation.
• Grid connection waiting time in some regionsMore than 5 yearsLong lead times increase the attractiveness of on-site generation solutions.
• Potential incremental municipal bond supplyAbout $100BThe report believes higher costs and more renegotiation of tax incentives may drive municipal bond financing demand.

Impact & implications

From an investment perspective, the political backlash turns data center construction from a simple capex expansion issue into a rebalancing among power, local public finance, residential electricity prices, environmental permitting, and national competitiveness. At the equity level, companies related to on-site generation and grid support benefit, while existing colocation data center REITs are relatively resilient; at the credit level, front-loaded near-term capex may increase issuance pressure, but persistent opposition could ultimately reduce future supply; at the securitized credit level, existing ABS and CMBS have limited exposure to early-stage development political risk, and slower construction may actually help reduce oversupply; at the macro level, the impact is more concentrated in electricity prices and household burdens in data center-dense regions.

Risks

• Local moratoriums shift from temporary measures to long-term or permanent restrictions.
• State-level or federal-level policies expand, causing systemic slowing of AI data center construction.
• Concerns over residential electricity prices, grid stress, water resources, waste heat, noise, dust, and traffic continue to strengthen community opposition.
• AI capex comes in below Morgan Stanley's 2026 estimate of $877B.
• Power interconnection, equipment, and labor bottlenecks drive costs even higher.
• On-site generation solutions face risks related to permitting, emissions, fuel supply, or community acceptance.
• Regional electricity inflation intensifies household burdens and triggers stronger ratepayer protection policies.
• Front-loaded near-term capex leads to rising issuance pressure in credit markets.

What to watch

• The number, duration, and scope of data center moratoriums at the local, state, and federal levels.
• Whether policy actions continue to focus on pauses and guardrails, or evolve into permanent bans.
• Changes in resident attitudes toward halting construction versus continuing construction in surveys such as Morning Consult.
• Whether the amount of canceled or delayed projects tracked by Data Center Watch continues to rise.
• 2026 and 2027 capex guidance from hyperscale cloud providers and Neoclouds.
• Residential electricity prices and wholesale power prices in data center-dense regions such as South Atlantic/Northern Virginia.
• The adoption pace of on-site generation, energy storage, demand response, and grid support services.
• Subsequent rating, earnings, and order changes for related names such as SEI, INIO, BE, EQIX, and DLR.
• Incremental municipal bond supply, renegotiation of tax incentives, and realization of local fiscal revenues.
• How US policymakers balance AI competition, national security, and community costs.