Data center expansion faces dual bottlenecks from power and communities, benefiting assets with the ability to “secure power quickly”
AI summary card
Data center expansion faces dual bottlenecks from power and communities, benefiting assets with the ability to “secure power quickly”
Morgan Stanley believes that state scrutiny and community resistance toward US data centers are intensifying, but the power shortfall driven by computing demand continues to support medium- to long-term opportunities for Powered Shell Providers, on-site generation, and certain power assets.
- Texas is auditing data center projects in the ERCOT interconnection process; approximately 90% of more than 474GW of pending interconnection requests are data-center-related.
- Virginia is advancing the “cost causer pays” principle, requiring large new loads to bear clearer funding responsibility for dedicated transmission facilities.
- Base Load designation in ERCOT Batch Zero can be viewed as a green light for interconnection; Studied Load faces system upgrades, capacity uncertainty, and risks of prolonged studies.
- The report estimates a 38GW US data center power gap in 2026—2028; even after incorporating rapid power-delivery solutions, a 1—11GW gap may remain.
- Preferred beneficiaries include CIFR, GLXY, and VST; grid uncertainty also benefits on-site power solutions such as SEI, BE, and INIO.
Report interpretation
Overview
The report updates the policy, community, and grid constraints facing US data center expansion and summarizes conclusions from expert discussions. The core view is that community opposition has evolved from a pure communications issue into a project-development constraint; meanwhile, AI computing-driven power shortages continue to create structural opportunities for companies with grid-connected resources or rapid power-delivery capabilities.
Core views
Texas project audits are more likely to delay high-quality projects by 1—3 months rather than weaken overall demand. Regulatory changes in Virginia increase the share of dedicated infrastructure costs borne by large data centers. Projects qualifying as Base Load in ERCOT Batch Zero will materially improve interconnection certainty and may serve as catalysts for Powered Shell Providers. For conventional power assets, near-term project delays depress forward power-price expectations, but long-term supply-demand tightness continues to support generation and contracting opportunities.
Analysis framework
The report conducts thematic research by combining state-level policy updates, ERCOT interconnection processes and timelines, expert interviews, project capacity and deposit data, as well as power supply-demand gap and EV/Watt valuation comparisons.
Methodology notes
Incorporates political and community risks into project-development assessments alongside land, power, and labor.
Distinguishes water, emissions, and land issues that can be mitigated through engineering or incremental investment from social opposition that is more difficult to address through project design; it also uses early municipal meeting records and local social media as leading risk indicators.
Uses Base Load and Studied Load classifications to determine project interconnection certainty.
Base Load represents mature and sufficiently committed load that can generally advance toward interconnection; Studied Load must undergo system-upgrade studies and is not assured of all requested capacity or an interconnection timeline.
Assesses Powered Shell Providers based on 15-year leases, 15% capital expenditure returns, and $15 of equity value creation.
The report uses this framework to conclude that the potential value of PSP assets remains insufficiently priced by the market and compares them with mature data center operators using EV/Watt.
Asset mapping & comparison
Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).
- CIFRPreferred beneficiary of ERCOT Batch Zero and Powered Shell Providers
- Strengths
- The report expects its approximately 2GW interconnection request may qualify as Base Load, with existing power resources that can be converted to data center use.
- Weaknesses
- Valuation and project advancement depend heavily on interconnection classification and execution of AI leases.
- Comparison
- Relative to GLXY, the report has a stronger preference for CIFR based on Base Load results and thematic positioning.
- Risks
- If classified as Studied Load, if audits are prolonged, or if customer demand weakens, value realization will be delayed.
- GLXYPowered Shell Providers beneficiary
- Strengths
- Considered relatively likely to receive a favorable Batch Zero notification and to benefit from converting existing power resources to high-performance computing.
- Weaknesses
- Uncertainty remains regarding interconnection status and development execution.
- Comparison
- Like CIFR, it is a potential Base Load beneficiary, but the report explicitly identifies CIFR as the preferred pick.
- Risks
- Study-queue, system-upgrade spending, and lease-execution risks.
- VSTKey power-generator beneficiary of rising ERCOT electricity demand
- Strengths
- The report believes that if larger projects enter Base Load, demand confirmation would support forward power prices and contracting opportunities for existing units.
- Weaknesses
- Recent weakness in ERCOT forward prices for 2027—2028 creates marginal valuation pressure.
- Comparison
- The report believes VST is best positioned for upward power-price dynamics.
- Risks
- Data center project delays, weaker forward prices, and policy uncertainty.
- SEI、BE、INIOBeneficiaries of on-site generation and rapid power-delivery solutions
- Strengths
- Grid interconnection uncertainty can drive data centers to adopt on-site power and create synergies with PSP or Allocated-category projects.
- Weaknesses
- Project economics, equipment deployment, and fuel supply still require validation.
- Comparison
- Compared with projects relying on long-term grid upgrades, on-site power offers a faster time-to-power advantage.
- Risks
- Equipment delivery, permitting, cost, and project customer-execution risks.
- EQIX、DLRColocation data center operators
- Strengths
- Their scale, critical-infrastructure status, and community-engagement experience make them relatively defensive; both have accelerated development and raised capital expenditure guidance.
- Weaknesses
- Development returns and revenue growth still depend on sustained expansion and successfully securing capacity.
- Comparison
- Compared with greenfield or smaller projects, mature operators have greater resilience to local opposition.
- Risks
- If political resistance directs incremental capacity to lower-friction US regions or international markets, growth and development returns may be constrained.
Key data
- ERCOT pending interconnection requestsMore than 474GWApproximately 90% are data-center-related.
- Potential delay for high-quality ERCOT projects1—3 monthsThe report judges that the audit primarily affects timing, while the April 9, 2027 study-results date remains unchanged for now.
- Preliminarily qualified large loads in Batch ZeroApproximately 205GWStill subject to confirmation of financial deposits.
- Expected Base Load project scaleApproximately 65GWThe report believes the scale could exceed market expectations.
- US data center power gap in 2026—202838GWEstimate before incorporating rapid power-delivery solutions.
- Gap after incorporating rapid power-delivery solutions1—11GWIndicates that the power bottleneck is not fully eliminated.
- ERCOT forward power prices and incentive level for new baseloadApproximately $50/MWh, versus more than approximately $95/MWhThe report believes greater Base Load demand could support forward power prices.
- PSP EV/Watt range$1.44—$5.09 per wattThe report continues to assume these companies can trade at $15 per watt.
- WULF—Anthropic transaction net value creationApproximately $19 per wattThe report estimates that the project requires approximately $3 per watt of equity funding.
Impact & implications
Investors should distinguish demand risk from interconnection-timing risk: tighter policies and community resistance raise requirements for project selection, siting, and cost of capital, but do not necessarily weaken AI-related power demand. Companies with mature interconnection qualifications, reusable power infrastructure, or rapidly deployable on-site generation have stronger relative advantages; large colocation data center operators are relatively protected by their scale, critical-infrastructure status, and community-engagement experience, although they still face the risk of capacity migrating to lower-friction regions in the US or overseas.
Risks
- Community opposition may translate into long-term bans, zoning restrictions, or stricter environmental requirements.
- Large-load “cost causer pays” mechanisms may increase project capital expenditures and interconnection costs.
- ERCOT audits and Batch Zero classifications may delay projects, with Studied Load projects particularly exposed to long-term uncertainty.
- Forward power-price volatility may affect near-term earnings expectations for independent power producers and utilities.
- Data center development may migrate to lower-population, colder regions or overseas, potentially changing beneficiary regions and asset allocation.
- PSP valuations depend on long-term AI leases, interconnection qualifications, and capital expenditure returns; weaker-than-expected execution would create valuation risk.
What to watch
- ERCOT is expected to release Batch Zero designation results by September 2, 2026 at the latest.
- Progress in projects remedying dynamic-model or data deficiencies by August 31, 2026.
- Allocated MW and commercial operation dates in the ERCOT Batch Zero study report on April 9, 2027.
- The duration of the Texas audit, compliance requirements, and execution by large developers.
- Virginia's final treatment of system-level transmission-upgrade costs and GS-5 large-load rates.
- Policy developments in Ohio and other states regarding data center taxes, environmental standards, resident compensation, and approval moratoriums.
- ERCOT forward power prices, data center lease economics, and on-site generation project orders.