Structural Overview
Mobilization capacity is the operative link between ambition and outcome. Capital can be committed, policy can be declared, and private investment can be announced — but none of it translates into functioning infrastructure without the physical capacity to build. That capacity is what Pillar 2 measures.
The central question is not whether the United States can fund the AI infrastructure buildout. It clearly can. The question is whether it can translate capital, policy intent, and private investment into physical infrastructure at scale and at speed. The current evidence is mixed, and the trajectory is insufficient relative to the pace of AI demand acceleration.
The regime assessment for Pillar 2 is more optimistic than Pillar 1 — structural strength rates at 3, trend is gradually improving — but the improvement is insufficient relative to the demand acceleration documented in Pillar 1. Unless permitting reform and transmission buildout materially accelerate, the resource constraints of Pillar 1 remain binding.
Permitting and Regulatory Friction
Mine and Energy Project Timelines
The permitting environment in the United States is structurally misaligned with the pace of AI infrastructure demand. The average new copper mine requires approximately 17 years from discovery to production. Transmission lines routinely require 7 to 10 or more years. These timelines are not primarily a function of engineering complexity — they are a function of the multi-agency review process that operates at the intersection of federal, state, and local authority.
The deterioration is measurable and tracked. The median duration from generator interconnection request to commercial operation surged from under two years in 2008 to over five years by 2022 — a 150 percent increase in processing time with no corresponding increase in project complexity. Lawrence Berkeley National Laboratory's annual "Queued Up" report is the definitive public source for this signal, and the1 trajectory has not reversed. Litigation risk compounds the institutional delay: any project large enough to matter is large enough to attract legal challenge, making friction a structural feature rather than an edge case.
Fragmented Authority and Reform Attempts
Federal industrial policy initiatives exist and have expanded meaningfully in recent years. But federal intent does not override state and local zoning constraints. Virginia — the world's largest data center market — has seen municipalities tighten approval processes in direct response to community concerns about power usage and noise. Texas has shifted grid upgrade cost burdens onto large-load customers through SB6 and related legislation. These are not aberrations; they are the predictable output of a decentralized system.
FERC Order 2023 represents the most significant structural reform to the interconnection process in decades. Its shift to a "first-ready, first-served" cluster study process is designed to clear the queue of speculative projects and accelerate genuine capacity. Transmission providers now face strict study deadlines — 60 days for affected system restudies, 150 days for cluster restudies — with financial penalties for non-compliance.2 The reform is directionally correct. But it addresses queue management, not the underlying fragmentation of siting authority across federal, state, and local jurisdictions that remains the deeper constraint.
Transmission and Grid Buildout Capacity
Required Expansion versus Actual Pace
The simultaneous demands of AI data center expansion, broader electrification, and industrial reshoring require a transmission buildout that materially exceeds the current pace of deployment. The 2,600 gigawatt interconnection backlog documented in Pillar 1 is not merely a demand signal — it is a measure of institutional congestion. The system is not processing applications at the speed they are being submitted.
The DOE's 2023 National Transmission Needs Study quantifies the gap precisely. Median scenario results indicate the U.S. requires a 64 percent increase in within-region transmission capacity and a 114 percent increase in interregional transfer capacity by 2035 to meet current policy goals — with high-demand scenarios requiring interregional expansion of up to 412 percent.3 At the current build rate, large data center construction backlogs stand at 11 years.4 The math does not close without a fundamental change in deployment speed.
The AI infrastructure pivot since 2023 has compounded this pressure. AI now accounts for 50 percent of new data center orders — and AI-specific infrastructure requires higher power density and specialized liquid cooling solutions, shifting demand from general power management to high-density thermal management that the existing grid equipment supply chain was not sized to deliver.4
Labor and Supply Constraints
Grid scaling is constrained not only by permitting but by the physical inputs required to build. Skilled trades — lineworkers, electrical engineers, high-voltage technicians — face genuine shortages. Transformer manufacturing is a particular bottleneck: lead times have extended 2 to 3 times above 2020 levels due to fragmented regional manufacturing and import reliance.5 The U.S. faces a $3 trillion mega-project pipeline requiring a massive expansion of the craft-skilled workforce that the domestic labor pipeline does not yet supply at scale.6
Productivity gains are partially compensating. Major manufacturers have increased transformer output by approximately 50 percent while labor hours rose only 39 percent — meaning automation and process improvement are doing real work.6 But productivity-driven output growth has a ceiling that new labor formation does not. The craft-skilled platform remains the highest-risk constraint for project completion timelines across the mega-project pipeline.
discovery to production
increase needed by 2035 (DOE)
backlog at 2025 build rates
since 2020 — supply fragmentation
Defense Industrial Scaling
AI infrastructure expansion does not operate in isolation from the defense industrial base. Both draw on overlapping pools of capital, materials, skilled labor, and manufacturing capacity. As geopolitical risk rises, the competition between these demands intensifies.
Defense Production Constraints
Munitions production scaling has remained slow relative to the pace required by post-Ukraine threat assessments. U.S. shipbuilding capacity has diminished substantially relative to Cold War levels, and reconstituting it requires sustained multi-decade investment. Semiconductor fabrication ramp — the linchpin of both AI capability and defense electronics — remains dependent on the interplay of federal subsidy and the availability of highly specialized skilled labor that the domestic workforce pipeline does not yet reliably supply.
Defense-Energy Overlap
The electrification of military systems increases copper intensity in ways that directly compound the civilian demand pressure documented in Pillar 1. Rare earth dependencies introduce geopolitical fragility at the intersection of defense and clean energy supply chains. And defense budget expansion — while essential on security grounds — competes for the same fiscal space as infrastructure funding at a time when both are urgently needed.
Capital Allocation Efficiency
Mobilization capacity is not only physical. It is also a question of whether capital is being allocated with discipline or being distorted by policy incentives and demand projections that may not prove durable.
Industrial Policy Push
Recent years have produced a meaningful expansion of federal industrial policy instruments: bonus depreciation for AI infrastructure, incentives for semiconductor fabrication through the CHIPS Act framework, and a range of strategic mineral initiatives. These represent a genuine shift in the orientation of U.S. industrial policy toward physical infrastructure as a national priority.
Fiscal Guardrails and Ramp-Up Friction
Two constraints limit the translation of policy intent into deployed capital. First, most discretionary budget authority operates under caps established by the Fiscal Responsibility Act of 2023, with separate limits for defense and nondefense spending.9 Industrial policy ambition is real; the fiscal envelope constraining it is equally real.
Second, the private sector is experiencing significant ramp-up margin friction as it expands capacity to meet demand. Major infrastructure manufacturers are absorbing front-loaded costs for new factories and labor training that compress near-term margins even as backlogs reach record levels. Eaton projects a 130 basis point operating margin impact in 2026 specifically from capacity expansion ramps.10 This pattern — record backlog, near-term margin compression — is the setup that thesis-price divergence investors should monitor closely.
Distortion Risk
The risk is overinvestment. AI infrastructure buildout is driven by demand projections that are, by their nature, uncertain. If those projections overshoot, stranded infrastructure is the result — capital deployed at scale into assets that generate insufficient return. Political allocation compounds this risk: visibility often matters more than efficiency when public money is involved, leading to geographic or sectoral distributions that optimize for optics rather than output.
Institutional Competence versus China
The most important structural comparison for assessing U.S. mobilization capacity is not against historical benchmarks but against the alternative: China's centralized industrial mobilization model. The comparison is instructive precisely because it reveals asymmetric strengths and weaknesses that shape the competitive landscape.
| Dimension | United States | China |
|---|---|---|
| Decision model | Decentralized, market-led, litigation-prone | Centralized, state-directed, fast-executing |
| Capital markets | Deep, liquid, globally dominant | State-channeled, less transparent |
| Infrastructure speed | Slow — multi-year permitting typical | Fast — state override of local resistance |
| Innovation | Private sector-led, globally competitive | State-subsidized, rapidly improving |
| Regulatory fragmentation | High — federal / state / local conflict | Low — central authority overrides |
| Political risk | Polarization slows consensus infrastructure | Succession and opacity risk |
The U.S. advantage lies in deep capital markets, private sector innovation, and flexible enterprise. The weakness is structural: slow infrastructure approval, regulatory fragmentation, and political polarization that makes durable consensus on physical buildout difficult to sustain. Mobilization effectiveness — not technology or capital — ultimately determines whether structural constraints persist.
Investment Transmission Mechanism
The pace of mobilization directly shapes which assets benefit and which face continued pressure. The two scenarios — slow mobilization and accelerating mobilization — produce meaningfully different capital allocation implications.
| Scenario | Asset / Sector | Direction | Reasoning |
|---|---|---|---|
| Mobilization remains slow | Commodity producers (Cu, steel) | ↑ Benefit | Bottlenecks persist; pricing power strengthens |
| Mobilization remains slow | Defense primes | ↑ Benefit | Durable spending from geopolitical risk |
| Mobilization remains slow | Grid optimization software | ↑ Benefit | Maximize existing grid capacity when expansion is slow |
| Mobilization accelerates | Infrastructure engineering firms | ↑ Benefit | Direct exposure to buildout volume |
| Mobilization accelerates | Transmission equipment mfrs | ↑ Benefit | Transformer and conductor demand surge |
| Mobilization accelerates | Broad productivity equities | ↑ Benefit | Constraint relief enables broader AI value unlock |
| Mobilization accelerates | Commodity pricing power | ↓ Moderates | Supply response reduces structural scarcity premium |
Key Signals to Monitor
The regime is gradually improving but remains insufficient. The following indicators provide the earliest evidence of whether mobilization is accelerating toward adequacy or remaining structurally constrained.
- 01 Median interconnection timeline (LBNL "Queued Up") — easing if trending below 5 years Annual
- 02 Transmission circuit-miles energized annually vs DOE 2035 need targets Annual / DOE
- 03 Transformer lead times (ETN / GEV quarterly supplements) — easing if below 12 months Quarterly
- 04 Quanta / GEV backlog liquidation rate vs margin trend — labor constraint proxy Quarterly
- 05 FERC Order 2023 cluster study completion rates and penalty enforcement Quarterly / FERC
- 06 Data center construction backlog rate — tightening if queue exceeds 11-year pace Semi-annual
Primary data sources: DOE National Transmission Needs Study (2023), LBNL "Queued Up" annual interconnection series, FERC Order 2023 implementation filings, and quarterly earnings supplements from GE Vernova, Eaton Corporation, and Quanta Services. The trajectory of permitting reform and grid scaling will determine whether resource constraints become chronic or temporary.
- Lawrence Berkeley National Laboratory, "Queued Up: Characteristics of Power Plants Seeking Transmission Interconnection," annual series (most recent: 2024). Interconnection timeline deterioration, 2008–2022. Publicly available at emp.lbl.gov. § II
- Federal Energy Regulatory Commission, Order No. 2023, "Improvements to Generator Interconnection Procedures and Agreements" (July 2023); CITAP Program implementation filings. First-ready first-served cluster study mandate; 60/150-day deadlines; financial penalties. § II
- U.S. Department of Energy, "National Transmission Needs Study" (October 2023). 64% within-region and 114%–412% interregional transfer capacity requirements by 2035. Publicly available at energy.gov. § III
- Eaton Corporation, Q4 2025 Earnings Supplement and Investor Presentation (February 2026). 11-year data center construction backlog at 2025 build rates; 50% of data center orders for AI infrastructure; 54% of mega-project activity in data centers; $15.3B electrical backlog; 130bp margin impact from capacity ramps. § III, § V
- McKinsey & Company, "Global Energy Perspective 2023"; supplemented by Eaton and GE Vernova quarterly earnings commentary on lead times. Transformer lead times extended 2–3× above 2020 baseline; pre-pandemic range 4–16 months; current range 15–24+ months. § III
- Eaton Corporation, Q4 2025 Earnings Supplement; Quanta Services Q4 2025 Earnings Release. $3 trillion mega-project pipeline figure; 50% output increase on 39% labor hour growth as productivity proxy. § III
- GE Vernova, Q4 2025 Earnings Release and Investor Presentation (February 2026). $150B total order backlog; $2B+ in data center orders (3× prior year); 24 GW new gas contracts Q4 2025; Prolec GE acquisition adding $3B revenue and doubling transformer output. § III, § IV
- Quanta Services, Q4 2025 Earnings Release and Remaining Performance Obligations disclosure (February 2026). $44B total backlog; $36.2B Electric Infrastructure Solutions segment backlog; Wilson Construction and Tri-City acquisitions expanding craft-skilled platform. § III, § VIII
- Fiscal Responsibility Act of 2023, Pub. L. 118-5 (June 2023). Discretionary budget caps for FY2025; separate defense and nondefense spending limits. § V
- NERC, Electricity Supply and Demand (ES&D) Database; U.S. DOE, "National Transmission Needs Study" (2023). Circuit-miles energized annually; 2011–2020 decadal baseline for easing threshold. Primary monitoring source for § VIII Signal 02. § VIII