The 2,000 GW Backlog Exposing America’s Grid Infrastructure Crisis

Electricity demand across the United States is climbing toward record highs. The infrastructure meant to move that power is more congested and more underfunded than at any point in modern memory.

That contradiction is not a policy abstraction. It is a structural bottleneck that is already delaying projects, repricing development pipelines, and loading asymmetric risk into U.S. power markets.

The International Energy Agency (IEA) frames the scale bluntly: global annual grid investment, currently around $400 billion, must rise by roughly 50% to approximately $600 billion per year to meet 2030 demand. That implies a persistent annual shortfall of around $200 billion.

In the United States, the shortfall shows up as backlog. According to Lawrence Berkeley National Laboratory (LBNL), 2,061 GW of generation and storage capacity sat in interconnection queues at the end of 2025, median wait times now exceed five years, and more than 1.8 TW of projects have been withdrawn since 2020. Data centres have ended a decade of flat U.S. power consumption and become the dominant driver of long-term demand growth, a demand shock the grid was never planned to absorb.

Here is the breakdown that matters for anyone with capital exposed to U.S. power: where the bottleneck comes from, what regulators are doing about it, and what the grid infrastructure crisis means for energy investment risk and opportunity through 2035.

A 2,000 GW backlog: how U.S. interconnection queues became a structural crisis

The number itself is the story. At the end of 2025, LBNL’s “Queued Up: 2026 Edition” counted 2,061 GW of active generation and storage capacity waiting for grid connection across roughly 8,200 projects.

To put that in perspective, that queued capacity is larger than the entire installed generating fleet of the United States. The system is being asked to connect more than it has ever physically built, all at once.

The composition reveals where capital wants to go. The LBNL queue breaks down as follows:

• Solar: 773 GW
• Storage: 749 GW
• Gas: 253 GW, up 86% year-over-year
• Wind: 220 GW

That 86% jump in gas is not a quiet technical shift. It tracks directly with Trump Administration policy signals favouring natural gas as the preferred source for meeting new electricity demand, and it tells you the forward generation mix is bending under political pressure, not just developer economics.

The queue’s sheer size also overstates real demand, because much of it never gets built. Withdrawals exceeded 750 GW in 2025 alone, and more than 1.8 TW has been pulled from queues since 2020, predominantly renewables and storage. That withdrawal rate is effectively a live measure of how many projects are uneconomic to finish once upgrade costs and multi-year delays are priced in.

Blocked data centre capacity compounds the withdrawal problem: Goldman Sachs projects U.S. data centre electricity demand will hit 66 GW by 2027, more than double the 2026 baseline, yet over $170 billion in planned capacity has been stalled by the same interconnection constraints driving the 1.8 TW withdrawal figure.

For projects that did reach commercial operation in 2025, the median wait from interconnection request to switch-on topped five years. Queue exposure, in other words, is no longer a background planning detail. It is a pricing variable in any U.S. generation development thesis.

Regional queue dynamics: where the pressure is most acute

The pressure is not evenly distributed. ERCOT, the Texas grid operator, saw its active queue climb to roughly 421 GW by mid-2026 from 245 GW in 2023, a 72% jump in three years.

That surge saturated an energy-only market fast enough to trigger a temporary pause on connecting new data centre projects to the Texas grid. When a demand shock can force a connection freeze in the fastest-growing power market in the country, it tells you how little slack the system is carrying.

CAISO sits at the opposite pole. California’s operator roughly halved its active queue between 2023 and 2026 through aggressive reform, a genuine processing win. The caveat, as analysts note, is that fast triage risks disqualifying viable projects alongside the speculative ones, so a shrinking queue is not automatically a healthier one.

Three structural failures driving the bottleneck

Knowing how large the backlog is explains the symptom. It does not explain the disease. The bottleneck persists because three structural failures have compounded over years, each reinforcing the others.

1. Generation growth has outrun transmission. Developers have proposed vastly more capacity than the physical grid can connect, because transmission build-out has not kept pace with the generation pipeline. Insufficient grid capacity is driving higher congestion and slowing the deployment of new generation, storage, and demand resources, according to the IEA.
2. The planning framework is reactive by design. The existing interconnection process evaluates projects one at a time rather than planning a grid for the future resource mix over multi-decade horizons. That short-term framework was never built to absorb a generational demand surge, and the U.S. Energy Information Administration (EIA) has identified data centre load as the dominant long-term driver of American electricity demand growth.
3. Cost allocation is gridlocked. State regulators, FERC, and regional transmission organisations (RTOs) have repeatedly clashed over who pays for major upgrades, stalling investment even when the physical and economic case is clear.

IEA framing Insufficient grid capacity is creating higher levels of network congestion and slowing the deployment of new electricity generation, storage, and demand resources.

That third failure carries the sharpest investor consequence. A project with strong standalone economics can be rendered unviable by an upstream transmission dispute it has nothing to do with.

PJM describes this as a “transition gap,” where retiring thermal generation and surging data centre load outpace the speed at which new resources can connect. The attrition it produces is measurable: seven ISOs collectively withdrew 460 GW of projects between 2023 and 2026, per S&P Global.

Emergency grid directives have become a routine rather than exceptional tool: the Department of Energy issued its seventh order to PJM alone in 2026, commandeering backup generators across the Mid-Atlantic, a pattern that quantifies what the transition gap looks like in operational terms rather than planning documents.

What this tells you is that these are not administrative delays that clear with better paperwork. They are structural, which reframes the timeline entirely. Resolution demands regulatory reform and capital mobilisation at a scale measured in years, not quarters, and that distinction should anchor any assumption about how fast the system can heal.

What FERC Order 1920 does, and what it cannot fix

The regulatory response is real, and it is significant. It is also structurally incomplete, which matters more for your risk horizon than the headlines suggest.

FERC Order 1920 is the Commission’s landmark long-term transmission planning and cost-allocation rule, published in the Federal Register on 11 June 2024 and effective 12 August 2024.

How the rule works

The order replaces the reactive, project-by-project approach with proactive planning. Transmission planners must now conduct long-term regional planning at least every five years across a 20-year horizon.

It introduces structured cost-allocation methods, directly targeting the gridlock that has stalled upgrades for years. It also requires planners to consider “right-sizing” existing facilities rather than defaulting to minimal patches.

Taken together, these provisions are the most consequential grid planning reform in decades. The problem is not the design. It is the clock.

The implementation timeline reality

Compliance filings are staggered across ISOs well beyond the rule’s effective date, and many deadlines have already slipped in practice.

The practical read is uncomfortable. If new planning cycles only take effect as late as 2028 in some regions, then capital deployed in projects today is still operating under the old framework.

Order 1920’s benefits arrive too late to de-risk assets needing interconnection decisions in the next two to three years. It functions as a medium-term structural improvement, not a near-term risk mitigant, and you should not price its benefits into short-horizon development timelines.

The gap Order 1920 cannot bridge

Even assuming full compliance, three constraints sit entirely outside the rule’s reach.

The first is physical supply. Gas turbine supply constraints and record-high infrastructure costs are complicating heavy reliance on gas-fired generation, and no planning rule can manufacture turbines.

The second is policy fragmentation. There is no federal mandate on renewable targets, so state-level responses remain inconsistent, leaving the development environment patchy by design rather than by transition.

The third is pace itself. As Wood Mackenzie analysts frame it, the tension between bringing capacity online quickly and keeping it affordable is driving reform across regional markets, but no single comprehensive solution exists. A well-designed framework still cannot compress the multi-year lead times that transmission construction physically requires.

Investment implications: pricing the grid constraint into U.S. energy exposure

Diagnosis is useful only if it changes how you position. The grid constraint reprices risk unevenly, and that unevenness is where the opportunity and the exposure both sit.

• Reliability and capacity-shortfall risk. The “transition gap” creates credible capacity shortfall risk in PJM and ERCOT, where retiring thermal generation and surging load collide. That should factor directly into how you assess utility reliability exposure in those regions, with PJM the clearest named example in a major RTO.

The PJM capacity shortfall is not a hypothetical: the most recent capacity auction cleared at its $325/MW-day price cap and still produced a 6.8 GW reserve gap, with data centres accounting for roughly $6.3 billion of the $16 billion total cost, a live demonstration of how thermal retirements and surging load interact in the transition gap the article describes.

• Stranded-development risk. With more than 1.8 TW withdrawn since 2020 and attrition running high, developers without secured interconnection or behind-the-meter solutions carry a materially higher risk profile. Projects with secured interconnection command valuation premiums over queue-exposed assets, and that spread should be reflected in how development-stage assets are valued.
• Transmission and grid-enabling opportunity. Infrastructure that moves power benefits from grid stress regardless of which fuel mix ultimately wins.

The data points that reframe the picture are counterintuitive. Despite the renewable transition narrative, Wood Mackenzie projects gas will supply the majority of incremental U.S. power.

Wood Mackenzie projection Natural gas is projected to supply 52% of the incremental U.S. power generation needed through 2035.

Regional concentration sharpens the point. The EIA’s September Short-Term Energy Outlook projects the West South Central region, dominated by Texas, will represent roughly 20% of U.S. electricity sales growth in 2026 and approximately 40% in 2027.

What that tells you is that even inside a queue-saturated, capital-constrained system, specific regional and fuel-type exposures carry materially different risk-reward profiles. Queue position, behind-the-meter status, and regional footprint have become first-order valuation inputs, not secondary considerations.

Where opportunity sits in a constrained grid environment

Three positions carry structural advantage, and it is worth holding a clear mental map of which benefit from grid stress versus which depend on queue outcomes.

Gas and gas infrastructure sit on the demand-clarity side, backed by the 52% incremental-supply projection and explicit policy support, though not immune to turbine supply constraints.

Transmission infrastructure operators benefit from volume regardless of the generation mix, because every fuel scenario still requires more wires. Grid-enabling technology companies sit closest to the regulatory reform cycle, with a demand signal tied to system-wide change rather than any single project clearing its queue.

This is directional positioning, not a recommendation to buy any specific name. The point is that queue-dependent generation assets and grid-enabling infrastructure now sit on opposite sides of the risk ledger.

What the $200 billion gap means for the energy transition timeline

Pull the threads together and a revised picture of the transition’s pace emerges, one that reconciles the optimistic demand trajectory with the physical limits on supply.

The IEA’s $200 billion annual investment shortfall is not a financing abstraction. It is a direct statement about how much slower U.S. and global capacity additions will run than demand-side models assume, and that gap widens every year the money fails to arrive.

The decarbonisation dimension carries its own risk. Renewables and storage make up the majority of the more than 1.8 TW withdrawn since 2020, and if that pattern holds, states and utilities will miss statutory clean-energy targets.

For investors whose business models assume a specific decarbonisation trajectory, that attrition translates into real policy and regulatory risk. Anchoring a transition timeline to demand forecasts alone, without accounting for the supply-side grid constraint, systematically underprices delay risk and overprices how fast clean capacity can displace thermal generation.

The grid bottleneck limiting nuclear deployment in the western United States illustrates how the same interconnection constraints shaping the 2,061 GW queue extend beyond renewables and gas into every generation technology seeking connection, making the physical grid rather than any individual fuel policy the binding constraint on the transition.

IEA investment benchmark Global annual grid investment must rise by roughly 50%, from about $400 billion to $600 billion, to meet 2030 power demand.

IEA Director of Energy Markets and Security Keisuke Sadamori has stressed that expanding system flexibility and resilience is critical alongside raw grid expansion. Four variables will determine whether the constraint tightens or loosens through 2035:

1. The pace of Order 1920 implementation
2. Resolution of gas turbine supply constraints
3. State-level policy consistency
4. How fast co-location and behind-the-meter solutions scale as workarounds

This article is for informational purposes only and should not be considered financial advice. Investors should conduct their own research and consult with financial professionals before making investment decisions.

Three variables that will set the pace of U.S. grid resolution through 2035

Rather than a closed conclusion, treat the road ahead as a monitoring framework. Three variables function as leading indicators, and when each moves materially, the risk profile of U.S. generation and transmission assets moves with it.

1. Order 1920 implementation pace. Whether ISOs file on time and new planning cycles actually begin by 2028 as targeted will decide whether the regulatory framework finally catches up to the demand shock or stays chronically reactive. Slippage here keeps today’s projects under the old, slower regime for longer.
2. Gas turbine supply and infrastructure costs. Wood Mackenzie projects gas at 52% of incremental U.S. power through 2035, but that outlook is contingent on supply chains, not just policy support. Turbine bottlenecks and record infrastructure costs could delay even well-queued gas projects, compressing the reliability buffer gas is expected to provide.
3. State-level policy consistency. With no single comprehensive federal solution, the investment environment stays fragmented. States that resolve transmission cost-allocation quickly and clearly should attract disproportionate capital, making this a structural feature rather than a passing transitional condition.

Watching these three lets you separate episodic market noise from genuine structural shifts in the grid constraint. That is the difference between reacting to headlines and recognising when the underlying picture has actually changed.

Financial projections are subject to market conditions and various risk factors, and these forward-looking statements are speculative and subject to change based on market and regulatory developments.