📊 Full opportunity report: The queue. Why the grid, not the chip, is the binding constraint on AI. on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
The primary constraint on AI infrastructure expansion has shifted from semiconductor chip availability to grid interconnection delays. This causes a bifurcation in data-center development, with private solutions bypassing the grid, but shifting costs onto ratepayers. The situation is reshaping geography, costs, and policy debates.
The main constraint on the expansion of AI data centers in the US has shifted from chip shortages to the interconnection queue for the electrical grid, with delays averaging five years or more. This change is reshaping the geography of data-center deployment, increasing costs, and creating political tensions over who bears the burden of grid expansion.
For two years, the dominant narrative was that chip shortages limited AI infrastructure growth. However, recent data shows that the bottleneck has moved to the grid, with approximately 2,300 to 2,600 gigawatts of generation and storage capacity stuck in US interconnection queues. The median wait time for projects to reach commercial operation has increased from under two years in 2008 to nearly five years today, with some projects facing delays up to twelve years.
Demand for power from data centers is surging, with US projections reaching 76 gigawatts in 2026, up from 50 gigawatts in 2024. Globally, data-center energy consumption could exceed 1,000 terawatt-hours annually by the early 2030s. In Texas, interconnection requests have increased by 700% in a single year, from 1 gigawatt to 8 gigawatts, illustrating the scale of the demand wall.
As a result, capital is increasingly bypassing the grid by building private power sources such as behind-the-meter gas plants and co-located nuclear facilities. Companies like Microsoft are restarting nuclear plants like Three Mile Island to secure baseload power, effectively sidestepping the grid constraints. However, this shift shifts costs onto ratepayers, with utilities and policymakers raising concerns about the political and economic impacts of cost externalization.
The queue.Why the grid, not the chip,
is the binding constraint on AI.
more than total installed capacity
up to 12 years for data centers
vs grid access maybe 2035
ratepayers · the cost-shift, concrete
in a single year
Virginia ratepayers (2024)
across PJM consumers
The grid is the bottleneck. The private grid is the response. And the seam between them — who pays for the public infrastructure the private builders still lean on — is where the economics and politics of the AI buildout are now decided.Thorsten Meyer · The Queue · AI Energy & Infrastructure 02
Impacts of the Grid Bottleneck on AI Infrastructure
This shift from chip to grid constraints fundamentally alters the landscape of AI infrastructure development. It prioritizes geography based on proximity to power, inflates the cost of data-center leases by 15-25%, and fosters a bifurcated buildout: self-powered sites versus grid-dependent projects. The political debate centers on who should pay for the necessary grid upgrades, with ratepayers increasingly bearing the costs of bypass strategies. This dynamic could slow overall AI deployment, increase costs, and intensify policy conflicts over energy infrastructure funding.
private behind-the-meter gas power plant
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
From Chip Shortages to Grid Delays: The Changing Bottleneck
Historically, the AI buildout focused on securing advanced semiconductor chips, with supply constraints limiting expansion. Over the past two years, attention shifted as the chip supply chain stabilized and the interconnection queue emerged as the new bottleneck. The US has an abundance of power generation capacity, but the process to connect new projects to the grid is slow, bureaucratic, and physically constrained, creating a significant delay in deploying new data centers.
China’s rapid capacity additions—about 430 gigawatts annually—contrast sharply with US delays, which are measured in years. The US’s interconnection process involves lengthy permitting, infrastructure upgrades, and transformer supply chains that slow down project deployment despite available capital and demand. This has led to a strategic shift where private power generation is used to bypass the grid bottleneck.
Recent data indicates that nearly 80% of projects in the queue withdraw, highlighting the severity of the delay. Meanwhile, data-center energy demand is rising sharply, with some projects opting for co-location at nuclear plants or building behind-the-meter facilities, effectively sidestepping the grid but raising questions about cost distribution.
“The grid is the bottleneck; the response is a private grid; and the seam between them — who pays for the transmission and capacity — is where the politics of the AI buildout now lives.”
— Thorsten Meyer
nuclear backup power generator for data centers
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Unclear Long-Term Impact of Private Power Strategies
It remains unclear how widespread and sustainable private power solutions will be in the long term, especially regarding their economic viability, regulatory acceptance, and impact on the shared grid. The political debate over cost sharing and ratepayer impacts is still evolving, and future policy interventions could alter this dynamic.
grid interconnection delay mitigation solutions
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Next Steps in Grid Expansion and Policy Response
Expected developments include increased policy focus on fast-tracking grid upgrades, potential reforms to interconnection procedures, and greater scrutiny of private bypass strategies. Utilities and regulators may face pressure to balance the need for rapid AI infrastructure deployment with equitable cost sharing. Monitoring these policy shifts will be essential to understanding how the bottleneck evolves.
AI data center energy storage systems
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
Why has the interconnection queue become the main bottleneck for AI data centers?
The process to connect new power projects to the grid has become slow and bureaucratic, with delays of up to five or more years, despite abundant generation capacity and rising demand. This has shifted the primary constraint from chip supply to grid access.
How are companies bypassing the grid constraint?
Many are building private power sources, such as behind-the-meter gas plants or co-located nuclear facilities, to secure reliable power without waiting in the interconnection queue. These solutions often shift costs onto ratepayers and raise political questions.
What are the political implications of shifting costs to ratepayers?
Utilities and policymakers are increasingly concerned about who bears the financial burden of grid upgrades and capacity expansion. This has led to debates over cost allocation, with some projects passing billions in transmission costs to consumers, fueling political tensions.
Will the private grid solutions be sustainable long-term?
It is uncertain whether private, behind-the-meter power generation will be a sustainable solution, or if regulatory changes and policy interventions will force a more integrated, shared grid expansion in the future.
Source: ThorstenMeyerAI.com
