📊 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 bottleneck for AI infrastructure expansion has shifted from chip supply to the power grid’s interconnection queue. Capital is bypassing the grid, creating private power solutions that shift costs onto ratepayers. This development reshapes the landscape of AI infrastructure deployment.
The US power grid’s interconnection queue has become the dominant bottleneck for AI infrastructure development, surpassing chip supply constraints. This shift is prompting private power generation solutions that bypass the shared grid, with significant political and economic implications.
For the past two years, the narrative centered on chip shortages and GPU availability as the primary limiting factors for AI buildout. However, recent data indicates that the bottleneck has moved to the power grid, specifically the interconnection queue. Currently, between 2,300 and 2,600 gigawatts of generation and storage projects are stuck in US interconnection queues, a volume greater than the country’s total installed power capacity.
The median wait time for projects to reach commercial operation has increased from under two years in 2008 to nearly five years in 2026, with some data-center projects facing quoted timelines of up to twelve years. About 80% of projects in the queue withdraw before completion, highlighting the severity of the delay. Meanwhile, US data-center power demand is projected to reach approximately 76 gigawatts in 2026, up from 50 gigawatts in 2024, with global consumption potentially surpassing 1,000 terawatt-hours annually by the early 2030s.
As a result, capital is increasingly bypassing the grid. Some hyperscalers are co-locating with nuclear plants or building behind-the-meter gas plants to avoid long interconnection delays. For example, Microsoft’s deal to restart Three Mile Island Unit 1 aims to supply 835 megawatts of carbon-free baseload power, sidestepping the grid. This approach shifts costs onto ratepayers, as utilities and ratepayers bear the burden of expanding transmission capacity. The capacity auction in PJM, a major US grid operator, saw costs balloon from $2.2 billion to $14.7 billion in a single year, with $4.3 billion of transmission costs passed onto consumers in 2024.
The core argument is that the grid is now the binding constraint on AI infrastructure growth, leading to a bifurcation: those building private, self-powered solutions and those dependent on the slow-moving shared grid. This dynamic is reshaping the geography of data centers, the economics of power procurement, and the political landscape surrounding infrastructure costs.
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
Implications of the Grid Constraint for AI Infrastructure Expansion
This shift signifies a fundamental change in how AI infrastructure is deployed. The bottleneck moving from chip scarcity to grid interconnection delays means that capital is increasingly building private power sources to bypass the constraints, which may lead to a bifurcated infrastructure landscape. The costs associated with this bypass are ultimately borne by ratepayers, raising political and regulatory challenges. Furthermore, the re-pricing of geography and costs influences the location strategies of data centers and the economic viability of projects, potentially accelerating the privatization of power generation for AI needs.
CyberPower CP2000PFCRM2U PFC Sinewave UPS Battery Backup
- Power Capacity: 2000VA/1200W PFC Sine Wave UPS
- Number of Outlets: 8 NEMA 5-20R outlets with surge protection
- Input Power Cord: 6-foot NEMA 5-20P cord
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
From Chip Shortages to Grid Delays: The Evolving Bottleneck
Initially, the AI buildout was constrained by the availability of GPUs and chips, with supply chain issues dominating the narrative. Over the past two years, industry focus shifted to the power sector, where the interconnection queue has become the new choke point. The US faces a backlog of thousands of gigawatts in projects waiting to connect to the grid, with median wait times increasing fivefold since 2008. China, by comparison, adds hundreds of gigawatts annually, highlighting the US’s unique infrastructural bottleneck. This disconnect between available capital and the ability to connect to power is reshaping the development landscape.
Developers are responding by building private power sources, such as co-located nuclear or behind-the-meter gas plants, to avoid the delays. These solutions, while effective for individual projects, shift the costs onto the broader grid and ratepayers, fueling political debates over infrastructure financing and cost allocation.
“The grid is the bottleneck; the response is a private grid; and the seam between them — who pays for the transmission and capacity the private builders still lean on — is where the politics of the AI buildout now lives.”
— Thorsten Meyer
Unclear Long-Term Impact of Private Power Bypass
It remains uncertain how widespread and lasting the shift towards private power solutions will be, and whether regulatory changes might address the interconnection backlog. The political response to cost-shifting onto ratepayers is still evolving, and the long-term effects on grid reliability and fairness are not yet fully understood.
Next Steps in Addressing Grid Constraints and Political Debates
Policy discussions and regulatory reforms aimed at streamlining interconnection procedures are likely to intensify. Additionally, utilities and developers may pursue further private solutions, potentially leading to a bifurcated infrastructure landscape. Monitoring changes in interconnection timelines, costs, and political responses over the coming year will be critical to understanding how the US manages this bottleneck.
Key Questions
Why has the bottleneck shifted from chips to the power grid?
While chip shortages limited AI hardware availability, the power grid’s interconnection queue has become the new bottleneck, with delays of up to twelve years preventing new power capacity from coming online quickly enough to meet rising demand.
How are developers bypassing the grid constraint?
Developers are building private power sources, such as co-located nuclear or behind-the-meter gas plants, to avoid the long interconnection delays and ensure faster project deployment.
What are the political implications of shifting costs onto ratepayers?
The costs of expanding transmission and capacity are increasingly being passed onto consumers, leading to political debates and proposals for reform, such as the White House ‘Ratepayer Protection Pledge.’
Will regulatory reforms resolve the interconnection backlog?
It is uncertain; ongoing policy debates aim to streamline processes, but significant changes are still in development, and their effectiveness remains to be seen.
What does this mean for the future location of data centers?
The search for megawatts now prioritizes proximity to available power sources, including nuclear plants or private generation sites, over traditional factors like fiber latency.
Source: ThorstenMeyerAI.com
