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.

The US interconnection queue has become the dominant bottleneck for AI infrastructure growth, surpassing chip supply constraints. With roughly 2,300 to 2,600 gigawatts of projects waiting for grid connection—more than the entire US power capacity—the median wait time has grown to nearly five years. This shift is prompting developers and large-scale data-center operators to seek private, behind-the-meter power solutions, bypassing the grid entirely.

For two years, the industry focused on securing GPUs and fabrication capacity to meet AI demand. That narrative has shifted; now, the primary constraint is the slow and congested US electrical grid interconnection process. The queue for connecting new power generation and storage projects has ballooned to over 2,300 gigawatts, with median wait times approaching five years, up from under two years in 2008. Some projects, particularly data centers, face quoted timelines of up to twelve years.

As a result, capital is increasingly routing around the grid. Large data-center operators are co-locating power generation at nuclear plants, such as Microsoft’s deal to restart Three Mile Island Unit 1, or building private gas plants that can be operational in 18 months. Meanwhile, utilities like PJM report that more gigawatts of data-center applications are in the queue than their historic peak demand, illustrating the scale of demand and the urgency for alternative solutions.

This bypassing comes at a cost: when private power generation is built, the transmission and capacity costs are shifted onto ratepayers, fueling political debates. For example, PJM’s transmission costs passed to consumers reached $4.3 billion in 2024, with Virginia bearing nearly $2 billion. The industry’s response to the grid constraint is effectively creating a bifurcated buildout: self-powered, private solutions for capital-rich players, and a slow, congested public grid for others.

Implications of the Grid Constraint on AI Infrastructure

This shift fundamentally alters how and where AI infrastructure is built. The grid interconnection queue’s delay reprices geography, favoring locations with immediate power access or private generation. It also revalues project economics, with queue position becoming a key cost factor, leading to premium leasing rates for sites with faster power access. Politically, the externalization of grid costs onto ratepayers raises questions about fairness and policy responses, shaping the future of US energy infrastructure and AI development.

From Chip Shortage to Grid Bottleneck

Historically, the focus of AI infrastructure expansion was on securing semiconductor chips—GPU supply and fabrication capacity. Over the past two years, the narrative shifted as chip shortages eased and supply chains stabilized. The new bottleneck emerged from the physical and bureaucratic constraints of the US electrical grid, where the interconnection process has become a choke point. The US has over 2,300 gigawatts of projects waiting in the queue, with median connection times rising sharply, contrasting with China’s rapid capacity additions of around 430 gigawatts annually.

This bottleneck is not due to a lack of capital or generation capacity but stems from the slow pace of grid upgrades, permitting, and transformer supply. As a result, developers and data-center operators are increasingly building private, behind-the-meter generation or co-locating at existing nuclear plants, bypassing the grid to meet their rapid deployment needs.

“The constraint has shifted from silicon to the grid — the interconnection queue is now the primary bottleneck for AI infrastructure buildout.”

— Thorsten Meyer

Unclear Impact of Private Grid Solutions on Policy

It remains unclear how policy will evolve to address the externalization of grid costs and the political backlash against ratepayer burdens. The long-term consequences of widespread private generation bypassing the public grid are still being debated, and regulatory responses are uncertain.

Future Developments in Grid and Infrastructure Policy

Next steps include potential policy interventions to manage costs and ensure equitable grid access. Industry efforts to streamline interconnection processes and expand capacity are also expected to accelerate. Monitoring how the political landscape responds to the cost shifts and private grid proliferation will be critical in shaping the future of AI infrastructure buildout.

Key Questions

Why is the interconnection queue now the main bottleneck for AI infrastructure?

The queue’s median wait time has grown to nearly five years, delaying project deployment despite abundant capital and generation capacity, due to bureaucratic, physical, and permitting delays in the grid infrastructure.

How are companies bypassing the grid constraint?

Many are building private power generation—such as co-locating nuclear or gas plants—and relying on behind-the-meter solutions to avoid the long interconnection delays.

What are the political implications of these private solutions?

Private generation shifts costs onto ratepayers, fueling political debates over fairness, cost allocation, and the future regulation of grid infrastructure.

Will policy changes address the interconnection backlog?

It is uncertain; policymakers are considering reforms, but the long-term impact remains to be seen as industry and political interests evolve.

Source: ThorstenMeyerAI.com