America's AI ambitions depend on more than chips and algorithms. An ageing power grid and slow infrastructure development could threaten its global lead.
The global race for artificial intelligence is usually discussed in terms of advanced chips, powerful algorithms, computing capacity and skilled talent. But there is another factor that could determine who leads this race: electricity.
The United States is investing heavily in AI infrastructure, but its power grid is struggling to keep pace. The problem is not simply generating enough electricity. It is also about moving that electricity across long distances to the places where it is needed.
An ageing transmission network, fragmented decision-making, regulatory hurdles and lengthy legal disputes are making it increasingly difficult to build the infrastructure required to support America's AI ambitions.
A 102-Mile Transmission Line That Took 13 Years
The Cardinal-Hickory Creek project offers a striking example of the problem.
The 102-mile transmission line was designed to carry wind-generated electricity from Iowa to Wisconsin, crossing the Mississippi River and a protected wildlife and fish refuge along the way. Its purpose was to ease congestion on the Midwest electricity grid and improve the movement of renewable power.
The regional grid operator approved the need for the project in 2011. However, construction was completed only in 2024.
The intervening years were marked by route studies, objections from landowners, environmental reviews, court proceedings and disputes over how the transmission line should cross the protected refuge. A judge initially blocked a proposed land swap, an appeals court later reversed that decision, and the project proceeded through several stages before completion.
Taking 13 years to build a transmission line of just over 100 miles is more than an administrative inconvenience. It reflects the difficulty the United States faces in delivering large infrastructure projects.
Electricity Demand Is Rising as Construction Slows
The timing is particularly challenging.
After nearly two decades of relatively flat electricity consumption, demand in the United States is rising again. Data centres, artificial intelligence, electric vehicles and industrial expansion are contributing to the increase.
Major technology companies are expected to spend close to $900 billion on AI-related projects this year. One consultancy estimates that the country could need an additional 166 gigawatts of electricity capacity by 2030, roughly equivalent to the consumption of 15 New York Cities.
Yet the expansion of the transmission network is moving in the opposite direction.
In 2013, the United States added nearly 4,000 miles of high-voltage transmission lines. By 2024, annual additions had fallen to fewer than 900 miles.
Large transmission projects now take approximately a decade from planning to completion, a timeline comparable to the construction period of the Hoover Dam.
Generating electricity is only half the challenge. It must also reach consumers and industries reliably. Much of America's existing transmission infrastructure was built during the 1960s and 1970s and is approaching the end of its useful life.
Expanding and modernising the network would allow cheaper electricity to reach areas facing shortages, reduce the risk of blackouts and help contain household and industrial power costs.
A Fragmented System With No Clear Authority
Why is building transmission infrastructure so difficult in the United States?
One major reason is the fragmented nature of its electricity system.
Grid planning has evolved through a combination of private power companies, local utilities, state governments, regional operators and federal agencies. Each institution has its own responsibilities, priorities and regulatory processes.
This decentralised structure has encouraged competition, local solutions and experimentation. However, it becomes a serious obstacle when a project involves multiple states and regions.
A long-distance transmission line may require approvals from several regulatory authorities. These authorities may disagree over whether the project is necessary, how its costs should be distributed and which communities should bear its impact.
Landowners can challenge proposed routes. Local utilities may resist projects that allow cheaper electricity from other regions to enter their markets. Environmental groups can raise concerns, while legal challenges can keep construction tied up in courts for years.
The underlying problem is the absence of a coordinated national approach to transmission planning.
Electricity does not follow political or administrative boundaries. Yet the institutions responsible for regulating the grid often operate within precisely those boundaries. This creates a system in which numerous stakeholders can influence a project, but no single authority has the power to ensure its timely completion.
The result is an infrastructure process where delaying a project can be easier than delivering it.
China Is Building the Infrastructure America Needs
The grid challenge also has a geopolitical dimension.
Over the past 15 years, China has expanded its high-voltage transmission network on a much larger scale than the United States. Its centralised planning system allows the government to coordinate routes, investment and construction across regions with fewer institutional obstacles.
Washington operates under a very different political and regulatory structure. State governments, private utilities, federal agencies and local communities all have a say in major infrastructure decisions.
The difference matters because the AI competition is not being fought through software alone. It also depends on the physical infrastructure required to operate increasingly powerful computing systems.
Industry leaders have warned that the United States risks losing a competitive advantage if it cannot build the electricity infrastructure needed to support its technology sector.
The concern is straightforward: a country cannot maintain leadership in AI if its data centres cannot secure reliable and affordable power.
Who Will Pay for the AI Power Demand?
The electricity challenge is also becoming a public issue.
A poll found that roughly half of Americans believe data centres are bad for household electricity costs. The concern is understandable. Technology companies are investing billions in facilities that require enormous amounts of electricity, while ordinary consumers may ultimately bear part of the infrastructure cost.
An estimated $1.3 trillion could be invested in power generation and transmission infrastructure by 2030.
The central question is how that investment will be financed and who will benefit from it.
If the cost of expanding the grid is passed on to household consumers through higher utility bills, public resistance to new data centres and transmission projects could intensify.
Communities may reasonably question why they should pay more for electricity to support private technology companies.
This makes cost-sharing and transparency essential parts of the debate. The benefits of expanding the grid may be national, but the financial burden should not fall unfairly on ordinary electricity consumers.
Temporary Solutions Cannot Replace Grid Expansion
With transmission infrastructure failing to keep pace with demand, technology companies and data-centre developers are exploring alternatives.
More than a quarter of planned data-centre capacity is reportedly being designed with dedicated on-site power generation. Some developers are also searching for locations where existing electricity infrastructure has spare capacity.
Another approach is to make data centres more flexible electricity consumers.
Certain computing tasks can be postponed during peak demand periods, shifted to facilities in other locations or temporarily supported by backup power. Google has already entered into agreements in several states to make its electricity consumption more flexible.
Such arrangements could help utilities manage demand without immediately building new generation and transmission capacity.
However, these measures have limitations.
On-site power generation is expensive and still requires reliable backup arrangements. Shifting computing workloads cannot solve every electricity shortage, particularly when data centres need continuous power for critical operations.
Moreover, the availability of spare grid capacity is shrinking. Connecting a new facility to the electricity network can itself take years, with some projects facing timelines of around a decade.
These workarounds may buy time, but they cannot replace a modern and adequately expanded transmission network.
What Needs to Change?
The United States needs a more coordinated approach to electricity infrastructure.
First, long-distance transmission planning must move beyond fragmented state-level decision-making. Regional and national requirements need to be considered together, with clearer responsibilities for the institutions involved.
Second, permitting and approval procedures need to become faster and more predictable. Projects should not remain trapped in overlapping reviews and repeated legal proceedings for years.
Third, landowners and local communities must receive fair treatment. Their concerns about property rights, environmental damage and local disruption cannot simply be dismissed in the name of economic growth.
Finally, the country needs a more transparent system for distributing infrastructure costs. Technology companies that generate substantial new electricity demand should bear a fair share of the investment required to serve them.
The challenge is to achieve these reforms without weakening environmental safeguards, public consultation or the rights of affected communities. Faster infrastructure development should not mean removing accountability.
The Real Test of America's AI Ambitions
America's artificial intelligence ambitions are usually measured through chip production, computing power, research investment and technological breakthroughs.
But another test is becoming increasingly important: can the country build the physical infrastructure needed to support its ambitions?
The United States may have some of the world's leading technology companies, researchers and AI developers. Yet none of these advantages can compensate for an electricity network that cannot deliver enough reliable power to where it is needed.
China's experience demonstrates the strategic importance of large-scale infrastructure development. America's challenge is to find a way to combine its decentralised institutions and democratic safeguards with the ability to complete major projects on time.
The next phase of the AI race may not be decided by algorithms alone. It may also depend on who can build the power lines, modernise the grid and deliver electricity before demand overtakes supply.
The global race for artificial intelligence is usually discussed in terms of advanced chips, powerful algorithms, computing capacity and skilled talent. But there is another factor that could determine who leads this race: electricity.
The United States is investing heavily in AI infrastructure, but its power grid is struggling to keep pace. The problem is not simply generating enough electricity. It is also about moving that electricity across long distances to the places where it is needed.
An ageing transmission network, fragmented decision-making, regulatory hurdles and lengthy legal disputes are making it increasingly difficult to build the infrastructure required to support America's AI ambitions.
A 102-Mile Transmission Line That Took 13 Years
The Cardinal-Hickory Creek project offers a striking example of the problem.
The 102-mile transmission line was designed to carry wind-generated electricity from Iowa to Wisconsin, crossing the Mississippi River and a protected wildlife and fish refuge along the way. Its purpose was to ease congestion on the Midwest electricity grid and improve the movement of renewable power.
The regional grid operator approved the need for the project in 2011. However, construction was completed only in 2024.
The intervening years were marked by route studies, objections from landowners, environmental reviews, court proceedings and disputes over how the transmission line should cross the protected refuge. A judge initially blocked a proposed land swap, an appeals court later reversed that decision, and the project proceeded through several stages before completion.
Taking 13 years to build a transmission line of just over 100 miles is more than an administrative inconvenience. It reflects the difficulty the United States faces in delivering large infrastructure projects.
Electricity Demand Is Rising as Construction Slows
The timing is particularly challenging.
After nearly two decades of relatively flat electricity consumption, demand in the United States is rising again. Data centres, artificial intelligence, electric vehicles and industrial expansion are contributing to the increase.
Major technology companies are expected to spend close to $900 billion on AI-related projects this year. One consultancy estimates that the country could need an additional 166 gigawatts of electricity capacity by 2030, roughly equivalent to the consumption of 15 New York Cities.
Yet the expansion of the transmission network is moving in the opposite direction.
In 2013, the United States added nearly 4,000 miles of high-voltage transmission lines. By 2024, annual additions had fallen to fewer than 900 miles.
Large transmission projects now take approximately a decade from planning to completion, a timeline comparable to the construction period of the Hoover Dam.
Generating electricity is only half the challenge. It must also reach consumers and industries reliably. Much of America's existing transmission infrastructure was built during the 1960s and 1970s and is approaching the end of its useful life.
Expanding and modernising the network would allow cheaper electricity to reach areas facing shortages, reduce the risk of blackouts and help contain household and industrial power costs.
A Fragmented System With No Clear Authority
Why is building transmission infrastructure so difficult in the United States?
One major reason is the fragmented nature of its electricity system.
Grid planning has evolved through a combination of private power companies, local utilities, state governments, regional operators and federal agencies. Each institution has its own responsibilities, priorities and regulatory processes.
This decentralised structure has encouraged competition, local solutions and experimentation. However, it becomes a serious obstacle when a project involves multiple states and regions.
A long-distance transmission line may require approvals from several regulatory authorities. These authorities may disagree over whether the project is necessary, how its costs should be distributed and which communities should bear its impact.
Landowners can challenge proposed routes. Local utilities may resist projects that allow cheaper electricity from other regions to enter their markets. Environmental groups can raise concerns, while legal challenges can keep construction tied up in courts for years.
The underlying problem is the absence of a coordinated national approach to transmission planning.
Electricity does not follow political or administrative boundaries. Yet the institutions responsible for regulating the grid often operate within precisely those boundaries. This creates a system in which numerous stakeholders can influence a project, but no single authority has the power to ensure its timely completion.
The result is an infrastructure process where delaying a project can be easier than delivering it.
China Is Building the Infrastructure America Needs
The grid challenge also has a geopolitical dimension.
Over the past 15 years, China has expanded its high-voltage transmission network on a much larger scale than the United States. Its centralised planning system allows the government to coordinate routes, investment and construction across regions with fewer institutional obstacles.
Washington operates under a very different political and regulatory structure. State governments, private utilities, federal agencies and local communities all have a say in major infrastructure decisions.
The difference matters because the AI competition is not being fought through software alone. It also depends on the physical infrastructure required to operate increasingly powerful computing systems.
Industry leaders have warned that the United States risks losing a competitive advantage if it cannot build the electricity infrastructure needed to support its technology sector.
The concern is straightforward: a country cannot maintain leadership in AI if its data centres cannot secure reliable and affordable power.
Who Will Pay for the AI Power Demand?
The electricity challenge is also becoming a public issue.
A poll found that roughly half of Americans believe data centres are bad for household electricity costs. The concern is understandable. Technology companies are investing billions in facilities that require enormous amounts of electricity, while ordinary consumers may ultimately bear part of the infrastructure cost.
An estimated $1.3 trillion could be invested in power generation and transmission infrastructure by 2030.
The central question is how that investment will be financed and who will benefit from it.
If the cost of expanding the grid is passed on to household consumers through higher utility bills, public resistance to new data centres and transmission projects could intensify.
Communities may reasonably question why they should pay more for electricity to support private technology companies.
This makes cost-sharing and transparency essential parts of the debate. The benefits of expanding the grid may be national, but the financial burden should not fall unfairly on ordinary electricity consumers.
Temporary Solutions Cannot Replace Grid Expansion
With transmission infrastructure failing to keep pace with demand, technology companies and data-centre developers are exploring alternatives.
More than a quarter of planned data-centre capacity is reportedly being designed with dedicated on-site power generation. Some developers are also searching for locations where existing electricity infrastructure has spare capacity.
Another approach is to make data centres more flexible electricity consumers.
Certain computing tasks can be postponed during peak demand periods, shifted to facilities in other locations or temporarily supported by backup power. Google has already entered into agreements in several states to make its electricity consumption more flexible.
Such arrangements could help utilities manage demand without immediately building new generation and transmission capacity.
However, these measures have limitations.
On-site power generation is expensive and still requires reliable backup arrangements. Shifting computing workloads cannot solve every electricity shortage, particularly when data centres need continuous power for critical operations.
Moreover, the availability of spare grid capacity is shrinking. Connecting a new facility to the electricity network can itself take years, with some projects facing timelines of around a decade.
These workarounds may buy time, but they cannot replace a modern and adequately expanded transmission network.
What Needs to Change?
The United States needs a more coordinated approach to electricity infrastructure.
First, long-distance transmission planning must move beyond fragmented state-level decision-making. Regional and national requirements need to be considered together, with clearer responsibilities for the institutions involved.
Second, permitting and approval procedures need to become faster and more predictable. Projects should not remain trapped in overlapping reviews and repeated legal proceedings for years.
Third, landowners and local communities must receive fair treatment. Their concerns about property rights, environmental damage and local disruption cannot simply be dismissed in the name of economic growth.
Finally, the country needs a more transparent system for distributing infrastructure costs. Technology companies that generate substantial new electricity demand should bear a fair share of the investment required to serve them.
The challenge is to achieve these reforms without weakening environmental safeguards, public consultation or the rights of affected communities. Faster infrastructure development should not mean removing accountability.
The Real Test of America's AI Ambitions
America's artificial intelligence ambitions are usually measured through chip production, computing power, research investment and technological breakthroughs.
But another test is becoming increasingly important: can the country build the physical infrastructure needed to support its ambitions?
The United States may have some of the world's leading technology companies, researchers and AI developers. Yet none of these advantages can compensate for an electricity network that cannot deliver enough reliable power to where it is needed.
China's experience demonstrates the strategic importance of large-scale infrastructure development. America's challenge is to find a way to combine its decentralised institutions and democratic safeguards with the ability to complete major projects on time.
The next phase of the AI race may not be decided by algorithms alone. It may also depend on who can build the power lines, modernise the grid and deliver electricity before demand overtakes supply.
Leave a Comment