Answering the call to build
Let people build to meet load growth
A new report from Grid Strategies on load growth updates their 2023 report. To summarize its findings in a word, more.
For topline predictions, Grid Strategies suggests:
Load growth of 128 GW over the next five years.
This represents about 16% electricity demand growth from 2024 to 2029.
“Strategic industries” are behind most of that growth:
Data centers with around 90 GW
Advanced manufacturing at around 20 GW, and
Electrification of transportation and buildings adds approximately 20 GW.
With the big picture in mind, consider this: Electricity is the only industry that panics when you say they will have more business next year. Other industries would be glad to know they have a certain future!
As has been the case for most of the public conversation around load growth, reporting on rising demand ignores that there are plenty of would-be generators. The call is coming from inside the house. We have much more that wants to enter the system than even these new, larger load growth projections suggest.
There are 2.6 terawatts of generation waiting to be approved in the interconnection queue. Historically, only about 14% of what’s in the queue makes it to the generating stage. That still puts us at more than enough–360 gigawatts! It’s a bit less than three times the load growth that Grid Strategies expects. Another recent analysis found more than 1,200 GW of sites that could power AI data centers off-grid with a combination of solar, batteries, and gas backup. So, the US is not lacking opportunity.
The real problem is “an inability to build.” As Alex Trembath and I put it in Heatmap, “Obstacles to technology licensing, permitting, interconnection, and transmission are the key bottlenecks,” preventing energy suppliers from rising up to meet energy demand.
Of course, most of the projects waiting are wind and solar. The data center industry is clearly in the market for dispatchable generation to backup intermittent energy supplies. They’re investing in gas generation for today, and making a string of nuclear investments. If you zoom in on the Three Mile Island restart, as an example, remember that Governor Shapiro in Pennsylvania had to ask the grid operators to speed up the plant’s reconnection to the grid!
Let people build
Casting the central problem as our inability to build also reveals the path forward. The future will be possible only by cutting red tape holding back energy entrepreneurs.
The right policy solutions fall under one big tent: remove the barriers to building. In practice, this big tent holds several approaches at every level:
Streamline permitting for all energy technologies–generation, transmission, and residential technologies like rooftop solar or home batteries.
Streamline interconnection processes (moving towards ERIS is the concrete advice).
Investigate avenues for flexibility in demand and new solutions like co-location that can help make data centers and large loads grid assets (interconnection processes were built when load was always thought of as firm, these solutions require new pathways and policies to enable).
NEPA reform should replace vetocracy with substantive rules.
Enable energy parks with clear, certain rules (either off-grid, private, or tied into the grid).
Streamline the evolution of retiring plants into new generation, such as coal to gas, nuclear, renewables, or some combination of any of these.
Create new funding mechanisms, like Google’s Clean Transition Tariff, to allow companies to pay for what they want and need.
Improving federal permitting where states play a role, such as in the Clean Air Act.
Pre-approve sites for nuclear development and energy development (Utah has been working on reports about strategic pathways for nuclear development that will be useful guide for other states).
These kinds of steps are good policy changes that improve the energy system for everyone. They’re important because, as the North American Electric Reliability Corporation’s (NERC) John Moura recently put it, “Simply put, our infrastructure is not being built fast enough to keep up with the rising demand.” NERC’s recent 10-year assessment suggests shortfalls become more likely without changes. Lynne Kiesling’s series on data centers identifies the same troubles.
Look back when setting a path forward
With all of these projections of the future, look back at history too. In 1999, a leading energy commentator said that 50% of electricity would be from computers “within the next decade.” Today, data centers use around 4% of US electricity.
Researchers regularly rein in more sensationalized reporting from consultancies and public reporting. The 2024 Berkeley Lab report on data center usage, for example, forecasts a smaller range of potential growth than many others. They predict that future data center electricity consumption will likely sit between about 7% and 12% of US electricity consumption by 2028.
The lesson is simple: Extrapolating today’s trends can lead to more foul balls than useful ballpark figures. The crystal ball is always foggier than history.
Setting that historical view aside again–the human condition requires guessing at the future! Far from fearing load growth, cross your fingers that even the high projections are too low. That would signal AI’s massive value to lifting the world to new heights.
Policy should set an open road so that every entrepreneur can make a run at their idea. Letting entrepreneurs try out their ideas allows, as the economist Arnold Kling puts it, the process of experimentation, evaluation, and evolution that drives progress to begin. In the case of the load growth predicted by Grid Strategies, letting people build is the most promising way to meet the challenge.
Good thoughts, Josh!