Can AI help connect more solar and wind farms to the power grid?

The US Department of Energy (DOE) believes AI can speed up the process of connecting new energy projects to the power grid.

It announced $30 million in funding now available through its Artificial Intelligence for Interconnection (AI4IX) program. The DOE is interested in fostering partnerships between grid operators and software and energy project developers to bring new sources of power online faster.

The plan is to use AI to streamline the interconnection application process that is currently very slow. Delays are holding back efforts to get new solar and wind farms operational. The pressure to clear that backlog is mounting as power demand grows — ironically, a large part of it is due to the artificial intelligence boom.

As it is, it takes up to seven years to connect a new power generation project to the grid in the US, and the wait time is increasing. As a result, there's a huge backlog of 2,600 gigawatts of new energy generation and storage projects awaiting interconnection. That's nearly twice the capacity currently available for power generation in the U.S., according to the DOE.

Why does it take so long to connect to the grid? Utilities and grid operators require project developers to complete a series of studies aimed at determining what kinds of grid upgrades might be necessary and how much they'll cost.

The slow interconnection process is also a relic of an energy system designed around fossil fuels — when developers had relatively few large power plants in the queue. Power grids with more renewable energy are less centralized, with power coming from a more distributed network of smaller solar, wind, and battery projects. That means more applications to review.

New solar and onshore wind farms are generally cheaper sources of electricity than coal or gas, and they don't produce air pollution or contribute to climate change. You can reap those benefits by investing in new power plants in the U.S. This could be reflected in the type of energy infrastructure being built in the U.S. More than 94 percent of new energy capacity waiting to be connected to the grid is carbon-free, primarily solar, wind and batteries.

The DOE suggests that if utilities and grid operators could use AI, they could complete interconnection applications for those projects faster. It says the current manual process of reviewing applications is labor-intensive — a problem made worse because submitted applications are often incomplete. Communicating with developers so they can fill in missing information to correct their applications can cause significant delays, the DOE says. Through AI4IX, the DOE wants to use existing AI algorithms to quickly identify applications with deficiencies and notify their authors. For example, utilities could use AI software trained on accurate applications. The DOE is accepting proposals for the first round of AI4IX funding until January 10, 2025, and hopes to announce award winners sometime in the winter of 2025. Federal agencies are bracing for upheaval once President-elect Donald Trump takes office next year. Elon Musk and Vivek Ramaswami's plans for Trump's "Department of Government Efficiency" include "massive staff cuts" across federal agencies. And Project 2025 calls for eliminating programs within the DOE's Grid Deployment Office, which is overseeing funding for the AI4IX program.

But despite Joe Biden's promises to abandon the clean energy legacy, Trump's team has not specifically targeted funding from the bipartisan infrastructure legislation that funds the initiative.

In addition, Trump's pick to lead the Environmental Protection Agency, Lee Zeldin, says he wants to "make the US the AI capital of the world." This pledge reflects the incoming Trump administration's regulation agenda, particularly when it comes to any rules that could hinder the growth of energy-hungry AI data centers.

According to Goldman Sachs Research, power demand from data centers could increase 160 percent by 2030 with the boom in AI. So bringing more power generation online to meet that demand may remain a priority for the DOE going forward.