These Companies Want To Use AI To Make Cheaper and Cleaner Energy—If the Government Lets Them

Donald Trump promised on the campaign trail to halve electricity prices within 18 months of taking office. With more than half of that period now elapsed, it seems that the president will fall well short of his goal.

In October, the most recent month for which data are available, the annual price of residential electricity jumped by 5.2 percent, according to the Energy Information Administration. While there are myriad reasons for these price hikes, including Trump’s own trade policies, AI and data centers have taken much of the blame from policymakers. Sen. Bernie Sanders (I–Vt.) recently suggested enacting a federal moratorium on data center construction, and Sen. Elizabeth Warren (D–Mass.) has opened an investigation into the role that data centers play in driving up electricity costs.

The public, too, seems to be souring on AI: An October poll from Arbor, an electricity software company, found that nearly two-thirds of respondents think AI is causing higher utility bills. Meanwhile, over 70 percent of Americans say they are concerned about the environmental impact of AI, according to an Associated Press–NORC poll conducted in September.

It might be convenient to blame AI for these challenges, but government intervention in the space could hinder the solutions that several companies are developing to make the grid cleaner and more affordable.

One such firm is Everstar, which is looking to address one of the biggest roadblocks to nuclear power: excessive paperwork.

While reducing paperwork may seem like a trivial fix, it’s an important one; a reactor license application can easily exceed 10,000 pages and undergo up to two years of review from federal regulators. And simple errors in these documents can set projects back and cost thousands of dollars. For one of Everstar’s clients, fixing an error in the licensing documentation, which CEO Kevin Kong tells Reason was “essentially a typo,” required “developing and getting approval for a formal License Amendment Request.” This request cost the developer “tens of thousands of dollars in engineering time and external consultants” and added months in regulatory review, according to Kong.

Gordian, the company’s AI-enabled platform, aims to eliminate cases like these by “automat[ing] compliance, technical documentation, and regulatory navigation for the nuclear industry,” says Kong. Since launching earlier this year, the technology has yielded impressive results. After Last Energy was given federal funding in August to demonstrate its advanced nuclear reactor, it partnered with Everstar to write a 50-page environmental assessment. What would normally take eight weeks was completed in one. The system was also able to turn around a 200-page ecology report—a revision that usually takes a few weeks—in one night.

Kong says his clients have been able to cut “30-40% of the time spent on major regulatory deliverables,” which can be the “difference between projects penciling out or not.” The company plans to scale up operations in the coming year.

But getting the federal government to green-light new sources of energy generation is only one fix for the electricity price problem; developers must also be able to get this power onto the grid. This process can take up to five years to complete, keeping large swaths of power off the grid.

Tapestry, a Google X moonshot, is working on what it hopes will be a fix to this issue with its Grid Planning Tool. Before a grid operator approves new interconnection requests, it often has to run simulations of the request, sometimes decades into the future. The Grid Planning Tool can take this process down from weeks to days and enables operators to identify the “most affordable, reliable, low carbon [energy sources],” according to the company. Tapestry’s tech is being deployed by PJM Interconnection, America’s largest grid operator, to speed up interconnection times.

Meanwhile, the firm’s GridAware platform automates inspections of physical grid infrastructure, like poles and wires, to improve reliability, prevent outages, and reduce maintenance costs for utility companies (which often get passed on to ratepayers). GridAware has been deployed in New Zealand, where it cut down inspection times for the nation’s biggest utility “from 45 minutes to around five minutes per asset,” according to Latitude Media.

AI may be the boogeyman for high electricity prices these days, but the technology is unlocking solutions to deliver cleaner, cheaper, and more abundant energy. That is, unless the government gets in the way.

The post These Companies Want To Use AI To Make Cheaper and Cleaner Energy—If the Government Lets Them appeared first on Reason.com.