Clean energy projects fail, stall, and stumble all the time. As 2030 grows closer — and associated clean energy targets grow more urgent — a gap is emerging between announced projects and projects that actually get funded and deployed.
Around 80% of projects never make it to construction, due to barriers like local opposition and grid interconnection issues. That’s a “very, very high failure rate,” said James McWalter, CEO and co-founder of Paces, which is an artificial intelligence-fueled geographic information system and data platform that aims to help renewable developers pick the best sites for their projects.
The startup uses AI to sort through the mountains of data on a developer’s key considerations when selecting a location: its size, its access to the grid, and its likelihood of getting the local permits it needs. And ultimately, the goal is to save time.
Those three elements together result in a yearslong siting process. In New York state, for example, siting large-scale renewable projects averages more than three years; elsewhere, it can stretch to up to eight.
But Paces claims it can cut down the site finding and evaluation process “from months to minutes,” saving developers the time, money, and energy they might spend on a project that maybe never had a shot in the first place.
“What we do is pull data across these three buckets of information from lots and lots of different sources,” McWalter told Latitude Media. “We make sense of it, and then we say, ‘Okay, this is a great place to build solar, and it’s not just because the sun is shining, but because you’re much more likely to get the project built here.’”
The startup recently announced it has raised an $11 million Series A led by Navitas Capital, roughly two years after it raised $1.9 million in pre-seed funding. It has a subscription-based revenue model, and around 100 customers, including EDF Renewables, AES, and Third Pillar Solar.
Low-hanging fruit
McWalter and Charles Bai, the two co-founders, both have backgrounds in data analytics and AI. McWalter started his career working for FactSet, a financial data and analytics platform, and then worked with data and machine learning for various startups across the U.S. and Mexico; Bai worked as an applied machine learning software engineer at Meta.
Before co-founding Paces in 2022, though, McWalter was considering agriculture-based decarbonization solutions, and started calling farmers across the Midwest to get their thoughts. What he heard instead, though, was a lot of complaining about all the clean energy developers reaching out to them to inquire about their land.
“One of the farmers I was cold-calling in Indiana was like: ‘Oh, are you one of these bloody wind developers pestering me?’” he said.
Those complaints signaled to McWalter a need on the developers’ side that AI’s data processing capabilities could be well-suited to meet — a realization that caused him to pivot to an entirely different corner of cleantech. According to Matt Casey, managing director of Latitude Intelligence, siting and permitting processes are “low-hanging fruit” for immediate and impactful AI utilization.
Paces is not the first to go after this space. Software companies with what McWalter characterizes as a “very map-based approach” to finding siting solutions have been around for years. But according to McWalter, it was only recently that the clean energy sector developed enough to create a market for the level of specialized detail that Paces offers — and helpfully AI advanced dramatically at right about the same time.
“It was not a big enough industry for venture-backed companies to look at until maybe two or three years ago, and [it’s been] incredibly fueled by the Inflation Reduction Act,” McWalter said, adding that he’s expecting competition to increase. “Like any solid, good service software company, we’re very paranoid about moving ever faster than anyone else.”
Permitting predictor
A key part of Paces’ competitive edge is the number of layers it adds to its maps to evaluate how likely a parcel of land is to see a successful project.
Geographic layers with information about floodplains and forestry are straightforward and have been used for years. Information about local permitting, though, is harder to crack, given that there are over 20,000 permitting jurisdictions in the United States. Paces’ system collects data from thousands of jurisdictions and uses large language models to go through towns’ and municipalities’ meeting minutes and collect information about zoning, moratoriums, permitting requirements, and environmental restrictions, among other things.
“This ability to take all of [this] textual data and make sense of it, and then connect it to the geography, is one of the key pieces that we built,” McWalter said.
Given how hard it can be to access permitting information, how costly grid upgrades can be, and how notoriously long grid interconnection times have become, developers tend to pick a project spot based on local grid capacity, according to McWalter. The status quo is that they start the interconnection process and see if it moves forward before starting the local permitting process, which means that “developers flood in” where there’s grid capacity.
“They all start to connect at the same time, and when they start doing the permitting, their activity overwhelms the local permitting office, which is then very likely to pass an anti-solar ordinance or some moratorium,” McWalter said. “So we see that the places with [the] most capacity on the grid… have the worst ‘permittability.’ It’s a hidden time bomb sitting in the interconnection queue.”
Community opposition ordinances limiting renewable development are considered the leading causes of project cancellations on par with grid interconnection issues, so much so that Paces runs a “sentiment check” on project sites. However, the company doesn’t find it a very useful metric, because the sentiment is negative in most jurisdictions.
When it comes to clean energy projects, McWalter added, “we need to do a better job communicating with those communities about the upside.”
