Path to ZeroEnergyClimate ChangeElectric VehiclesSupply Chains

To solve the water crisis, companies are increasingly turning to AI 

July 11, 2022, 5:00 PM UTC

Utah’s Great Salt Lake is drying up: Withering under the Southwestern U.S.’s worst drought in 1,200 years, the lake is now only a third its usual size and threatening to fill the air in nearby Salt Lake City with alarming levels of arsenic.

The risk, which a Republican state lawmaker in Utah recently called an “environmental nuclear bomb,” is just the kind of danger wrought by climate change that’s likely to discourage capital from flowing into the Southwest, costing private companies millions of dollars, said Matt Ellis, the CEO of Measurabl, which uses machine learning and vast amounts of data to help companies monitor their buildings’ climate emissions as well as energy and water use.

“You’re going to have a growth problem, a home value problem and a business problem,” he said, noting investors are likely to worry about building in Salt Lake City. “This goes really fast all the way through the capital stack of the real estate business, and we’re seeing our customers literally start to change their buying and selling behavior because of it.”

Southwest’s water shortage and other climate-related events, which are increasingly frequent and severe, are accelerating demand for more accurate measures of climate change’s impact, said industry observers and executives at the companies providing these technologies. Companies across all sectors of the economy are increasingly turning to AI and data models like those offered by Measurabl to better assess how climate will threaten their operations and profits.

The impact of the drought reaches far beyond the real estate business: It has dried the Southwest’s major reservoirs to record lows, threatening crops, supply chains and manufacturing lines, experts said. The consumer staples sector, which includes food, beverage and non-durable household goods, could lose $200 billion as a result of water shortages alone, according to a note issued in June of last year by U.K. bank Barclays. Meanwhile, big tech companies, whose data centers are cooled by millions of gallons of fresh water each day, have had to promise deep water conservation measures as Southwestern cities pushed back against plans to build new cloud infrastructure.

Investors, insurance providers and creditors are relying more and more on climate data to make key strategic decisions, which is putting pressure on companies to identify, disclose and address climate-related risks. 

Market research firm Gartner said sustainability efforts, known among corporations and investors as Environmental, Social and Governance (ESG) initiatives, rank among corporate board of directors’ top three priorities for this year and 2023, according to a survey of boards conducted late last year. It didn’t have hard numbers illustrating growing private-sector adoption of AI and data technologies. 

”There is a growing ecosystem of startups that are using AI and cloud services” for climate-related risk assessments,“ the market research firm said in a report published in December. “Use of machine learning and other AI techniques for climate change mitigation is already well underway.“

Microsoft is using data and machine learning models to both stanch the millions of gallons of water cooling its data centers and to help others identify and address water usage problems.

London-based Cervest said it uses machine learning and a vast array of data collected by satellites, weather stations and remote sensing methods to assess risks to a company’s portfolio of assets, including factories, corporate campuses, distribution and data centers.

Lotic Labs deploys similar technologies to monitor weather-related risk and translate that data into financial risk models.

The agriculture business, one of the world’s biggest consumers of freshwater, is also turning to AI to operate more efficiently, from field to grocery produce aisle.

Periods of drought have bedeviled the Southwestern U.S. for millennia, but this latest stretch of dry weather coupled with extreme heat has proved especially persistent – and perilous. Dry streaks lasting more than two decades are known as “megadroughts,“ and the drought drying up the Southwest is now rolling into its 22nd year.

Scientists say that’s in large part because of human-caused climate change, which is responsible for 42% of the lack of moisture in the region’s soil over the past two decades, according to a University of California, Los Angeles study published in February.

Federal officials announced last year that they were holding back water for the first time in the Colorado River Basin’s history, restricting water allocations to the states relying on the 1,450-mile waterway. Officials said in June that more restrictions may be on the way.

“It’s pretty bad,” said David Feldman, a political scientist who is the director of Water UCI, a water science and policy think tank at UC Irvine.

Growing global issue

“What we’re seeing in the Southwest is becoming more globally common,” said Feldman, who also works as a government consultant on water issues and has written several books about water policy and management. “The unpredictability of precipitation, the growth in various demands for water, not just for agriculture but for urban use, for industry, for manufacturing–it’s a problem that’s becoming common worldwide.”

Insurance provider Swiss Re estimates drought, wildfire, flooding and other climate-related events will hack up to $23 trillion off global economic output–a reduction of between 11% and 14%–by 2050.

Figures like that are underscoring the need for more accurate climate-related financial models, which insurers can use to build better and more broadly available climate insurance products, said Valentina Fomenko, who advises the World Bank on environmental initiatives and is the founder and CEO of business consultancy Strategy DNA.

“There needs to be this shift in thinking about how we assess risk,” she said. “We need analytics to be able to project water availability and potentially translate that into financial terms.”

Insurance companies are already citing drought and wildfire risk as reasons to hike rates or cancel policies. And credit ratings firm S&P Global said in a report last year that water shortages are increasingly stunting industrial production and gutting companies’ access to capital.

The agriculture industry, one of the world’s biggest consumers of freshwater, is trying to stave off those problems and protect its crops by adopting more data- and AI-backed technologies.

Growers, who since the presidency of George Washington have relied on the Farmers Almanac’s weather predictions to know when to plant crops, now turn to petabytes of data in the cloud, which AI models harness to make far-more-accurate predictions and automate a farm’s operations.

Tractors drive themselves in razor-straight lines. Instead of spraying outsized expanses of cropland with pesticides that can run off into local water supplies, cameras equipped with computer vision target clusters of weeds with high precision. Sensors in the ground can trigger the flow of irrigation water, also with far greater precision, which preserves water supplies.

All of these measures make it possible to grow more food on less land, which also saves water, said Curt Blades, a senior vice president at the Association of Equipment Manufacturers, an industrial and agriculture trade group.

“There are no silver bullets on this. Technology is only going to go so far to solve some of these challenges,” Blades said. “We can manage it. We can do more with less.”

Cervest, which assesses climate risk to corporate assets using complex machine learning models and vast repositories of data, recently demonstrated its software by zeroing in on the Hoover Dam, which straddles the Nevada and Arizona border and helps channel the waters of the Colorado River to more than 40 million people across the Southwest.

For 87 years, the dam has been a pillar of the Southwest’s distinct anything’s-possible promise of prosperity. Colorado River waters have long rinsed the salty Pacific from surfers along Los Angeles beaches, formed the ice cubes rattling the glasses of Las Vegas card players and irrigated the region’s sprawling ranches and farms. The dam’s hydropower-generated electricity, racing along 2,700 miles of copper transmission lines, has helped make Los Angeles County the U.S.’s top manufacturing center, with a GDP of $747 billion as of 2020.

Cervest’s software, called EarthScan, can zero in on a single corporate campus, factory, distribution center, data center – or even a monolithic piece of federal infrastructure like the Hoover Dam. It then uses artificial intelligence, poring through petabytes of data, to determine whether heatwaves, drought, wildfires and flooding are likely to strike those assets.

What did EarthScan’s AI see when it looked at the Hoover Dam? Trouble.

If the seven states tapping the Colorado River continue doing business as usual, according to Cervest’s data models, “heat stress” around the Hoover Dam will keep rising. The software’s algorithms currently give the dam a risk rating of “high” and project that rating will spike to “very high” by 2025.

Among the data the AI examined to reach that conclusion: Lake Mead, the U.S.’s largest reservoir, which was created when the Hoover Dam sent the Colorado River sloshing over the Mojave Desert just 30 miles east of Las Vegas, has receded to unprecedented lows–about a third its normal level. Lake Powell, created by the Glen Canyon Dam, also on the Colorado, has similarly shrunk.

The Hoover Dam is owned by taxpayers, not private interests. Still, its current woes could cost billions of dollars to the industries that rely on it for water and power, said Karan Chopra, Cervest’s chief operations officer. A further-escalating Southwestern drought, he said, could plow under farming operations, choke supply chains and stall manufacturing lines.

In fact, he said, Cervest’s 75 customers are using its EarthScan software, not just to monitor risk to their own buildings, but to vet supply partners. Institutional investors, he said, are using the software to assess risk across their portfolios and prospective investments.

‘No place to hide’

In March, the SEC proposed new rules that would require public companies to disclose in SEC filings the climate-related risks facing their enterprises, particularly where they relate to operations and finances. The proposed rules would also require those companies to reveal greenhouse gas emissions. The agency extended the comment period for proposals in May.

“There will be no place to hide,“ Chopra said of the expectations for increased climate disclosure. “My advice is to get ahead of it and use it as an opportunity for competitive advantage.”

Climate models are complex and difficult to interpret, and that work has typically been left to academics and other experts who draft thick reports for policymakers, “which is really important,“ Chopra said. But he added that Cervest is trying to make the models easy to apply across a portfolio of individual assets “so we can all get smarter.”

The software, he said, can identify current and potential risks based on different scenarios. It can highlight climate risk patterns as far back as 1970 or project risks into the future as far as 2100, he said.

“Imagine if everybody has access to this intelligence,” said Chopra, who said Cervest plans to make its software more broadly available in the near future. “Because that’s actually what we need in the world to start planning and preparing for what we are living through and what we will be living through.”

This story is part of The Path to Zero, a special series exploring how business can lead the fight against climate change.