AI has a water problem

In high school, I was president of the Science Club for a year. And I participated in the Science Olympiad for two years.

The Science Olympiad is a series of competitions designed to test the ability to apply scientific knowledge. The Science Olympiad calls itself the premier STEM competition in the country. When I participated, the phrase STEM didn't exist. We were kids who liked science and wanted an extracurricular activity for our college applications. We also liked to win. And we would go to extreme lengths to win. But that's a story for another day.

I participated in three events -- Egg Drop, Science Bowl and Water, Water Everywhere. Just in terms of cool factor, Egg Drop was the best. There were two versions of the competition. But the goal was always the same. Drop an egg from at least two stories up and make sure it lands as quickly as possible without breaking.

Science Bowl is like Quiz Bowl. But there are only science questions. It was usually the last event in the competition and drew a serious crowd.

Water, Water Everywhere was my specialty. The first half of the competition was a quiz about water. The second half was building a device to measure how much salt was in a variety of water samples.

I loved building those devices. It was a chance for my imagination to run wild. I always wanted to build a device that measured salinity so precisely that the competition never stood a chance. We (I had a teammate each year) won Water, Water Everywhere both years that I competed. And it was that competition that instilled my lifelong curiosity about water, water systems and the different uses for water.

That's why two headlines caught my eye recently…

• Google Is Really, Really Thirsty

• Western states reach 'historic' deal to help save Colorado River

Most people don't spend much time thinking about water beyond the basics -- drinking, residential use and farming. But water is used for much more than that. It has industrious uses including fabricating, processing, diluting and cooling. And Google needs incredible amounts of water for cooling.

"In 2022, total water consumption at our data centers and offices was 5.6 billion gallons — the equivalent of what it takes to irrigate 37 golf courses annually, on average, in the southwestern United States," wrote Google in its 2023 environmental report. "Wherever feasible, we try to use non-potable sources of freshwater and alternatives to freshwater," Google added. And as Business Insider reports, that's a 20% increase in water usage from 2021.

Much of the increase in water usage can be attributed to artificial intelligence. Artificial intelligence is insanely energy intensive. And we're only in AI's infancy. As Google, Microsoft, Amazon, Meta and a host of others push AI usage and adoption to stratospheric levels, the need for water to serve as a coolant will increase significantly.

The significant need for water as a coolant comes at the same time water scarcity is becoming a major issue on the West Coast.

Arizona, California and Nevada agreed to give up 13% of their Colorado River water allotment because water levels in the river are dropping to dangerous levels. And this is just the beginning. The seven states that use Colorado River water have to reach a new agreement by 2027.

The original Colorado River water compact was signed in 1922. It was a particularly wet period then, making the treaty easier to negotiate and sign. All sides thought -- or pretended to think for the sake of negotiations -- that the river system had much more water than it did. And it allocated water accordingly.

But a historic drought that began in 2000 combined with an explosion in usage -- Las Vegas has grown by 800,000 people since 2002 for example -- made water a scarce commodity.

How bad are water levels? In 1999, before the drought began, reservoirs were about 92% full. Now they're about 25% full.

If it hadn't rained significantly this year, reservoir levels would have fallen so low that hydroelectric power production could have stopped and downstream communities could have lost their water supply.

Now we're adding AI to the mix. And we're not stopping to ask questions about what this means for the future of water.

The answer could be it will have little impact because tech companies will largely use freshwater alternatives. The answer could be it will have a large impact because water technology isn't developing quickly enough. We don't know. But there's no free ride here. AI is going to expand at a rapid pace. And that's going to increase the demand for water.

Dozens of startups are already trying to solve this problem. The problem is developing water technology that can scale is both fairly expensive and time consuming. A smart governor or politician would enter a public-private partnership to solve this problem. But there's a shortage of smart politicians.

That's why smart investors have an opportunity in this space. Investors that can back the right startup (presumably one with great technology and access to regulators and politicians) have a reasonable chance to cash in on a moonshot. And for startup investors, those are pretty good odds.

The next time you look at a glass of water, take some time to think about how it got there -- and what other uses water has. It can unlock a whole new way of thinking in life -- and business.