Americans intuitively know that, all things being equal and no sudden large-scale solutions emerging, we’re headed toward an energy crisis. That’s one reason that data centers are unpopular on a bipartisan basis. How do we change the math, to make AI a viable part of our futures, given the needs of the grid and the new energy customers popping up across the country?

I wanted to touch on three presentations that I heard at our Planet Action event this year. (Disclaimer: I help to host these events.) Each of these speakers provided their own ideas for mitigating energy problems due to emerge soon.

One such idea was to make data centers more flexible, so that they can run on renewable energy sources with more dynamic supply.

“These facilities are not just energy-hungry brains,” Ayse Coskun, Chief Scientist at Emerald AI, said, of this concept. “They can also be the muscles of the grid, flexing on command. Unlike our homes or hospitals, data centers run jobs that are predictable, controllable, and often playable. That makes them ideal to help balance supply and demand on the grid by making AI data centers power flexible.”

He noted the variance in task demand.

“Not all computing tasks are urgent,” Coskun continued. “Some can wait for minutes or hours, or can be slowed down without anyone really noticing. … Instead of asking how can we compute as fast as possible, we asked, how can we match the needs of the grid in our computing system while at the same time maintaining user performance agreements? This shift led to new strategies.”

“AI can learn patterns, anticipate grid needs, and orchestrate across data centers, across utilities, and even nations in real time,” Coskun said, invoking a musical metaphor. “Imagine a data center, or a whole network of them, as an orchestra, with hundreds of instruments all playing at once. Left on their own, it can sound like chaos, but bringing a conductor, all that noise suddenly turns into music.”

This conductor, Coskun suggested, is AI.

“AI can direct data center operation so that data centers can precisely match the grid needs depending on what power is available, depending on what the users need,” she added. “And the result is harmony, efficient computing, reliable electricity, and a system that works beautifully together.”

In closing, Coskun asked us to reframe.

“The real question isn't how much energy AI consumes,” she said. “The real question is how much flexibility, resilience, and clean power can AI unlock? If we are bold enough to rethink AI data centers, the very machine that now seems like a burden could be our greatest asset in building a sustainable AI future.”

Sourcing, and Resourcing, Rare Elements

“This green revolution that we've envisioned for so long is built on a backbone of metals and materials,” said Tomás Villalón Jr., “some of which we are just beginning to really get a grasp of what the supply chains look like and how we actually get them.”

Villalón is co-founder and CTO of Phoenix Tailings, a company that works to mine waste for rare earth metals. At Planet Action, he was presenting on how to get useful elements back out of e-waste: pointing to the scale of the problem, where an estimated 2.8 billion tons of metal material was sourced in 2021.

“What we've had to really do within the past 20 years or so is have some soul-searching as a scientific community, as an engineering community, and understand what it actually takes to get these materials out,” he said. “And only just now are we beginning to hear about some of these things.”

Villalón mentioned using bacteria to pull copper out of waste.

“Yes, we are in a crisis,” he said. “But in every crisis, there’s a beautiful opportunity, and we have to rise to the occasion.”

Later, Villalón mentioned the Russian disaster in Chernobyl, where a reactor core meltdown irradiated many acres and killed thousands.

“Chernobyl is a watershed moment in our history, because it was one of the greatest crises that we ever encountered,” he said. “We didn't know how to solve it. The reactor blew up. We had to deal with this immediately. And in nine months, through the effort of tens of thousands of people, the problem was fixed. So now, as we think about these next-generation technologies, the supply chains, the geopolitics, all of these factors that roll together, we fundamentally have to have that same level of innovation that the people at Chernobyl had. We don't have the option or the luxury to say otherwise.”

The clock, he noted, is ticking.

“This isn't the future,” Villalón concluded. “This is now. Let's go make the world better together.”

Emmanuel Kasseris, a mechanical engineer by trade and serial clean technology entrepreneur, started by addressing a major problem in the twenty-first century: waste.

“Our cities make waste,” Kasseris said. “Our agriculture and our animals make waste. … Humans make waste. And if you don't understand what I mean by ‘humans make waste,’ you're not thinking like my four-year-old. So waste is everywhere. And one of the worst things about waste is that as organic material decomposes, it generates this nasty molecule called methane. And methane is a super pollutant.”

Kasseris noted that one methane molecule can do up to 84 times the damage to the environment of a CO2 molecule.

But, he suggested, all is not lost.

“Methane can become a carbon negative super fuel,” he said.

The gist of Kasseris’s idea is the goal of taking the gas flaring, turning it into liquid, and porting it to where it can be used.

“(Make it) a liquid that can be put in a truck and taken to market, where it can be used to decarbonize shipping, aviation, trucks, or further processed into other chemicals,” he said. “This system would be on site. It would be modular so you can move it because the sources change. It would be deployed quickly.”

Here’s how Kasseris described implementation in conclusion:

“We're used to building these big plants because we know how to reduce costs by using economies of scale,” he said. “Another way we can reduce costs is leveraging economies of mass production. If we, instead of building field projects where everything is a one-off design, build these systems in a manufacturing facility where the design is standardized, the parts are off the shelf, we can really get to these sort of payback periods.”

The gains, he noted, will be substantial.

“About 60% of U.S. methane emissions are actually recoverable,” Kasseris estimated. “We can convert them using this type of technology to something like 27 million tons per year of green fuel. This would power 5% to 10% of our aviation fuel requirements globally, or more than 10% of our shipping fuel requirements globally, generating more than $27 billion of revenue.”

All of this, he said, can be done with waste from humans and animals.

“All of that can be recoverable,” he said. “What about the remaining 40%? Well, let's just say that that is lost in the wind. So imagine if we extrapolate this concept to the entire world and use these resources. What could we do? Let's stop wasting our waste.”

What do you think of these important pivots? Drop me a comment.