Energy Shots #107:

Demand is infinite.
Conservation isn’t a solution.

Tech giants’ race to capture AI market share faces one critical dilemma:

• Prioritize the energy transition at the expense of technological progress? Or pursue technological advancement and eventually unlock new energy frontiers?

For the past three years, Silicon Valley’s response to this dilemma aligned with the IEA’s Net Zero by 2050 — i.e., limit investment in traditional energies like natural gas.

However, as long understood by industrial sectors, these unrealistic policies are not suitable for energy-intensive processes. With the energy needs of new technologies like AI expanding exponentially, Silicon Valley’s response to this dilemma has subtly shifted to a new perspective:

The Hierarchy of Energy Needs — energy must be abundant, affordable, and reliable.

This ‘about-face’ was evident to observers of last week’s inaugural SCSP AI+Energy Summit, where former Google CEO and SCSP chair Eric Schmidt candidly addressed the tech industry’s evolving stance in a 45-minute Q&A.

The Takeaway: Using Schmidt as a proxy for broader AI leaders indicates a material shift in energy priorities favoring near-term adoption of proven thermal generation resources like natural gas.

• Renewables, batteries, and nuclear are either too inconsistent or too undeveloped for large-scale bets that require reliable power.

• Climate goals are 1) not the industry’s top priority and 2) should not be used to limit AI and technological advancement.

• AI-related power demand growth is “infinite” and potential efficiency gains will be “swamped by the enormous needs of this new technology.” The industry is “not going to get there through conservation.”

Seven excerpts from this discussion that warrant energy industry participants’ attention are included below.

ES #107: Eric Schmidt on AI's Energy Needs

• Where does ‘energy’ rank in AI’s top problems?

• Why does energy pose a problem?

• What is AI’s power mix?

• What about nuclear?

• Can conservation solve AI’s energy problem?

• What about climate goals?

• What does AI need from the energy industry?

Where does ‘energy’ rank in AI’s top problems?

I do a weekly call with the heads of all the AI companies… And, in this call, we rank the problems… Energy is the number one or number two problem. The other one is Washington.

Why does energy pose a problem?

[In] terms of energy, the current belief is that we run out of power in the U.S. in roughly four years.

People come to those numbers by various ways and they're all approximations, but it goes something like this.

Nvidia built this huge chip, which on most metrics uses less power… and this huge chip puts out more than a kilowatt… That package needs liquid cooling. So that's causing the data centers to have to be upgraded, which is a non trivial expense.

But the important thing is when you do the math… and you assume a cost learning curve, you get huge improvements in energy efficiency per calculation.

The problem is that if you look at what they want to do with it.

They are 10 times more [computationally] expensive for the thing that they're substituting…So, you have this enormous benefit that the physicists are giving us to be larger chips. Which means that you want to be on the latest chips, and the latest chips in aggregate use more power.

What is AI’s power mix?

Let's go through the choices. New solar and new wind are materially cheaper than new gas. And yet you have to put in gas peaker plants for the two weeks in which the wind doesn't blow.

How long does it take to go from the lab and AI to a better battery? Five years, six years, seven years. The learning cycles are just too slow… And the people are working hard on batteries, there’s a lot of money there. If you could build large storage batteries, you eventually run out of the ingredients of those batteries — the demands are so great.

I think the most important thing that we can do right now is to get on a war footing about producing more energy in the places where people are going to use them. There’s plenty of energy. It’s in the wrong places… I look at the Renewable Portfolio Standards, and here’s the problem:

If you assume the RPS is maybe, depending on what numbers and assumptions you make, 20 or 30 percent of American planning and current planning, that still leaves 70 percent in the traditional power plants.

And we're running out of baseload unless you're willing to build a lot more coal, which we're not willing to do, and more gas, which we are willing to do.

I don't see how you can get much above 50 percent in renewable. Which means you're always going to have the intermittency problem, this is my opinion. The intermittency problem between the two, and that problem has to get solved.

What about nuclear?

History will record the German decision to shut down their perfectly fine working fission plants. It's a terrible idea… I appreciate all the concerns, but this is a hungry world and we're going to build it. We're gonna build the energy one way or the other.

So I’m going to put in my purchase order [for modular nuclear plants] today. Okay? Today. And for purposes of argument, the estimate is two billion dollars. We all know it'll be ten, but I put in my order for two. Because that's how much money I convinced the banks to give me. Who do I give it to? Westinghouse just pulled out of South Carolina.

The best SMRs, small reactors, are basically in Korea. So they're not fully certified here… I can go on.

Can conservation solve AI’s energy problem?

“I was at a dinner with Bill Gates and Andy Grove once. And they had a huge fight, it was years ago. And the saying was, Grove giveth, Gates taketh away. So the hardware guys get better, and they do a really good job.

And the software people just need more. And I have every reason to believe that in this phase of AI, the same is true. It won't be Grove and Gates. It'll be some other people.

But the fact of the matter is the demand is infinite, even assuming everything that you're implying in your question. So let's go through that.

1. First, we've already discussed better materials for batteries. That takes a while, but that's important.

2. You move to DC power lines for better transmission and loss control.

3. You put the data centers next to the big power lines, as I described earlier.

4. Architecturally, there are huge improvements to be made in, essentially, the cost effectiveness per transaction.

All of that will be swamped by the enormous needs of this new technology… I can assure you that we’re not going to get there through conservation.

That’s the key point, and the economics will drive it anyway.

No large company wants to have a huge power bill. Most of the people I’ve talked with say the power bill is becoming a very large component of their expense.”

What about climate goals?

My own opinion is that we're not going to hit the climate goals anyway, because we're not organized to do it. We're not, and the way to do it is the things that we're talking about now.

And yes, the needs in this area will be a problem, but I'd rather bet on AI solving the problem than constraining it and having the problem.

What does AI need from the energy industry?

1. What the industry needs from you all is more power that is predictable, that is baseload or baseload equivalent as soon as they can.

2. They need more places to site.

3. They need more ways of getting things built.

4. And they need more [ways] to get them connected.

This commentary contains our views and opinions and is based on information from sources we believe are reliable. This commentary is for informational purposes, should not be considered investment advice, and is not intended as an offer or solicitation concerning the purchase and sale of commodity interests or to serve as the basis for one to decide to execute derivatives or other transactions. This commentary is intended for Mobius clients only and is not considered promotional material.