Ep. 026 - PJM's $12B Modeling Mistake Is Hitting Ratepayers Again (Datacenter, Energy) | Robert Boswall, Jordan Nanos

**Robert Boswall** (0:00)
This is my mother's background. We got crystals that help you sleep.
We've got Buddha. We've got a kiwi or an...

**Jordan Nanos** (0:08)
Looks like a kiwi. Yeah, that's a kiwi from New Zealand, right? They lay the little green eggs that turn into fruit.

**Robert Boswall** (0:18)
Yeah, that's the one.

**Jordan Nanos** (0:19)
I love those, they're really tasty.
Hello everyone, welcome back to SemiAnalysis Weekly. We've got episode 26 and we're here with Robert Boswall. We're going to talk all about US rate payers wasting $12 billion of money that they don't have to waste because PJM is making a modeling mistake and they want to do it again. Rob, welcome to the show, man. I appreciate that you're there in your man cave with the crystals and the kiwi, it's looking great.

**Robert Boswall** (0:56)
Yeah, I got to balance out, bring out my feminine side, I guess. Yeah. I mean, I love the crystals. He doesn't love crystals.

**Jordan Nanos** (1:06)
We're going to have a nice time in the podcast where we talk about the electricity grid in Pennsylvania, in Maryland today, in Virginia, okay? Okay. Let's start with the basics. The high-level call that you guys are making here is that, well, first of all, maybe you can explain what PJM is, like, really basic, what's happening in the electricity grid in the US right now, and why Americans are going to see their power bills rise.

**Robert Boswall** (1:32)
Well, yeah, that's the thing, they don't necessarily have to. So PJM stands for Pennsylvania, New Jersey, Maryland, but it's actually 13 states. It's Delaware, Illinois, Indiana, Kentucky, Maryland, Michigan, New Jersey, North Carolina, Ohio, Pennsylvania, Tennessee, Virginia, West Virginia. And on top of that, the District of Columbia. It's not all of all of those states, some of them are partial, but yeah, it's a bit of a mouthful for the acronym to be all of the states.
So this is-

**Jordan Nanos** (2:03)
But this is US East 1, like this is data center alley.

**Robert Boswall** (2:08)
There's a 66 million people, this is data center alley, this is the largest grid by electricity delivered in America, or the largest sort of operational area in America.
And it's also what's driving a lot of the narrative around data centers driving up electricity prices. Like hundreds, maybe thousands of articles. And I always kind of just check this whenever it says that data centers lead to increased electricity rates, I always see what the source is, and it's almost always specific to PJM. Like this is not- there are some very small instances of this happening outside of PJM, but PJM is a particularly large driver of both the narrative, but also it's the area that is seen by far the largest rate increases due to increase in demand. And that's, yeah, we are take, and we've written about this before, and we're writing about this again from a slightly different perspective this time, is that that's due to market design, or not even market design, more like sort of bureaucratic auction design.

**Jordan Nanos** (3:12)
Okay, let's dig into that specifically. PJM is unique when compared to other grids even in the US in the way that they do these auctions, right?

**Robert Boswall** (3:20)
Yeah, so it's normal in electricity, in electricity markets to have kind of real-time markets and then like more ahead-of-time markets. So you just, you want to build in a buffer to how the electricity system runs. And so you run these auctions years in advance to make sure that you have enough spare capacity on the system so that in the worst-case scenario, it operates well. And normally, you actually only run those markets for either new capacity because they need long-term contracts, they need a degree of certainty to get built, or you're running them for like refurbishments or yeah, life extensions, things like that. You're basically just trying to keep some give in the system. And PJM is unusual where they run this for the entire system. And so every power plant can get into this, like fully existing ones that are completely paid off.
And what that means is that any price increase has like a massive multiplier effect across the, if you think about like an X and Y axis, you have a small price increase on the Y axis that's multiplied across the entire X axis. And because this is a very large grid, because it's the large grid in America, that can lead to some quite large price spikes, which is what we've seen.

**Jordan Nanos** (4:39)
Yeah. And this is definitely why maybe people talk a lot about how as AI data centers or just energy demand in general in the US increases over time, we need to solve this with base load, but also variable capacity because there's fluctuations in demand throughout the year.