When you look back at the report this time last year, what were the things that you either missed or the things that you got wrong? And it's gonna sound like I'm marking your homework, but maybe, Thomas, you wanna start us off.

I think this is an important exercise for, for everyone to do when they're trying to analyze a topic, uh, that is so complex and that is moving so quickly.

So I would say, you know, broadly speaking, our report last year was very comprehensive, and it got most things right, uh, and it covered most of the issues that we've seen emerge over the last, uh, twelve months.

I would say, uh, there's one that we maybe missed, um, which we look at in depth in the, in the new report, and that's the rise of, uh, on-site power generation options for data centers.

Um, so this is essentially a powering option where the data center is not connected to the grid, but rather has built its own power plants on-site, uh, in order to supply its power needs, largely, uh, through natural gas.

And essentially, since our report in April 2025, we've seen a huge number of project proposals, uh, m- still mainly at the proposal stage, uh, emerge in the United States.

So we wanted to look at this issue and to see whether this was, uh, a real viable trend, uh, for, for data centers.

Maybe you could both talk us through what you think the big themes could be.

I, uh, wonder if you have anything to add, uh, on that front.

Yeah, I, I-- So I, I think, David, this is a good example of, you know, a trend, uh, that is being driven by AI that will impact on energy technologies and energy innovation.

So first of all, you know, why are data centers and, and AI, uh, data centers in particular becoming more power dense? That's really driven by the needs of, of AI.

So the largest models today are too big to fit on a, on a single chip.

They need multiple chips to work, uh, in, in tandem in order to train them and, uh, a- and run them.

And those chips need to be spatially proximate to each other, uh, in order to reduce, uh, data losses and data latency so that they can really work as, you know, a, a single computing unit.

And that means you're getting more and more chips packed in, uh, together, close together, uh, in these, uh, latest, uh, um, uh, uh, AI, uh, server racks.

Talk a word about that, 'cause that's-- say a word about that, as it's getting so much attention here in the US.