00:00 Speaker A
If you look at the United States, more than half of the electrons produced each day in the country are produced by our equipment. So our installed base represents a very significant portion of the global system, approximately one-third outside of China. So I just want to give that context and say that we work across all industries and support all long-term dynamics. Of the current backlog, only about 10% of the backlog is hyperscalar and AI-centric. Over the next three years, this could grow to about a third, but now it's 10% of our business.
00:41 Speaker A
After all, there are many other drivers that are currently in demand. It's also just the electrification of buildings. It is the reindustrialization of the United States, the growth of many industries outside the United States, and ultimately the electrification of transportation. So there are a lot of drivers. AI is one of them, but one of many.
01:05 Speaker B
Well, this helps us understand the larger spectrum. But inside a gas turbine, it's going to be at a higher concentration, right?
01:14 Speaker A
yes. That's about one-third. Understood. This will determine about a third of your future business. As an example, this quarter we signed large contracts in Vietnam and Mexico. So there are certainly other international markets. Consider that Saudi has been a very big gas market for us in recent years. Taiwan is kind of AI-centric because we're driving TSMC's growth, which is ultimately related to data centers, but if you really want to say hyperscalar direct or near direct, about a third of our business is in the gas business.
01:58 Speaker B
So even if you look beyond hyperscalers to the larger scope that you're talking about, you're obviously familiar with the big debate around the spending that's currently being spent on computing, AI computing.
02:11 Speaker A
yes.
02:12 Speaker B
So it can be a double-edged sword. While this is great for business, there are also concerns about overcapacity and curtailment. Not worried about it at all?
02:22 Speaker A
of course.
02:24 Speaker B
Yeah.
02:24 Speaker A
Well, I'd like to start with what we see in our work. And the reality is part of how we're growing our margins and, you know, as we discussed with our investors yesterday, our EBIDA margin this year is 8.5%, increasing to 20% by '28. Part of that is the significant volume growth and, to a lesser extent, engineering growth, the variable cost productivity that we're driving. That's because we're investing in AI to improve productivity. So let's take a step back and consider that the $130 trillion global economy and half of its workforce and services are in one direction. If AI impacted 10% of that workforce's productivity, it would generate benefits of $6-7 trillion per year. And you hear a lot of discussion about investing hundreds of billions of dollars. What I'm saying is that it's still early and we think there's a lot of attractive returns in what we're seeing within Vernova and in general the whole interaction with hyperscalers. Therefore, I have confidence in the construction. And frankly, we're probably responding to a smaller demand than most people are reading about.
