Outerbounds’ Post

Transcript

As AI and analytics capabilities continue to advance, what do you see as the next big frontier for data processing technologies, Josh? Um, I mean, this isn't even self-serving the the next frontier. It is around we, we jokingly, we call it space mice. So we have the space charts and we talk about mice modulating interoperability, composability, extensibility, space scale performance and cost efficiency. But is that data is getting bigger and it's being forced by AI generates more data, people interact more with data. Data begets more data and so the data set size is just getting bigger. And there is this like how do I get data faster into these systems? Now they can ask better questions. My data, how do I analyze it better? And so speed and performance matter, but we have finite energy and so. You know, another thing I'm excited about is green. You know, I've, I spent, you know earlier in my life about 30 days in the Alaskan wilderness backpacking around Denali and then went back then area. You know, 17 years later and like. I don't care where people stand on global warming or any of this other stuff, the, the, the Alaskan, you know, mountain caps, snow caps are like retreating away. And like the dog sledding is like you're like sledding on mush. And like, I'm just like, I know 17 years ago, it's very different than this. And you talk to the locals, they're like, yeah, it looks very different a decade ago. And so I, I do think, you know, being energy efficient matters a lot. And so, you know, even a Voltron data, we, we recently were, you know. Our first cluster, you know, bunch of A1 hundreds and, you know, some wildly impressive TCO energy consumption, you know, 50-70 X faster, you know, attempt the energy consumption. And then we really looked at like some of the data and some of our design such things and we went back to the drawing board and like we could probably do this even greener, like, you know, really push energy consumption down and make it faster. And so yeah, when I when I talk to engineering team and I I saw some of their. Results and not gonna do a spoiler alert for kind of our next quarter launch, but faster and less energy consumption. And I think that's where we have to go. We have to bring down the the energy use of these large scale systems. And so it's not just to be like, oh, who cares? This job is you know, if it's 20 seconds or one second, the real thing we should be asking is like how many joules did that job take? You know like how much. Energy did you use to get that answer? And so I think the more energy conscious we get, we get this byproduct of performance, uh, kind of for free. And so I think energy, thinking about energy, energy consumption, data center consumption, you know, we, we, we like to joke about curse on, we can't be Curzan. We can't have a, you know, a city, you know, globe where, you know, there's just data centers everywhere.