Yeah, let's say our goal is we want to aggregate and turn heterogeneous fragmented compute into a unified AI platform to make compute unified, productive, and liquid. 

Explain what that means. Yes. So we have two layers. The bottom layer is we're building the first global compute grid. So if you think about this, right, like compute is the next power that drives AI. 

So it's basically similar to electricity. We're starting to see a lot of people building new clouds, AKA power plants. They have GPU clusters, AKA GPU clusters, AKA generators. 

And so if you look at the power market, they're like a lot of financialization built on the grid system. 

If there is no way to connect the financial world with the physical world, they will introduce a lot of like a basis risk. 

They're not much volume, et cetera. And so what we think the most important thing is to actually build the first grid system for ai compute so that now every ai companies and enterprises will be able to draw compute elastically. 

And then also whenever, whenever they have idle compute, they can sell it back.