AI. And we like to refer to it as the nervous system for AI, right? If you think about the GPUs as the brain, The brain can only function with a nervous system that can alert it, that can help the body react, and in this case, AI's body react. 

So we see a world where the network has intelligence built in to know, hey, wait a minute, those GPU clusters are nearing capacity or there's an energy limitation in that geographic location. 

Let's just reroute the traffic to a different location where there is energy, where there is capacity. 

But to do that, you have to have a programmable layer that lets you do that with little to no latency, on demand, from anywhere to anywhere. 

And that's what we're building.