What is that piece really arguing? And why are we watching the same pattern again with organizations buying AI by the token?

I've been in the industry for coming up to 20 years now.

So I've been through the phase of customers and organizations moving from on-premise into the cloud, getting into the cloud and thinking, oh, actually, this is a lot more expensive than I thought.

And potentially I need to repatriate and go back into the cloud.

But the journey about moving into the cloud needs to be done properly, right? It's not a case of lift and shift.

It's around modernizing and optimizing from that on-premise state into the cloud.

And that's where cost control has spiraled.

What happens with the speed of AI developments to this stop?