A legitimate reason to do this is so that you can sell it all off or restructure it later without it being intertwined with your main books. So, for instance, it you’re an AI company, your main business is developing models and selling inference, but you need data centers to do that. You could buy those data centers yourself, or you could create a data center subsidiary that would take on debt and build your data centers and then rent them back to you. In the future, if you decide you don’t need that capability any longer, you just sell the subsidiary and you don’t have to tease apart the P&L and personnel to do it. It’s clean and separate because it was structured that way up front.

Now, that’s not to say that there aren’t other benefits of having a separate balance sheet related to moving numbers around. But it doesn’t have to be nefarious.

The structures theyre using reduce risk. The 2 big moves here are 1. agreeing to lease a datacenter from developer instead of building it yourself and 2. using subsidiaries to finance and own the datacenters. Both these moves exist so that the company isnt left holding the bag if something goes wrong with the construction. They dont need to worry about zoning and construction because their partners are doing that. This also means the companies dont need to take out debt themselves, since these agreements dont count.