> The decision to build or rent may depend on a user’s data intensity. Think of brick-and-mortars real estate (e.g. office space) which for most companies is c

> The decision to build or rent may depend on a user’s data intensity. Think of brick-and-mortars real estate (e.g. office space) which for most companies is completely outside their core value proposition hence they would be most inclined to lease. Accurate statistics are surprisingly non-trivial to find but it seems like the company owned segment is circa 20% (versus 80% leasing their office space needs). Now when you think of the role data centers play in a growing digital economy, for many companies increasingly their value proposition is linked to their digital infrastructure. So that would push the fraction of "owned" versus "leased" substantially higher. This would leave still a eye-wateringly massive amount to be served by "hyperscalers". This is really what the "cloud revolution" (renting somebody elses linux server) is mostly about. There is a certain naturalness to this state of affairs. But there is the major catch. Imagine if practially all the real estate in the world is owned by just a handful of mega-corps and anybody needing office space needs to rent from them. Companies small and large will quickly realize what the term "rent-seeking" [1] means... when the seller holds all the cards and the buyer holds none. Rationally a hyperscaler oligopoly would attempt to keep the price of digital infrastructure at just below the ownership cost, though greed and various frictions in migrating back to in-house might temporarily allow milking higher rents. In any case its a most interesting parameter to observe in these rapidly changing times [1] https://en.wikipedia.org/wiki/Rent-seeking https://en.wikipedia.org/wiki/Rent-seeking