Maintenance capex is 4% of revenue in one filing and 11% in another
I've been reading through public filings for the listed data center owners because I wanted to understand what a passive holder actually pays out of the rent before anything reaches them. Two of them report recurring maintenance capex on wildly different scales relative to revenue, roughly 4% in one and something over 10% in the other once I stripped out what looked like expansion spend.
I can construct two explanations and I can't tell which is true. One is a lease structure difference. If the tenant owns and maintains the mechanical and electrical equipment and the owner holds the shell and the power path, the owner's recurring spend should be low, closer to industrial. The other explanation is age and density. An older facility built for low kilowatts per rack needs cooling and distribution rebuilt to stay leasable, and that spend shows up as maintenance rather than expansion because it doesn't add capacity.
Why it matters for me: if I assume 4% and the truth is 11%, my distributable yield estimate is wrong by a margin that swamps every other assumption I've made. So the question I'd put to the room is what number a passive holder should actually carry as a reserve when the disclosure won't separate keeping the box leasable from making the box bigger.
What recurring capex reserve would you carry on a passive data center hold?
23 votes