Hohpe writes short chapters, almost columns, on understanding the cloud, organising for it, migrating, architecture, building and economics. His claim is that this is not an infrastructure purchase but a change in how you operate, and that treating it as a purchase is what produces higher bills and no improvement.
What I took from it
The idea I use in every hosting conversation: savings are not included, they are earned. Moving the same thing as it stands onto a rented machine usually costs more, because you pay for being switched on and nobody switches anything off. Savings appear when the way of working changes, and that is a separate project which has to be said out loud before anyone signs.
The second: nobody wants a server. What a company wants is its system available, backed up and up to date without anyone having to attend to it. Put that way, the conversation about where the ERP lives stops being a price comparison per machine and becomes a comparison of who answers when something fails on a Saturday.
The third made me think hardest: avoiding lock-in has a price too. Designing everything so you can switch provider means giving up what that provider does well, and often you pay that insurance premium for a move that never happens. It is a decision with numbers behind it, not a principle.
And a fourth, managerial: automation is not about saving people, it is about getting the same result every time.
What I left out
What the format does not offer: there are no steps here, no service comparisons, no migration guide. These are essays, and a few restate the previous one with a different metaphor. It is also written from large enterprises with whole departments; anyone serving mid-sized companies has to translate the scale. Worth it anyway: it is one of the few books that argues about the decision rather than the tool.