Cloud waste isn't a competence problem. It's a nobody's-job problem.
It comes up constantly in cloud forums, always with the same exasperated tone: “It blows my mind how many startups just let resources run 24/7. Doesn’t anyone actually review their cloud spend?” And the replies are always the most useful part, because they say the quiet thing out loud: the people running up the bills aren’t stupid. They’re outnumbered.
Waste isn’t incompetence. It’s an ownership gap.
Here’s the reframe that actually explains the six-figure surprise bill: cloud cost has no owner. The one person nominally doing “devops” is also doing infra, security, on-call, and the last three things that caught fire this week — for five teams. Cost is on that list somewhere, below “keep the site up.”
As one commenter put it, the same companies shocked by a huge AWS bill are the ones who decided a single overworked engineer could own infrastructure, security, cost, and on-call across the whole org. You don’t get disciplined cloud spend out of a team in permanent fire-fighting mode — you get whatever’s left after the fires, which is nothing.
That’s not a competence problem. It’s a leadership-and-staffing problem wearing a competence costume. And it’s why the usual advice — “just review your bill every month!” — fails: “we’ll audit it later” is, functionally, a decision to accept the waste. Later never comes, because later is also on fire.
Why the tools don’t fix it
The instinct is to buy visibility: a cost dashboard, anomaly detection, a nicer Cost Explorer. But those don’t create an owner — they create more alerts for the person who already can’t keep up. Another tab nobody has time to open. A tool tells you you’re bleeding; it doesn’t stop the bleeding, and it certainly doesn’t do the migration. The missing piece was never visibility. It’s someone whose actual job is this.
The fix is an owner, not an audit
There are only two honest fixes, and they’re both about ownership:
- Resource it for real — give someone the mandate and the protected time to own cost, not as the ninth thing on a firefighter’s list. Most early companies won’t; the headcount goes to features.
- Rent the owner — bring in someone from the outside whose entire job is to find the leaks, carry out the migration, and keep it boring month after month, so your one overworked engineer doesn’t have to fit it between fires.
That second one is the honest version of what I do. Not “you’re doing it wrong” — you’re not; you’re doing four jobs. It’s “nobody has the time to do this one at all, so let me be the person who does.”
And it’s about to get worse
Because that same drowning team just shipped an AI feature. Now there’s a second unowned meter — one that moves faster than cloud and can be abused on top of overspent. If nobody has time to review the AWS bill, nobody is watching the OpenAI bill either, and that one can 10× overnight. Same ownership gap, new and sharper surface. (More on that soon.)
If your one overworked engineer is the reason cost never actually gets looked at, that’s the exact gap I fill — from the outside, for cloud and now AI. Every message comes straight to me — I read and reply to each one myself, usually within a day, and what readers send shapes what I build next. It’s just me for now, so that’s genuinely true; it won’t be forever. Send me a recent bill and I’ll show you what’s leaking — free, within a business day.