For most of the last decade, cloud spend was a rounding error next to headcount. That era is over. With capital more expensive and boards asking harder questions, the cloud bill has become one of the first line items founders are told to defend, and one of the easiest to get wrong.
The good news: unlike layoffs, most cloud savings are painless. They come from waste, not muscle. Here is how the sharpest teams are finding 20–40% without touching a single feature.
Start by measuring what you actually use
You cannot cut what you cannot see. The teams that win at cost start by tagging every resource by team, environment, and service, then piping that into a single dashboard everyone can read. Once spend is attributed, the conversation shifts from ‘the cloud is expensive’ to ‘this one pipeline costs more than the rest of staging combined.’
Quick wins we look for first
- Idle and orphaned resources, unattached volumes, old snapshots, forgotten load balancers.
- Over-provisioned instances running at single-digit CPU all week.
- Non-production environments left running overnight and at weekends.
- Egress traffic that could be cached or moved closer to users.
Then go back to the negotiating table
Commitment-based discounts, savings plans, reserved capacity, committed-use contracts, are where the real money sits, but only once your usage is stable enough to commit. Bring twelve months of clean usage data to your provider and the discount conversation changes character entirely.
“The clients who negotiate hardest are the ones who show up with data. A clean usage baseline is worth more than any amount of leverage on a call.”
Make cost a shared, ongoing habit
One-off cleanups drift back within a quarter. The teams that keep savings put a lightweight cost review into their sprint rhythm, set budget alerts before overruns happen, and give engineers visibility into the price of the resources they spin up. Cost stops being a finance problem and becomes a design constraint, which is exactly where it belongs.
