Discount theatre: when your cloud bill needs a spreadsheet, not a deploy
You came to ship software. You stayed to model amortized 3-year commitments.
That is the moment most teams meet their cloud’s pricing for real. Not the sticker price on the instance page, the other price: the one that only appears after you’ve built a spreadsheet, asked finance for sign-off, and bet on what your workload looks like in 2029.
The maze
Every hyperscaler ships the same set of discount instruments, dressed up in different names. Learn one and you’ve roughly learned all of them.
- Reserved instances — commit to a specific shape, in a specific region, for one or three years. Get a discount. Get locked to that shape. Need something bigger next quarter? You’re now paying for the old one and the new one.
- Savings plans — commit to a dollar-per-hour spend instead of a shape. More flexible, harder to reason about. You’re guessing your future baseline and signing a contract against the guess.
- Spot / preemptible — cheap capacity that can vanish with a couple of minutes’ notice. Great, if your workload tolerates being killed mid-run. Now you’re writing checkpointing and retry logic to chase a discount.
- Committed-use discounts — same trade in a different wrapper: promise a minimum, save some, eat the overage or the shortfall.
Each one is a real lever. Each one also costs something the pricing page never lists.
The hidden tax
The discount is the part they show you. The tax is the part you pay either way.
Time. Someone on your team now owns “cloud cost.” They model commitments, track coverage, true up quarterly. That person is usually an engineer who would rather be writing code. You’re spending senior attention on procurement math.
Lock-in. A three-year reservation is a three-year bet that you’ll still want that exact shape, in that exact region, running that exact much. Architectures change. Reservations don’t. The discount is real only if the future cooperates.
Risk. Over-commit and you’ve prepaid for capacity you can’t use — a sunk cost dressed as a saving. Under-commit and you’re back on the expensive on-demand rate for the part you guessed wrong. Either way, the house wins. The whole structure exists because the variance pays them, not you.
And here’s the quiet bit: the on-demand price is set high on purpose. It’s the anchor that makes the discounts look generous. You’re not being rewarded for committing — you’re being penalized for not.
The honest number is the only number
We didn’t build a discount program. We built a price.
On Kaligon, every resource has one flat rate. No reserved tiers, no spot bidding, no savings plans to model, no committed-use math. Billing is per-second, capped monthly — after roughly 730 hours the meter stops, so anything you leave running just settles at a fixed monthly number. No prepayment to forecast, no shape to lock, no spreadsheet to maintain.
The number you see on the pricing page is the number you pay. It’s the same whether you commit for three years or three minutes. It already lands roughly 30–60% under hyperscaler rates on equivalent VM and storage shapes — without you signing anything, modeling anything, or betting on 2029.
No egress between your resources in the same region. No charge for snapshots, backups, API calls, or support tickets. Per-project budgets that actually stop spend, and a hard monthly cap so the bill can’t surprise you.
The cheapest way to run a cloud isn’t to out-negotiate it. It’s to use one that doesn’t make you negotiate at all.
Go back to shipping software. We’ll keep the price boring.