We Tested Guaranteed Commitments on Our Own AWS Bill. Here’s What Happened.

Most cloud cost conversations end the same way: the savings are obvious, the math is simple, but the commitment feels like too much of a bet. So teams stay on on-demand pricing — paying a premium for flexibility they may not even need.

We were one of those teams.

At Alphaus, we build cloud financial management tools for a living. We know the RI/SP savings playbook inside out — we help MSPs and their clients execute it every day through Ripple. And yet, for a long time, we hesitated to apply it to our own AWS infrastructure. The concern wasn't the discount. It was the lock-in.

So we enrolled our own accounts in Guaranteed Commitments through Ripple — the same product we offer to end customers — and tracked every dollar in Octo, our FinOps platform. What follows is exactly what happened.

Our AWS infrastructure includes our Malaysia operations — a consistent, predictable environment with stable baseline compute that wasn't going anywhere. On paper, that's a textbook case for Reserved Instance pricing.

The problem was familiar: our product roadmap was active. New features shipping. Architecture evolving. Onboarding new customers meant scaling in ways that were hard to forecast a year in advance. A 1-year or 3-year RI commitment felt like we were betting against our own growth — and if we got it wrong, we'd be managing stranded capacity on top of everything else.

So we defaulted to on-demand pricing. Flexible, low-friction, and quietly expensive.

What changed our position was the commitment structure itself. Guaranteed Commitments through Ripple does not require a 1-year or 3-year outlook. Terms renew on a 30-day rolling basis — which means the maximum downside at any point is a single month of commitment on whatever is currently enrolled. If usage drops below committed levels, the unused portion generates a cash rebate rather than disappearing as wasted spend.

That structure changed the risk calculation entirely. We enrolled our most stable workloads first — core compute and database infrastructure where usage patterns were consistent and predictable — and kept variable and experimental workloads on on-demand. No upfront capital. No multi-year negotiation. No changes to our architecture.

Using Octo’s Cost Groups — a feature that lets you segment cloud spend across any combination of accounts, tags, services, and regions — we monitored our Guaranteed Commitments savings in real time, period by period, without pulling a single manual billing report.

The July–September 2025 figure represents our on-demand baseline — near-zero savings because Guaranteed Commitments coverage had not yet been activated. Everything from October 2025 onward reflects the post-activation picture.

A note on context: this is an internal engineering account, not a large enterprise production environment. The savings rate — not the raw dollar figure — is what translates to other organizations. Across our enrolled workloads, savings have run consistently in the 28–32% range compared to on-demand pricing, which is in line with the RI-equivalent discount Guaranteed Commitments delivers. For organizations running $50K, $100K, or $500K/month in AWS spend, the mechanism works identically — the savings scale with the spend.

The savings didn't arrive all at once. They accelerated.

From Q4 2025 to the April–May 2026 period, our savings grew by a notable percentage — from $391.53 to $1,300.81. This pattern is consistent with how organizations typically approach coverage expansion: start conservative, validate the mechanics, then extend.

Each month, Octo's Cost Group data gave us a real-time read on savings performance — not a lagging quarterly report, but the actual numbers as each period closed. That feedback loop made it straightforward to decide: expand coverage, hold steady, or adjust scope. The 30-day term structure meant each of those decisions carried a 30-day horizon, not a year-long one.

Guaranteed Commitments comes in two forms. Understanding which to use — and when — matters for structuring coverage correctly.

Guaranteed Reserved Instances (GRI) carry a lower risk premium because they apply to specific instance types. This is the right choice for workloads with stable, defined compute profiles: core infrastructure, databases, always-on services where instance type isn't expected to change.

Guaranteed Savings Plans (GSP) apply across a broader range of compute usage regardless of instance type, which offers more flexibility — at a higher risk premium. This fits workloads where instance flexibility matters, or where a team wants to test the commitment model before committing to a more structured GRI scope.

In practice, the most effective approach combines both: GRI anchoring the stable core, GSP covering workloads where flexibility outweighs the premium difference. Both types run on 30-day rolling terms. Both generate cash rebates on unused commitment — nothing is forfeited if usage drops.

Our account is an internal environment, not a large production footprint. The dollar amounts are ours. The mechanism — and the savings rate — is available to any organization running predictable AWS workloads on on-demand pricing.

The question worth asking isn't whether the savings exist. They do. The question is whether the traditional path to access them — a 1-year or 3-year RI purchase, with the forecasting discipline and procurement overhead that requires — actually fits how your organization operates.

For most teams navigating active growth, infrastructure change, or procurement constraints, the answer is: not quite. Guaranteed Commitments through Ripple is a different path to the same destination: RI-equivalent discount rates, on terms that match how modern organizations actually make infrastructure decisions.

No upfront cost. No multi-year lock-in.

Want to see what coverage is possible in your environment before you commit? Talk to an Alphaus partner MSP or our team — we can walk you through a savings estimate in 20 minutes.

Learn more → www.alphaus.cloud/en/ripple