Reserved Instances automation: A strong promise, with a more nuanced reality

AWS cost optimization has become a top priority. According to the 2025 State of FinOps report, reducing waste is now the leading priority for FinOps practitioners, with roughly 30% of cloud spend estimated to be wasted on oversized or unused resources.

In this context, platforms such as nOps, DoIt and Pump have gained traction with a compelling promise: fully automated management of Reserved Instances (RIs) and Savings Plans, with no effort required from your team.

On the surface, the model is attractive. You connect your AWS account, the algorithm buys and sells commitments on your behalf, and you keep a portion of the savings generated. In practice, more and more companies are coming away from this experience frustrated, often left with commitments that no longer match their operational reality.

Here is why.

Blind automation: When the algorithm optimizes for itself

Pump.co, for example, operates on a model where the platform charges nothing upfront and takes 20% of the savings it generates on your AWS bill. At first glance, it sounds like a win-win. But that is exactly where the structural problem begins.

The issue is twofold: the commitments purchased on your behalf are not necessarily the right fit for your workloads, and 20% of your savings goes back to the platform. For a team that absolutely refuses to spend a few minutes validating its commitments, this may still be better than doing nothing. But in real terms, you are leaving a meaningful share of savings on the table, and doing so repeatedly, commitment after commitment.

With some of our clients who were disappointed by the results of Pump.co or nOps, the additional savings we were able to generate through Stable, by structuring commitments around their actual workloads, represented more than 18% in extra savings, kept directly by the client, without handing over the 20% cut taken by other platforms.

The real problem: You never capture 100% of your savings potential

The promise of full automation hides an uncomfortable financial reality. When a provider takes a share of every dollar saved, it has a structural incentive to maximize the volume of commitments purchased on your behalf, not necessarily to maximize your actual return.

Fully automated platforms tend to choose commitments that are either conservative, to limit their own risk, or aggressive, to maximize short-term savings and therefore their own revenue. In both cases, you do not get the optimal strategy for your specific consumption profile.

More importantly, you never had the chance to validate the choices, understand the reasoning behind them or factor in your internal knowledge of where your infrastructure is headed.

What companies are really looking for: Control, visibility and a strategy that fits

Companies leaving these platforms are not looking for less expertise. They are looking for the right expertise, at the right time, with the ability to understand and validate the decisions that will financially commit them for 1 to 3 years.

RIs are non-refundable and difficult to modify or exchange once purchased. That makes it essential to carefully assess your usage patterns and make informed decisions before committing. This kind of informed decision-making is exactly what blind automation cannot provide.

At Unicorne, we see a recurring need for three specific things.

1. A contextualized workload analysis. Not only historical billing data, but also your roadmap, migration plans, seasonal cycles and the actual stability of each service.
2. A clear view of the options available. Before making any commitment, you should be able to see which type of commitment fits which workload, the expected ROI for each scenario, and the risks if your infrastructure changes.
3. Human validation before every purchase. The goal is not to eliminate your involvement. It is to help you make an informed decision in 30 minutes instead of 3 weeks. You remain in control of your choices, and the full savings potential stays with you.

The right commitment strategy: neither fully automated nor fully manual

At Unicorne, our approach is based on a simple belief: AWS savings do not come from buying RIs alone. They come from buying the right commitments for the right operational reality.

In practical terms, a well-structured commitment strategy generally allows companies to achieve stronger savings on stable workloads while preserving the flexibility needed for environments that continue to evolve.

The difference compared with automated platforms is straightforward: you capture 100% of those savings, without redistributing a fixed share to an intermediary.

Regain control of your AWS commitments

The frustration with automated optimization platforms is understandable. It points to a deeper need these tools do not fully address: the ability to understand, validate and control the financial decisions that commit your company for several years.

At Unicorne, we do not push automatic commitments. We build an engagement strategy with you, based on your actual workloads, with clear visibility into your options, human validation before every purchase, and support that helps you capture your full savings potential.

If your current Reserved Instances no longer reflect your real infrastructure, or if you have never had clear visibility into how those decisions were made, now is a good time to speak with our FinOps experts.