How to reduce cloud waste: It boils down to 3 steps

It’s almost Newtonian; for every statistic about the growing amount of money spent on cloud computing resources, there is an equal and opposite statistic about how much is wasted. For example, Gartner’s latest forecast has worldwide end-user spending on public cloud services catapulting to $723.4 billion in 2025 from $595.7 billion in 2024, a 21.5% increase. Meanwhile, in a recent survey of 300 companies, 78% of respondents estimated that between 21% and 50% of their cloud expenditures are wasted annually. And so forth and so on.

No matter how it’s quantified, the aggregate of budgets wasted on unnecessary cloud resources is big — sometimes shockingly big within a single organization. This blog explores three steps to rein in laissez-faire cloud resource management (sometimes jokingly referred to as “ClickOps”) to achieve inherently leaner spending that’s more tightly integrated with the business (a “FinOps” approach). To illustrate one of these steps, I’ll share how a large insurer cut its $100 million annual budget to run a flagship analytics model on-premises down to just $800 per day in the cloud.

»Disclaimer: Extra capacity isn’t always waste

What does “cloud waste” really mean?

With some applications, a certain amount of cloud overhead is beneficial. In these cases, overprovisioning by 10% to 20% can alleviate IT teams from constantly resizing apps or scrambling to deal with ephemeral workloads. Given the tradeoff between marginal costs and potential mayhem, retaining some application-specific excess capacity is often better.

The more significant issue with many applications is that a simple “lift and shift” can cost dearly when migrated from on-premises resources to the cloud. In the planning process for moving any given app, the question “What do I want by moving this to the cloud?” should come first to ensure the architecture pattern meets or exceeds cost expectations. If migrations aren’t first considered in those terms, the benefit of moving to the cloud is often lost. Which brings us to …

»Step 1: Rethink apps from CapEx to OpEx

Migrating large, typically legacy apps as-is to the cloud incurs significant waste. Legacy apps become monolithic when years or decades of incremental functionality are bolted onto outdated architectures. Moving them directly to the cloud can cost more than keeping them on-prem, a problem that can only be solved with a strategic re-think.

The key to modernizing legacy workloads is to take advantage of the “rental model” of the cloud. This requires modifying an app’s architecture from an on-prem fixed cost (CapEx) to the cloud’s flexible cost structure (OpEx).

»A real-world example

A large insurance company had its “crown jewels” — a risk and pricing analytics system — running on-prem. It required over 100 servers, a large data lake farm, and about 150 engineers to constantly manage networking, data management, extract-transfer-load (ETL) processes, and more. The app was costly to maintain but couldn’t be retired; its capabilities are at the heart of the insurer’s business.

If the company had just lifted and shifted the app to the cloud, its inefficient architecture might have cost up to twice as much to operate, and it would still have to be maintained by dozens of engineers. Instead, the insurer modernized the app, re-architecting it for the cloud. The risk and pricing analytics app now lives in inactive cold storage in the cloud, which means it gets woken up when needed and dynamically hydrated with data in a few minutes. The analytics can now be run much faster, and afterward, the data instance is flushed, and the app goes back into glacier storage, incurring lower costs.

The benefits are immense: the app in the cloud is now managed by three to five people, and analytics run daily instead of every few months. Sensitive personally identifiable information (PII) is in production for only a few minutes; otherwise, it rests in secure cloud storage.

The cost-benefit of rearchitecting the app from CapEx to OpEx is equally immense. The risk and pricing analytics system costs almost $100 million a year to run on-premises.

In the cloud, it’s less than $800 per day.

$800 × 365 days = $292,000 per year (to be safe, we’ll round up to $300,000 to account for odds and ends!)

$100,000,000 - $300,000 = $99,700,000+ savings per year

»Step 2: Put guardrails in place

Under- or unused cloud resources can quickly add up. Zombie processes, orphaned resources, and abandoned accounts are readily identifiable as waste and can be remedied continuously and efficiently with automated solutions.

Although many developers and operations engineers have trouble admitting it out loud, they are often new to the cloud. They will struggle with cloud migration if they’ve been operating on-prem environments for much or all of their careers. If they haven’t yet learned how to optimize for the cloud, they may introduce more costs than they save as they try to rearchitect applications.

To make the transition easier, an experienced platform team putting guardrails in place is a huge cost-saver. They can set up solutions behind the scenes that automate and optimize cloud resource management, using infrastructure as code (IaC) to replace ClickOps with consistent corporate policies. Applications can be run through an image lifecycle management system and automatically pack resources efficiently into cloud instances based on pre-defined policies, optimizing cloud resource consumption without user intervention. As teams transition to the cloud, guardrails can help them avoid classic mistakes and build confidence with each new success.

»Step 3: Tightly couple cloud spending with business results

In many companies or large lines of business (LOBs), IT is a single, centralized cost center controlling all technology spend, including cloud. While these IT organizations may support a specific business, the budget is not tightly correlated with, or factored into, the business unit’s profit and loss (P&L) statement. If overall IT spending is on track, individual expenditures aren’t closely scrutinized — why slow down innovation?

Cloud FinOps brings technology, finance, and business together to influence operational processes based on the financial impact they will have. It’s an active state, not reflective, and naturally forces creativity when assessing the tradeoffs between cloud spend and the resiliency it can drive against revenue targets and margins.

For example, if an IT organization triples the cloud resources supporting an app to make sure it’s rock solid globally but doing so reduces operating margins by 30%, the tradeoff is considerable. Is it worth it to the business? Can the resilience be achieved in another, cheaper way? FinOps tools like Apptio and Infracost are part of a growing number of solutions in this space to help drive down infrastructure costs without sacrificing revenue targets or risk tolerances.

»Learn more

Is your organization’s cloud consumption growing? Is cloud waste, too? Learn more about the benefits of IaC from the HashiCorp webinar: Deploying policy-guarded infrastructure – Creating and enforcing guardrails.