Photo by Anders J on Unsplash

Can you have your cake and eat it too? Sustainability in Cloud vs Cost Optimisation

Part 2: And eat it too?

Picking one or the other

Welcome back! Thanks for joining us for part 2 of our series on cloud sustainability and cost optimisation, or as I like to say, Can you have your cake and eat it too?

In part 1, we talked about the how and why behind consumption and cost, now onto the what. What does this look like in practice and what are the various mechanisms that we can put in place to address consumption. The practices that follow are just the tip of the iceberg when it comes to reducing consumption, what I want to illustrate are a few examples where there may or may not be significant cross over between our two desired outcomes, lower emissions, or lower cost.

Recap

First, a quick recap for readers of part 1¹ or for those who are just joining us. The often bleak reality of sustainability is that without market incentives driving it’s adoption, when stacked up against other organisational priorities it can become less of a focal area — in my experience at least.

The question I want to pose and hopefully answer within this 2-part series, is why do we need to pick just the one priority? What if we instead look at the parallels between sustainability and cost, and identify opportunities to pursue both. Or, more succinctly put, can you have your cake and eat it too?

Rightsizing Compute Instances

Let’s start off nice and simple, rightsizing at it’s core is about aligning our cloud infrastructure with our actual consumption, this way we only pay for the resources we actually use. One of the core principles of FinOps is data-driven decision making, this means putting in place the mechanisms that allow us to tag and monitor resources, identifying key resource metrics, and placing accountability on the relevant business owner.

From a compute perspective this could mean utilising CSP or 3rd party monitoring tools to identify under or over provisioned VMs, typically looking at metrics like CPU usage, memory consumption, and network traffic over a set period. Analysing this data can then help identify instances where VMs are consistently using more resources than needed or struggling to meet demand due to insufficient resources. Instance types vary to suit different workloads, understanding these workloads and proactively monitoring our utilisation provides the opportunity to rightsize and save on cloud costs.

How does rightsizing weigh up against our desired outcomes? Good news! This is a perfect opportunity to lower consumption and cost through the one intervention. This highlights the core principle of cloud sustainability, better optimised utilisation leads to a smaller environment footprint, with the added benefit of reduced cost.

Choice of Region

In my previous article I outlined why choice of cloud region can have some impactful design considerations for when we look at emissions. In a nutshell, this is down to the respective locations, local energy mix, and access to Carbon Free Energy. Energy mix each location is determined by a range of complex factors:

• Geography and local environment can impact what energy production options are available, i.e. solar, wind, or hydropower.
• Infrastructure in place to take advantage of the local conditions. i.e. solar or wind farms.
• Local regulations that incentivise or subsidise the development and usage of renewable energy.

Transparency becomes key in understanding how these factors add up and impact the energy options available. The Carbon Free Energy for Google Cloud report² is a great example a CSP providing this level of insight, allowing for actionable data when determining where resources are hosted and identifying those with the lowest environmental impact.

But what about the cost perspective? Costs per region do vary as they’re operating within the market conditions local to that region. Likewise, there’s a mix of factors that determine the cost of hosting resources in a particular region, just to name a few:

• Costs related to building and operating data centres, including land, construction, power, and cooling, vary significantly by location. Regions with lower infrastructure costs often have lower cloud pricing.
• Energy costs are a major operating expense for data centres. Regions with cheaper and more sustainable energy sources tend to have more competitively priced cloud services.
• Government policies, tax rates, and regulatory requirements related to data storage and privacy can impact cloud pricing in different regions.
• Regions with robust and reliable network infrastructure tend to have lower latency and data transfer costs, which can influence overall cloud pricing.

So where does this leave us? Location has significant factors that can influence our two outcomes, not to mention the other considerations of available services, data privacy or security. Higher access to sustainable energy sources may be a factor that reduces cost but other factors will also enter into consideration. As for our headline question, does choice of region let us achieve both desired outcomes? Yes, but not always. This is a decision that might sometimes become a judgement call between our priorities.

Scheduling Resources

This is a challenge that I’ve recently had to tackle myself, environment costs across 3 tenants: Development, Testing, and Production, are each confined by an already set operational budget. Once the budget is gone, it’s gone. This presents the fun task of determining which resources need to be deployed and which can be scaled back. This again goes back to the principle of data-driven decision making and accountability of cloud usage. Starting with Dev and Test environments, odds are that these environments do not need to be in operation full time, 7 days a week.

Automated scheduling gives us the ability to identify when and where resources need to be deployed and allows us to shut them down outside of this window. If environments go from operating full time to just business hours, this represents a 60% decrease in utilisation. This also puts the onus on development teams to adopt better financial accountability for the resources that they’re responsible for, another tick in the box for the cultural shift that FinOps enables.

As we’ve addressed throughout this article, lower utilisation = lower emissions. The reduction in utilisation we can see from proper resource management and scheduling also means that this is a key driver of lowering cost. This means that scheduling is a win/win for achieving our outcomes!

Commitment Based Purchasing

Commitment based purchasing is exactly what it sounds like, by committing to long term compute requirements, users can save up to 70% on cloud costs when compared to On Demand pricing. They come in plenty of flavours across each CSP, for example, AWS Reserved Instances³ or GCP Committed Use Discounts⁴. Ideal for steady, long-running workloads, purchasing plans can achieve massive cost reductions for where we know compute resources are required, typically over a 1–3 year period.

Here’s where it gets complicated, making full use of commitment-based purchasing means that we first require detailed visibility into our workload utilisation running in its ideal state. For mature organisations this may not be as big as a challenge, done incorrectly though it can mean either unused capacity that’s been committed to upfront, or additional On Demand spend for resources beyond the committed threshold.

Once again, we go back to the FinOps Framework, when implemented correctly it means organisations should have mechanisms in place to get timely access to data and the collaboration between engineering, finance, and business teams to properly plan for their future consumption. In this case, when applied correctly, commitment-based purchasing represents a huge opportunity for cost reduction. From a sustainability perspective it’s a little trickier. The risk of over-consumption is still present, even if it is heavily discounted. In my opinion, I’d say that commitment-based purchasing edges out sustainability and more so favours cost optimisation.

Conclusion

Let’s go back to our original question, Can you have your cake and eat it too? In the case of the examples I outlined, the answer is mostly yes! Much like real cake, there’s a degree of moderation to be applied on these use cases. Sometimes we can achieve both outcomes, in other situations we may need a more critical lens applied. When we consider mechanisms that aim to reduce and optimise utilisation, then sustainability and cost become directly linked, enabling us to reduce spend while also creating a smaller environmental impact.

The key take away from this is that while these parallels exist to be leveraged, they come with the caveat that understanding your workloads and requirements play a crucial role in properly implementing some of these best practices. When applied correctly, developed, and matured, FinOps represents a framework that maximises the business value of cloud and supports more sustainable practices.

Thank you for reading our 2 part series on cloud sustainability and cost optimisation! If you’re looking to better consider Cloud and IT within your Net Zero strategy or are looking to implement FinOps, please reach out for a more detailed conversation on how Deloitte Engineering can support.

References

[1] White, C. (2025). Can you have your cake and eat it too? Sustainability in Cloud vs Cost Optimisation. [online] Medium. Available at: https://medium.com/deloitte-uk-cloud-blog/can-you-have-your-cake-and-eat-it-too-sustainability-in-cloud-vs-cost-optimisation-eac6d1e976f0

[2] Google Cloud. (2025). Carbon free energy for Google Cloud regions. [online] Available at: https://cloud.google.com/sustainability/region-carbon.

[3] AWS (2024). Amazon EC2 Reserved Instances. [online] Amazon Web Services, Inc. Available at: https://aws.amazon.com/ec2/pricing/reserved-instances/.

[4] Google Cloud. (2025). Committed use discounts | Documentation. [online] Available at: https://cloud.google.com/docs/cuds.

Note: This article speaks only to my personal views/experiences, is not published on behalf of Deloitte LLP and associated firms and does not constitute professional or legal advice. All product names, logos, and brands are the property of their respective owners. All company, product and service names used in this website are for identification purposes only. Use of these names, logos, and brands does not imply endorsement.