Energy Efficient Computing

In October 2024, Gartner released a report on the top ten technology trends for 2025. At number 6 on the list was Energy Efficient Computing.

The environmental impact of so many industries is under scrutiny, so it’s only right that the IT sector should play its part. It is projected that in 2025 alone, data centres will consume 20% of global electricity and emit 5.5% of the world’s carbon emissions. This excludes the use of IT systems in virtually every business around the world. The fact that industry influencer Gartner has flagged this as a concern indicates that we could (even should) be at a turning point with the industry pushing sustainability up the business agenda.

We’re not just ticking a box on the green agenda here. There are financial savings to be had too. By streamlining processes and systems, you use fewer resources and hence reduce your outgoings.

What is Energy Efficient Computing?

The Gartner report defines Energy Efficient Computing as technology that encompasses more efficient architecture, code and algorithms, hardware optimised for energy efficiency and the use of renewable energy in IT systems.

This is a broad definition and indicates the scale of the problem. It’s not something that is going to be solved overnight, and the biggest impact will be made by the largest enterprises making radical changes in their way of working.

However, it’s not just the responsibility of these tech giants to do all the work. Businesses large and small can make a positive impact.

Time to throw in a couple of concepts:

1. The aggregation of marginal gains;
2. The whole is greater than the sum of the parts.

The aggregation of marginal gains

Simply put, improving efficiencies of processes, systems and activities by 1% at a time can lead to a significant cumulative effect over time. This 1% is achieved by making small, continuous improvements, rather than seeking radical changes. This does require close attention to detail and a meticulous approach to identifying areas where even seemingly insignificant changes can make a difference. It also requires ongoing monitoring and analysis to ensure that improvements are sustained.

The whole is greater than the sum of the parts

Making a 1% improvement in the efficiency of one system can have a knock-on effect for other parts of an ecosystem. This effect could be positive, but it could be negative, hence the need to be selective about the changes that are made. Where the knock-on effects are positive, the initial 1% improvement has wider efficiency benefits. Taking all the 1% improvements together (plus ongoing improvements), the overall impact is much greater than it initially appears.

Where can these efficiencies be found?

I attended an event the other day, run by Yorkshire Azure User Group and Yorkshire DevOps. Of particular interest to me was a presentation by Simon Thurman, Senior Technical Specialist at Microsoft and member of the Green Software Foundation Developer Relations Committee.

Simon mentioned a handful of areas where positive changes could be made which aligned with the Gartner definition of Energy Efficient Computing. Three that struck a chord were:

1. Efficiency of code
2. Use of efficient energy sources
3. Hardware efficiencies

Code efficiency

Simon reflected that as a junior developer 30 years ago, if he’d written code in an inefficient manner back then that is still running today, then that’s 30 years of inefficient running. Even if this uses a small amount additional power, over an extended period, this all adds up to a significant quantity.

Over time, developers have become more experienced, languages have evolved, and the advent of AI is set to have an impact on the world of DevOps.

Code Assistants and Agents (available on open source platforms, OpenAI, Google, Anthropic, and many others) are transforming the way software is developed—dramatically improving speed, quality, and consistency. These intelligent tools support developers by automating repetitive tasks, flagging bugs early, and generating context-aware suggestions, effectively serving as tireless coding companions. But their value goes beyond productivity. By promoting best practices and eliminating inefficient patterns, they help developers write cleaner, leaner code—code that runs faster, consumes fewer resources, and scales more gracefully.

This, in turn, supports a more sustainable approach to software engineering. Efficient code requires less processing power, reducing strain on infrastructure and minimising energy consumption across development pipelines and live environments. As organisations begin to take digital sustainability seriously, code assistants are becoming silent allies in the fight against carbon-heavy computing. By weaving intelligence into the development process, these agents help engineers not just build faster, but build smarter—and greener.

Efficient Energy Sources

This was a bit of a lightbulb moment for me. For years we’ve been aware that cloud computing is independent of location. You can place your data pretty much anywhere in the world. Yes, there are issues that might influence this decision—data latency, security, support—but there are instances where businesses can benefit from energy efficiencies.

Simon shared an interesting website (Electricity Maps), which shows real-time, historical and forecast data on the global access to electricity mix and carbon intensity. This means that you can do a bit of cross-checking on where your cloud data centre is located and see whether there is an alternative hosted in a location with more sustainable energy sources.

For example, moving a data centre from the UK to France (where there is a greater proportion of nuclear energy), the carbon intensity of the resources drops from around 245g CO2eq/kWh to just 18g CO2eq/kWh. Or look for a country that makes greater use of wind, tidal or solar power.

As the Electricity Maps screenshot below shows, there is significant differences in the carbon intensity of electricity generation across the globe.

Moving to a cloud datacentre to a more sustainable country might make a marked difference to your carbon footprint. Of course, this might not be appropriate for time sensitive data, but where latency isn’t an issue, then possibilities exist.

Hardware Efficiencies

This is an indication that that many organisations currently underutilise their data centre infrastructure, with server utilisation rates often below 50%, suggesting significant inefficiencies that need addressing. Diligent use of performance analytics tools and automation platforms can identify these inefficiencies so that businesses can mitigate these issues, increases in these utilisation rates, defer the need for additional hardware purchases and reducing energy consumption. It’s not just a question of energy saving, but removing the need for additional hardware makes financial sense too.

Scheduling Your Computing

In addition to looking at the sustainability of energy sources, businesses can also look at energy costs. Electricity prices tend to drop during off-peak hours—typically at night— due to lower demand on national grids. While this won’t help with real-time data processing essential to the functioning of a business, shifting batch processing workloads to night-time hours presents a strategic opportunity for companies looking to cut operational costs and improve sustainability metrics.

By aligning non-time-critical tasks such as data warehousing, analytics, and financial reconciliations to these hours, businesses can unlock substantial cost savings.

For example, a financial services firm running large-scale reconciliations can reduce energy costs by scheduling these tasks post-business hours. Similarly, an e-commerce company could process overnight transaction data or reindex product catalogues during low-tariff periods, freeing up daytime capacity for customer-facing operations.

Geographic Considerations

The cost advantage varies globally. Regions with dynamic pricing models—like the UK, parts of Europe, and certain U.S. states (e.g., California, Texas)—stand to benefit the most. Additionally, countries with significant investments in smart grids and time-of-use pricing make it easier to automate and optimise these transitions.

Sustainability Impact

From a sustainability perspective, this practice aligns well with greener energy use. At night, grids are often powered more by renewable sources—particularly wind, which tends to generate more power during off-peak hours. By consuming electricity when renewables dominate the mix, businesses can indirectly reduce their carbon footprint and support the wider adoption of clean energy.

Counting the cost

By way of example, I used the Microsoft Azure pricing calculator to do some simple cost comparisons between the deployment of virtual machines in different locations around the world. In this example, I chose a B32as v2 VM with 32 vCPUs, 128 GB RAM and 0 GB temporary storage, paid for on a pay-as-you-go arrangement. Within these, I looked at the costs for VMs optimised for different processes.

Alongside this I noted the Carbon Intensity for each region both at midday and midnight on 5 May 2025. The value of carbon intensity fluctuates depending on the amount of green energy available to the grid, so these figures shouldn’t be taken a guaranteed.

I also used the pricing calculator to look at the cost differences for Data Lake Storage, in this case 100TB General Purpose storage for hot, cool and archive access tiers.

These tables highlight the options available to businesses to investigate alternatives in their cloud computing services and the possibility to either save money or to have a positive impact on the environment.

Where can DeeperThanBlue Help?

As an expert in cloud migration and hosting, and partners of the major hyperscalers, we are well placed to identify where you might find efficiencies in your cloud storage solutions.

We have experience in implementing serverless environments on Google Cloud and Microsoft Azure, whereby a cloud server remains dormant until a request is made, after which it becomes active for the period of utilisation. This way the customer is only paying for the compute power that is used.

We have also implemented a Kubernetes containerised environment for another client in which clusters remain active all the time but underlying nodes are shut down over night and at weekends when they are not needed.

Summary

With Gartner highlighting the energy efficiency as a trend this year, it seems unlikely that the issue will go away anytime soon. Indeed, it is probably that it will move higher up the priority list.

Business owners, shareholders and decision makers will be taking a closer look at ways in which they can tackle the problem in a way that is both manageable and effective.

The trick will be to find those parts of the business where efficiencies can be made, through close attention to detail and a meticulous deep dive into processes and systems.

But once you start finding suitable cases you can begin making the small changes, and gradually notice a larger overall effect for the business.

• Share