How To Drive Sustainable IT: Turn Laptops Into Infrastructure
Discover how software-managed resilience turns old laptops into enterprise-grade infrastructure – with real ROI and sustainability gains.
We turned 100 retired laptops into a cluster that saves over $136k a year, and kept 210 kg of electronics out of landfill.
It started as a pragmatic experiment: what if the laptops gathering dust in storage could do more than sit in recycling limbo? We stacked them in a custom rack configuration, wired them up, and orchestrated them with Kubernetes.
The result was a functioning, self-healing compute platform that’s now part of our development environment.
Key takeaways:
- A third path: Orchestration now makes reused hardware a viable alternative to buying servers or renting cloud.
- Software-managed resilience: Reliability can be delivered through orchestration and monitoring, not expensive hardware.
- Real ROI: The 100-laptop cluster saves $136k per year versus a 1-year AWS Savings Plan.
- Proven sustainability: Extending device life by 2.5 years avoided ~210 kg of e-waste while deepening in-house infrastructure skills.
“It started with an unpromising stack of forgotten laptops. But instead of seeing e-waste, we saw untapped potential. Now, that 'e-waste' is the core of our Development Cluster.” – Jason Rolles, Managing Director, BlueOptima
If you’re balancing performance, cost, and sustainability targets, read on to discover the lessons we learned from this project that could help you, too.
The (Expensive) False Binary
Most organizations assume they have just two options for compute: buy new servers or rent capacity in the cloud. Both feel safe, but both carry hidden drag, like capital lock-in, rising total cost of ownership (TCO), and an expanding hardware footprint.
This project proved a third path. When you treat resilience as something managed by software, not built into hardware, yesterday’s laptops become today’s infrastructure.
Kubernetes handles hard failures automatically; monitoring and ops catch the subtle degradations before they spread. Together, they turn unused assets into reliable compute.
Why Sustainable IT Matters Now
Cloud costs are under sharper scrutiny than ever. AWS Savings Plans typically offer around 34% off for one-year, no-upfront commitments, and up to 72% for using specific instance families in a region. Any on-premise model has to clear that bar.
At the same time, electronic waste is spirallng. The UN reported 62 million tonnes generated in 2022 (an increase of 82% since 2010), with only 22.3 percent properly recycled. By 2030, that number is expected to reach 82 million. Recycling won’t be enough to tackle this problem.
Extending hardware lifecycles is the practical route to measurable impact.
For enterprises facing sustainability mandates and keen to support the UN’s Sustainable Development Goal (SDG) 12: Responsible Consumption and Production, reuse is one of the few levers that delivers both environmental and financial returns.
"Before you budget for a new server rack, look at the hardware you're about to recycle. We found our entire Kubernetes development infrastructure in that pile." – Salman Kagzi, CTO, BlueOptima
What We Did
- Repurposed 100 retired laptops, applying circular-economy principles to extend their life and reduce e-waste.
- Reimaged, labeled, and networked each device into a shared Kubernetes environment with automated provisioning.
- Built a custom rack setup to optimise airflow and access, and deployed monitoring for temperature and disk health.
- Enabled self-healing operations: Kubernetes reschedules failed nodes automatically; our telemetry catches slower “grey failures” before users notice.
- Accounted for maintenance equating to roughly 0.25 FTE of engineering time to keep the cluster running smoothly and predictably.
The Full Cost Picture
We modelled everything: power, cooling, space, networking, and labour. We benchmarked against the efficiency of hyperscalers to keep assumptions honest.
Compared to a one-year AWS Savings Plan baseline cost of $206,736, the reused cluster came out at $69,811.
That’s a saving of $136,925 per year.
Not every organization will see the same savings due to local energy prices and facility efficiency, but the principle holds: software-managed resilience can shift the economics of compute.
You can find a detailed breakdown of the sustainable IT project stats in our whitepaper.
The Environmental Dividend
Extending those laptops’ lives by roughly 2.5 years prevented more than 210 kg of complex electronics from becoming e-waste, and avoided the upstream emissions of manufacturing new machines. That’s a tangible contribution to SDG 12.
Lifecycle extension isn’t a pledge; it’s a measurable, auditable action that any CIO can report on with confidence.
What Your Team Gains
Running this cluster deepened capability within our team. Managing bare-metal Kubernetes forced our engineers to understand the hardware realities that managed services obscure.
They now spot degradation earlier, design for failure, and think more holistically about performance.
It also diversifies our infrastructure strategy. Cloud remains essential for elasticity and scale, but this “third path” adds resilience and sovereignty without duplicating spend.
The Pragmatic Limits
This approach fits steady, fault-tolerant workloads – CI/CD runners, internal services, analytics jobs – not latency-critical production traffic. It requires proactive monitoring, disciplined maintenance, and a willingness to accept that some hardware will fail.
But we’ve found that when resilience is handled in software, those failures become routine events, not outages.
The Old ‘Buy or Rent’ Equation Doesn’t Apply Any More
When orchestration and monitoring work together, resilience becomes a software-managed property, and reuse becomes a strategic advantage.
We ran it. We measured it. It works.
Download the full paper for details on our assumptions, ROI calculation, and operational lowdown. It’s a transparent blueprint for any enterprise exploring sustainable IT and cost-efficient infrastructure at scale.
Read more here.
