We’ve been talking about the edge in terms of compute and workloads for years. However, there are emerging changes in where the intelligence needs to sit and how edge devices need to be managed, and those changes are starting to reshape the role of the edge.
For the past decade, most organisations have pulled everything back into the data centre or the cloud. Consolidation was the goal, with centralised control, management and cost savings, but AI is now reversing part of that trend. When you want to run AI inferencing workloads on a live camera feed or a sensor stream, the workload must be close to where the data is generated because the value lies in low response times. Sending that data back to a central data centre for processing means the moment has passed.
In this blog, I want to explore what is driving infrastructure back to the edge and why managing distributed environments at scale has become the real challenge. I’ll also look at how a more standardised approach can help organisations run AI and other modern workloads across large numbers of sites.
Think about a supermarket running a vision system at the self-checkout, or a station analysing security footage to flag an incident as it happens. Those scenarios only work when they run locally at the site in real time. That is what is now pulling infrastructure back out to the edge.
For many use cases, the edge makes sense, but pushing infrastructure to hundreds of sites becomes a complex operational problem, which is why many organisations avoid it.
Working with utility companies that live with this situation every day, they have edge infrastructure across many sites, and almost every site is different. There is no consistent build, so every time someone goes out to a site, they have to manage it as a separate configuration. Multiply that across a fleet and you have a support and skills burden that grows with every site you add. That is the real ceiling on distributed edge, and it has very little to do with the compute itself.
The shift lies in how this infrastructure should now be built and managed. Cisco are addressing this with their Unified Edge solution, bringing compute, networking, storage and security into a single modular platform within a chassis. Traditionally, each of those was a separate component to rack, cable and configure. Consolidating them removes much of the physical setup, which is a valuable, if short-lived, win.
The bigger change is in management. The unified edge chassis gives you up to five slots to fill with whatever the site needs, whether that’s compute, a GPU for AI processing, networking or storage. You tailor the box to the job rather than shipping the same oversized kit everywhere, then manage the entire fleet from a single plane using blueprints and templates so that, for example, a new warehouse deploys a consistently build every time they are built. You build a warehouse blueprint once, then change only the variables for warehouse one, two or three, and everything stays in sync across the fleet.
Deployment also gets simpler because our field techs can provision a unit from a mobile phone, so standing up or replacing a site no longer requires a specialist on the ground. Going back to that energy company, this is the difference between every site being a one-off and every site being a known, repeatable build.
That flexibility runs deeper than the hardware. Because you can also run the hypervisor and software of your choice across those slots, the platform gives you room to modernise your virtualisation strategy on your own terms.
One pattern we are seeing is that changes in virtualisation licensing models can force smaller edge locations into a minimum footprint designed for the data centre. Paying for data-centre-scale licensing at a site running a fraction of the workload is hard to justify.
A smaller, modular infrastructure footprint gives you the flexibility to right-size your edge environment, run the platform to suit your needs and evolve your virtualisation strategy over time. You get the infrastructure that suits you without being locked into a model built for the data centre.
None of this is confined to a single sector. Any organisation with distributed sites and a need to process data locally could benefit from a more flexible edge infrastructure approach. That includes retail, manufacturing, mining, healthcare and in parts of government like transport.
Operational technology (OT) is a good example too. OT systems have often been walled off from IT entirely, kept very local by design, with no real connection back to the data centre. A consolidated, well-managed edge platform gives you a way to bring structure to those environments without losing what makes them work. The point is less about a specific use case and more about what becomes possible once distributed infrastructure at scale is genuinely manageable.
For most organisations, running AI or modern workloads at the edge is already on the agenda. The harder question is whether you can operate that edge across your whole fleet without it turning into sprawl you cannot support.
If this is on your radar, check out this recently published IDC Whitepaper Simplifying Secure, AIReady Infrastructure for the Distributed Enterprise. As a Cisco Networking Preferred Partner and Cisco ANZ Partner of the Year (2025), we are also running one-on-one workshops, tailored to your industry and your use cases, to map this against your own environment. Contact us via the form below to start the conversation.
Contact our team today.
Information provided within this form will be handled in accordance with our privacy statement.