Yannick
Why your factory can’t wait for the cloud, and how to bring cloud innovation to the factory floor anyway.
The CIO of an international manufacturing company faces a difficult choice. Management wants innovation: AI, predictive maintenance, centralized dashboards. IT wants centralized management, standardized infrastructure, and predictable costs. But the factory runs on systems that cannot tolerate a single millisecond of delay, at locations where internet connections are unreliable, managed by teams that have been doing the same thing for years because it works.
Move everything to the public cloud? Too slow, too expensive, too dependent. Keep everything local the traditional way? No centralized management, no standardization, and no path toward innovation. This is the tension where Industry 4.0 breaks down.
Below, we show how manufacturing companies solve this. Not by choosing one or the other, but by combining both: local infrastructure with cloud connectivity. We start with a real-world case, dig into the technical and organizational challenges, and finish with what this means for governance, management, and the first steps.
The challenge: global management with local requirements
One of our clients produces animal feed at locations worldwide. Some of those locations are truly remote: production sites in Africa where internet connections are unreliable and local IT staff is not available. At the same time, the head office in the Netherlands wants centralized control over all operations.
They ran into three challenges:
Management.
With traditional infrastructure, you have to configure, update, and troubleshoot per location. That requires local IT expertise that simply does not exist at remote sites, and it does not scale when you manage dozens of locations.
Innovation.
They wanted to adopt modern approaches: Infrastructure as Code, streamlined application deployment, AI for process monitoring. Their legacy infrastructure was blocking that.
Latency.
A pure public cloud solution was not an option. MES systems need to run locally, close to the machines. If every request has to travel to a cloud region and back, that takes too long, especially at locations with unreliable connections. Milliseconds matter in production processes.
Azure Local is the answer. The platform runs physically in the factory but is connected to Azure Public. This allows the IT team in the Netherlands to manage all locations from a single management environment: monitoring, updates, policy management, and alerting for every site.
The result: far less dependency on local IT teams. Management overhead dropped through automatic updates and integration with the hardware vendor. Systems are prepared by Splitbrain and are plug-and-play on arrival at the remote location. At the same time, MES systems continue to run locally, with the low latency that production processes require.
Centralized control over worldwide operations, without depending on local staff. A single management environment for all locations. And a path toward cloud innovation such as Infrastructure as Code, AI, and predictive maintenance, without disrupting production.
This is not a unique situation. Operational technology, the systems that control machines and production processes, is often conservatively set up. Understandable: OT systems are directly tied to physical processes, and any change is a potential safety risk. So the tendency is to not touch anything. But that approach blocks innovation and makes management unnecessarily complex.
The market numbers confirm this. According to IDC, global spending on edge computing will reach approximately $261 billion in 2025, with expected growth to $380 billion by 2028. Manufacturing & Resources is the second largest sector in those expenditures after Retail, accounting for a quarter of the total. The reason: real-time monitoring of equipment and processes, reducing downtime, and improving operational efficiency.
OT/IT convergence: two worlds colliding
In most organizations, IT and OT live in separate worlds. IT focuses on centralized management, scalability, and standardization. OT focuses on availability, safety, and continuity. These priorities clash, especially when IT wants to manage systems remotely while OT policy dictates that nothing communicates outbound.
A practical example: IT wants to patch and reboot on a fixed schedule. OT wants the production line to keep running. IT wants firewalls and segmentation. OT wants the PLC to communicate with the MES system without delay. IT wants standardization. OT has legacy systems that have been running for twenty years and that nobody dares to touch.
What goes wrong when IT and OT do not work well together is illustrated by a case study from Belden, an American specialist in industrial networking. A manufacturer of architectural aluminum ran 24/7 operations. The entire production process depended on a network managed by IT.
The problem: nearly three times per week, the network went down for unexplained reasons. Each outage cost at least 20 to 30 minutes of downtime per production line. IT pointed to the PLC hardware as the cause, but the controls engineer on the factory side knew the PLCs were fine.
The real problem turned out to be the network architecture. The enterprise switches managed by IT were not designed for the demands of a production environment. Only when OT set up a dedicated industrial network did the outages stop. Six months without a single network issue. Productivity, revenue, and delivery times improved immediately.
OT networks require different hardware and different priorities than IT networks. Convergence does not mean merging everything onto the same stack. It means deliberately separating where needed and integrating where possible.
OT/IT convergence is not just about technology. It is about governance: who owns the OT network? Who decides on changes? Who carries the SLA toward production?
The technical solution, a platform that runs locally but is cloud-connected, solves part of the problem. You get centralized visibility without IT having to manage every switch on the factory floor. But without clear agreements on roles and responsibilities, the tension persists.
Latency in manufacturing: why milliseconds matter
MES systems sit in the chain between the PLCs that control machines and the business systems that manage orders and planning. For direct line control and interlock scenarios, latency must stay as low as possible. Once you get into hundreds of milliseconds or seconds, you risk incorrect decisions, quality issues, and emergency stops.
A MES sizing guide from Sepasoft puts it in concrete terms: at 10ms network latency and 200 queries per second, that latency stacks up to 2 seconds of effective delay in total processing. That is too high for critical interactions. The recommendation is to keep latency substantially below 10ms.
Cross-region or long-distance connections always add latency. The further the cloud region, the greater the delay. For MES systems that need to communicate with machines in real time, that quickly becomes too much.
This is not a theoretical problem. On a conveyor belt moving products at high speed, latency determines whether a quality check can intervene in time. In an extrusion process, latency determines whether pressure and temperature are adjusted quickly enough. Milliseconds make the difference between a good product and scrap.
Azure Local deliberately separates what must run locally from what can run in the cloud. The basic principles:
Local: MES systems, PLC communication, real-time quality checks, everything that is latency-critical. These workloads run on Azure Local hardware in the factory.
Cloud: Centralized dashboards, AI model training, analytics on aggregated data, software updates. These workloads run in Azure.
Management layer: Azure Arc connects the local infrastructure to the cloud. This lets you manage everything from a single location, even though workloads are distributed.
Think of AI for process monitoring: the model is trained in the cloud on large datasets, but inference runs locally. Or predictive maintenance: sensor data is processed locally, only aggregated insights go to the centralized dashboard.
Centralized management without local IT: how it works
Traditional approaches to management are focused on per-location maintenance. That does not scale for organizations operating worldwide. You need local IT expertise that often is not there. You depend on physical presence for updates and troubleshooting. And you have no centralized view of what runs across all those locations.
With Azure Local, you manage all locations from a single dashboard in Azure. You see the status of every node, every VM, every workload. Updates are planned and rolled out centrally. Alerts come into one place.
In practice, this means:
Automatic updates.
Firmware and software are updated without anyone needing to be physically present.
Integration with Infrastructure as Code.
Applications and configurations are managed and deployed as code, not manually per location.
Cloud services to local.
Azure services such as monitoring, policy management, and security run via the cloud but operate on the local infrastructure.
Integration with hardware vendor.
When hardware issues arise, the vendor can directly inspect and ship replacement parts.
What changes: You manage everything from a single dashboard. No more traveling for updates. Fewer fires to fight thanks to proactive monitoring.
What stays the same: You remain responsible for the workloads. You need to create runbooks. You need to set up escalation procedures.
What your team needs: Azure skills. Not specialized local knowledge per site.
Many IT managers fear that new technology means extra workload. With Azure Local, the opposite is true. Your team does not need to build specialized local expertise for every site. You do not need to travel for updates. And you have fewer fires to fight because you see problems before they escalate.
That does require an upfront investment. Your team needs Azure skills. Think training on Azure Arc, Infrastructure as Code, and the specific management tools of Azure Local. Organizations that do not have that knowledge in-house, or do not want to build it, can outsource management through managed services.
The question IT managers often ask: “Can my team run this easily?” The answer is yes, provided the implementation is set up properly. That means: clear runbooks, automated monitoring, and a well-defined escalation procedure.
Governance and risk: how do you demonstrate compliance and accountability?
For the CIO, technology is only part of the story. With major investments, you need to be able to demonstrate risks, compliance, and accountability. Three questions you must be able to answer:
With Azure Local, you have a shared model: the hardware sits in your own factory, the management layer runs via Azure. In practice, this means Azure Local often falls between the IT team and the cloud team. That requires clear agreements. Who is responsible when something goes wrong? Who decides on updates that affect production? Without those agreements, you end up in a gray area where neither team feels ownership.
Data stays local where required. That helps with data sovereignty and sector-specific regulations. Data on the local nodes is encrypted with BitLocker, both at rest and in transit. Azure provides logging and audit trails for the management layer. And with Azure Policy, you can enforce that configurations meet your own compliance requirements. The responsibility for compliance remains with the organization, but the platform gives you the instruments to make it demonstrable.
Legacy infrastructure does not become more secure over time. Without centralized visibility, you miss problems until they escalate. Without modern management tools, you depend on local staff that is increasingly hard to find. And without a path toward innovation, you fall behind competitors that are investing.
Conclusion: what now?
The smart factory market is expected to grow at 9 to 10% per year, reaching a size of more than $300 billion by the end of this decade. Industrial AI, real-time process monitoring, and predictive maintenance are no longer experiments but are becoming standard in manufacturing. That growth does not happen by itself: it requires infrastructure that bridges the gap between cloud and factory floor.
The risk of waiting is concrete. Hardware ages. VMware licenses get more expensive. Local IT staff becomes scarcer. And you increasingly need to demonstrate that innovation and compliance go hand in hand.
The first step? Map out which workloads are latency-critical and which are not. That determines what must run locally and what can move to the cloud. From there, you can design an architecture that combines both.
Are you working on a hardware refresh or a migration away from VMware? Then this is the moment to not just replace hardware, but to modernize the architecture as well. Get in touch to discuss the options.
1) Microsoft. Industry 4.0 technologies. microsoft.com/en-us/ai/manufacturing/resources/industry-4-0-technologies
2) IDC Worldwide Edge Computing Spending Guide, March 2025. Via Computer Weekly: computerweekly.com/news/366620991/IDC-global-edge-computing-spending-to-approach-380bn-by-2028
3) Belden case study, December 2023. Separating IT and OT Networks Helps Manufacturer Boost Productivity. belden.com
4) Sepasoft MES Server Sizing and Architecture Guide V6. docs.sepasoft.com (PDF)
5) Fortune Business Insights, Grand View Research, Mordor Intelligence: Smart Factory Market analyses 2024-2032. fortunebusinessinsights.com
Gerelateerde blogs
