Live data flowing in 7 days. No IT department required. See exactly what it looks like for your machines.
Calculate Your Downtime CostPractical guides, benchmarks, and shop floor strategies — no fluff, no spam, unsubscribe anytime.
Lights-out machining adds 50-75% capacity at near-zero labor cost. But you cannot run unattended without monitoring. Here is what to monitor, how to build confidence in stages, and the ROI math.
The CFO conversation: what machine monitoring actually saves, a worked ROI example for a 5-machine facility, payback period calculation, and a framework for presenting to ownership.
Point-to-point integrations do not scale. The Unified Namespace replaces spaghetti with a single event-driven hub — MQTT, Sparkplug B, and ISA-95 hierarchy. How it works, how it compares to historians and MES, and why small shops benefit too.
If you run a manufacturing plant, there is a good chance you are paying for Kepware. PTC's KEPServerEX has been the default way to connect PLCs, CNCs, and SCADA systems to IT networks for over two decades. It works. It is also expensive, Windows-only, and architecturally stuck in the point-to-point era that modern manufacturing is leaving behind.
In 2025, TPG Capital acquired PTC's Kepware division. Private equity acquisitions of industrial software follow a predictable playbook: raise prices, reduce R&D, extract margin. Manufacturers who rely on Kepware should expect 20%+ price increases starting in 2026 — and a product roadmap that prioritizes revenue extraction over innovation.
This article explains why the Unified Namespace (UNS) architecture makes Kepware's approach obsolete, what a migration looks like, and how to evaluate whether your plant is ready to make the switch.
Kepware is a protocol translator. It sits on a Windows server, connects to your PLCs via OPC-DA, Modbus, Siemens S7, Allen-Bradley DF1, or one of 150+ other industrial protocols, and exposes the data via OPC-UA or a proprietary API. Applications like SCADA, historians, and MES systems then connect to Kepware to read machine data.
This model made sense when the alternative was writing custom protocol drivers for every PLC-to-application connection. Kepware centralized the protocol translation problem into one product.
The problem is what happens next. Every application that needs machine data connects directly to Kepware. The historian polls Kepware. The dashboard polls Kepware. The MES polls Kepware. The AI model polls Kepware. You end up with a star topology where Kepware is the single point of failure, and every new consumer requires configuration, licensing, and testing.
Kepware Architecture (Star Topology)
PLC-1 ──┐ ┌── Historian PLC-2 ──┤ ├── SCADA CNC ──┼── KEPServerEX ───┼── MES Sensor ─┤ (Windows) ├── Dashboard Robot ─┘ └── AI Model Problem: Every consumer = new license + config Problem: Kepware down = everything down Problem: Windows-only = no edge compute
Kepware licensing is per-driver, per-channel, per-connection. A typical CNC shop with 10 machines, 3 PLCs, and a handful of sensors can easily spend $5,000-15,000 on initial licenses. Annual maintenance runs 20-25% of the license cost. Over 5 years, a mid-size manufacturer might spend $30,000-50,000 on Kepware alone — for software that does nothing but move data from point A to point B.
With TPG's acquisition, these costs are going up. Private equity firms do not acquire industrial software to lower prices. They acquire it because manufacturers are locked in — switching costs are high, and the data keeps flowing through Kepware every second of every shift.
Cost Comparison: 10 Machines, 5 Years
UNS uses MQTT (open standard) with Sparkplug B (open spec). No per-driver licensing. The cost is in the edge compute hardware and the platform that contextualizes the data — not in moving bytes.
The Unified Namespace replaces the star topology with a publish/subscribe model. Instead of every application connecting to Kepware, every system — machines, sensors, applications, databases — publishes to and subscribes from a single MQTT broker. The broker is the nervous system. Adding a new consumer is a subscription, not a project.
UNS Architecture (Publish/Subscribe)
PLC-1 ──┐ ┌── Historian PLC-2 ──┤ ├── SCADA CNC ──┼── MQTT Broker ───────┼── MES Sensor ─┤ (UNS) ├── Dashboard Robot ─┘ └── Edge Compute └── AI Model Adding a consumer = 1 subscription (minutes) Broker down = store & forward at edge Runs on Linux edge compute (no Windows)
The key differences:
The most common objection to leaving Kepware is: “We can't just rip out our entire data infrastructure.” You do not have to. UNS and Kepware can coexist. Here is the pragmatic migration path:
Install an MQTT broker and an edge compute node alongside your existing Kepware server. Configure Kepware's IoT Gateway (or a simple OPC-UA to MQTT bridge) to publish machine data to the MQTT broker. Your existing applications keep running through Kepware. New applications subscribe to the MQTT broker.
For each machine type, deploy an edge node that connects directly to the PLC or sensor — bypassing Kepware. Start with your highest-value machines. Each edge node publishes Sparkplug B messages to the MQTT broker with full ISA-95 context (enterprise, site, area, cell, equipment).
As edge nodes take over protocol translation, Kepware handles fewer connections. When the last application migrates to MQTT subscriptions, Kepware can be decommissioned. Your Kepware license renewal becomes your UNS operating budget — and it is smaller.
Migration Timeline
UNS is not the right move for every plant today:
If you decide to migrate, here is what matters:
Flowstate's Industrial Visibility Package connects your machines to a real-time dashboard in 7 days. No Kepware required. Sparkplug B + MQTT + ISA-95 hierarchy out of the box.
Kepware solved the right problem in 2005: connecting proprietary PLCs to IT systems when no open standard existed. In 2026, MQTT and Sparkplug B are that open standard. Kepware is now a toll booth on a highway that has a free lane.
TPG's acquisition makes the toll more expensive. The migration path to UNS is well-understood, non-disruptive, and pays for itself in the first year through eliminated licensing costs alone — before you count the value of real-time data, AI readiness, and edge resilience.
The question is not whether to move to UNS. It is whether you do it now — while you have budget and time — or later, when your Kepware renewal comes in 30% higher and your competitors are already running AI on their machine data.
