From SPS 2025, Looking at the Structural Turning Point in German Industrial Automation: Software-Defined, Industrial AI, and the Repricing of Manufacturing Competitiveness

Structural Turning Point in German Industrial Automation as Seen from SPS 2025: Software-Defined, Industrial AI, and the Repricing of Manufacturing Competitiveness

A Trade Fair Phenomenon That German Industry Should Be Wary Of

In November 2025 in Nuremberg, the SPS (Smart Production Solutions) trade fair drew more than 55,000 visitors, up 9% from the previous year; the number of exhibitors reached 1,175, an increase of 5%. But more noteworthy than the scale is the structural change in the content on display: control, drive, and sensor technologies were still present, yet no longer the focus. What truly took center stage was software, IT, and AI. At the same time, participation by Chinese exhibitors rose 41% year on year.

For German industry, these three figures taken together pose a question that must be answered: when the center of value in automation shifts from hardware to software, how much longer can the "mechanical precision + proprietary control" model on which German manufacturing rests sustain its competitive advantage?

This is not a question of a trade fair's product list, but a question of the underlying competitive logic of the German industrial system.

I. Software-Defined Automation: Control Shifts from Hardware to Platforms

The clearest signal from SPS 2025 is that Software-Defined Automation (SDA) has moved from an "innovation topic" to a platform strategy. Its technical core is decoupling control software from proprietary hardware.

Siemens anchors its SDA roadmap on three pillars: Industrial Edge as standardized software infrastructure, a virtualized control runtime (vPLC), and IT-like engineering (SIMATIC AX). Two points are noteworthy: its vPLC runtime can already run on any x86 device; at the same time, Siemens has launched an entry-level CPU 1511V, intending to lower the adoption threshold so that SDA is no longer limited to high-end lighthouse projects. Audi, one of its flagship customers, said that the vPLC running at its Neckarsulm and Ingolstadt plants has been in operation for six months with 100% availability.

Rockwell Automation's Logix Edge is heading in the same direction—the Logix engine runs as an isolated workload on industrial PCs, with release planned for the fourth quarter of 2026. Endress+Hauser's statement is even more symbolic: field instruments should be managed as IT assets, decoupled from hardware and kept updatable, similar to modern mobile devices.

Interpretation from a German perspective: The moat of the German automation industry over the past decades has been the lock-in effect created by "proprietary controller hardware + tied software ecosystem." SDA structurally weakens this lock-in: once control logic can run on general-purpose computing power, customer migration costs fall, and differentiation is forced to shift to I/O reliability, functional safety certification, engineering toolchains, and ecosystem integration capabilities.For a German leader like Siemens, this is not a case of passively taking a beating but a proactive strategic repositioning—using an open software ecosystem to gain platform-layer influence. But the cost is equally clear: differentiation at the hardware level is diluted, and competitors are no longer just European peers but also IT vendors and cloud service providers.

II. The cyclical recovery is real, but the "depth" is insufficient

Market signals are indeed improving. Siemens Digital Industries (DI) achieved 9% comparable growth in the fourth quarter of fiscal 2025 (about €5 billion), with its automation business growing 10% and software growing 8%, but on a comparable basis the full fiscal year was still -4%—indicating that the rebound is back-loaded rather than broad-based. Its fiscal 2026 guidance is 5%–10% comparable revenue growth and a 15%–19% profit margin, and management explicitly lists factory automation as a growth opportunity.

Beckhoff describes 2025 as "volatile": strong in the second quarter and weaker in the third; the previous 33% decline in 2024 mainly stemmed from inventory correction, and this factor has largely cleared, with the company expecting overall growth of 7%–10% in 2025. Data from the German Electrical and Digital Industry Association (ZVEI) also supports the "upward" narrative, with a September rebound in orders improving cumulative year-to-date performance, but it simultaneously warns that global trade barriers and geopolitical uncertainty remain persistent downside risks.

Underlying logic: This is normalization after inventory-cycle clearing and order pull-forward, not a structural expansion of demand. Siemens CEO Roland Busch's remarks are highly informative: the market is recovering, especially in factory automation, but machine building remains weak.

This point is crucial for understanding German industry. Mechanical equipment manufacturing is the domestic-demand foundation of Germany's automation industry—the investment cycle for machine tools, special-purpose equipment, and construction machinery directly determines the depth of domestic automation orders. If the base does not move, the automation recovery lacks depth: it looks more like inventory restocking and technology upgrading rather than a capacity expansion cycle.

III. Three Levels of Industrial AI Entering the Production System

At SPS 2025, AI is no longer a concept demonstration but is being implemented in layers.

First layer: Generative AI copilot. Many vendors showcased AI assistants for industrial scenarios, used for querying, diagnostics, and decision support. The barrier to entry at this layer has dropped rapidly and is becoming commoditized.

Layer 2: Agentic AI. Schneider Electric positions agents as autonomous engineering agents within the EcoStruxure Automation Platform: converting user specifications and technical specifications into control applications for EcoStruxure Automation Expert, supported by digital twin creation, testing, and simulation, and using the MCP framework for tool calls to complete deployment, simulation, and unit testing, ultimately subject to human approval. Siemens, through the low-code Insights Hub Copilot Studio (launched in March 2025), enables end users to build AI agents themselves using "system instructions + modular skills (such as OEE analysis)."

Layer 3: Edge AI. NPUs are embedded directly into industrial PCs and controllers to enable local real-time inference, avoiding cloud latency and bandwidth costs, with applications expanding outward from visual inspection. At the same time, the edge AI software stack is maturing.

Interpretation from a German perspective: One of the structural bottlenecks of German industry is the shortage and aging of engineering talent, especially in control engineering. If agents can compress the labor hours for control program generation, testing, and validation, their significance for Germany is not that of an "efficiency tool," but rather the replacement of a scarce factor of production.

But the risk is equally structural: once engineering knowledge is platformized and sedimented into the upstream software ecosystem, the tacit process knowledge on which German medium-sized machinery manufacturers (Mittelstand) have long relied may be absorbed by upstream platforms, thereby weakening their bargaining power in complete-machine and production-line integration. This is a strategic issue that German industry must manage at the same time as it embraces industrial AI.

IV. Data Infrastructure: The Path Germany Is Actually Good At

Another main thread of SPS 2025 is the data layer: Industrial DataOps and unified namespace applications are becoming more widespread; Single Pair Ethernet and Ethernet-APL are moving toward standardization.

These topics lack visual impact, but they are the precondition determining whether AI can scale. Without a unified semantic layer, industrial AI can only remain at single-point visual inspection; without continuous data at the field layer, agents cannot obtain verifiable context.

Interpretation from a German perspective: Standards, interfaces, and interoperability are precisely the areas where German industry excels. The discursive power of companies such as Siemens, Beckhoff, and Endress+Hauser has long been built on the ability to "write engineering specifications into standards." The standardization of SPE and Ethernet-APL is essentially infrastructure investment to open up field-layer data. Its payback period is long, but once established, it will determine the depth of assetization of factory data over the next decade.The problem is: German companies are accustomed to leading at the standardization level, yet lag behind US vendors at the data platform and cloud service level. If standards move first without a supporting commercialization platform, value may be intercepted upstream.

V. Two New Narratives: Physical AI and Data Centers

Two noteworthy narratives also emerged at the exhibition. The first is Physical AI—AI with the ability to act in the physical world—which is still at a forward-looking narrative stage and has not yet achieved scaled deployment. The second is that industrial OEMs are beginning to view data centers as a new growth vertical.

Interpretation from a German perspective: The latter has greater practical significance. Germany's mechanical manufacturing capabilities in power, cooling, power distribution, cabinets, and precision machinery can be transferred into data center infrastructure. This is an alternative market path when traditional industrial demand is weak. However, it has structural constraints: this market is driven by the investment cycles of US hyperscale cloud providers, and German companies are more equipment suppliers than rule makers, with a profit structure and bargaining position different from those with traditional industrial customers. In addition, the growth in data center electricity demand creates direct tension with Germany's persistently high energy costs and insufficient grid investment.

VI. OT Security: Boundary Moves Forward, Cost Attribute Changes

The enforcement boundary of OT security is moving closer to industrial assets themselves. This change is not isolated: the decoupling of control software and hardware, increased edge computing power, and the opening-up of the data layer—these three combined extend the attack surface from enterprise IT to production-line equipment.

Interpretation from a German perspective: As IT and OT converge, security is no longer merely a compliance cost, but a system availability cost. For continuous-production industries, the losses from unplanned downtime are far higher than security product procurement spending. German industry has previously been relatively lagging in OT security investment, and this shift in perception will gradually be reflected in the structure of capital expenditure.

VII. What This Means for the German Manufacturing System

First, the advantages have not disappeared, but the point of value capture has shifted. Germany's accumulated strengths in mechanical precision, engineering standards, industry know-how, and field-level standards remain valid, but profits are shifting toward software platforms, the data layer, and AI capabilities.

Second, the competitive landscape is a "pincer attack from above and below." Lower-tier edge hardware and mid-to-low-end automation face intensifying price competition—the 41% year-on-year increase in Chinese exhibitors' participation is a clear signal; upper-tier platforms are dominated by US cloud and software vendors. The realistic positioning of German companies is reliability at the field level and the industry semantic layer.

Third, the capability gap is widening. SDA and industrial AI require companies to have IT engineering capabilities, while many German mid-sized companies lack software teams. This may lead to digital divergence within German industry: leading companies accelerate platformization, while SMEs are forced to depend on external ecosystems.

Fourth, computing power and chips depend on external sources. The spread of edge AI means deeper dependence on chip and computing power supply, and Europe's local supply capacity still needs to be strengthened.

VIII. European and Global Impact

At the European level, SDA, DataOps, and the unified namespace are highly aligned in direction with the EU's push for digital sovereignty and its Industrial Data Space agenda. If Europe can establish certification rules for AI validation, functional safety, and industrial data governance, it will have an opportunity to turn "compliance burden" into "rule-setting advantage"—one of the few structural levers Europe has in platform competition.

At the global level, the landscape is stratifying: the US dominates AI platforms and cloud, China dominates cost and scale, while Germany and Europe focus on mid-to-high-end engineering capabilities and standard-setting. The spread of SDA will reduce the stickiness of proprietary controller ecosystems, thereby lowering barriers for new entrants—a double-edged sword for incumbents.

9. Trend Judgments for the Next 3–10 Years

1. Control-layer virtualization will become the default architecture. Hardware differentiation will shift to I/O, reliability, functional safety certification, and engineering toolchains.

2. Industrial AI will move from co-pilot to controlled autonomy. Humans will still retain approval nodes, and functional safety and AI validation standards will become new compliance barriers, with Europe having an opportunity to establish a rule-setting advantage here.

3. Engineering capability will be repriced. Hybrid engineers who understand processes, software, and data will become scarce resources, and the importance of talent development systems will rise.

4. Field-level standards will determine the speed of AI deployment. The installation progress of SPE, Ethernet-APL, and the unified namespace will directly determine the differences in depth of data assetization among companies.

5. Data centers and energy infrastructure will become the second growth curve for German mechanical engineering, but their growth potential is constrained by energy policy and the pace of grid investment.

6. OT security will shift from a cost item to a prerequisite for system availability, and will gradually be reflected in industrial enterprises' capital expenditures and organizational structures.

7. The automation market will recover moderately, but true differentiation lies not in the cycle, but in mastery of software platforms and the data layer.

Conclusion

What SPS 2025 truly showcased is not a product list, but a restructuring of the foundations of German industrial competitiveness: from "mechanical precision + proprietary control" to "open architecture + software capabilities + data assets."

In this transformation, Germany's engineering tradition remains an asset, but it needs to be proven anew—in a world where control logic can run on any compute resource, AI can generate control programs, and data can flow across layers, German manufacturing must answer a more difficult question: when hardware no longer locks in customers, what is truly irreplaceable?

Future observation indicators worth continuously tracking include: Siemens Digital Industries' order momentum and margin realization, the actual installation scale of vPLC, the conversion rate of industrial AI from trade show demonstrations to volume deployment, field-level installation progress of SPE and Ethernet-APL, changes in Chinese suppliers' share of the European automation market, and the frequency and nature of OT security incidents.