Local AI: Governance Before Hardware Purchase
Benedikt Langer
10 min readFour developments over two weeks show that locally operated AI goes far beyond the tech stack. ...
Read Article
7 min read
In April 2025, the Federal Network Agency allocated a total of 465 frequency licenses for private 5G campus networks. The market will remain well below the 13,500 networks initially planned for 2026. This is not a disappointment; it is an adjustment. Three years of pilot operations have demonstrated that 5G on campus works, but not for all initially registered use cases.
Key Takeaways
- 465 allocations, not 13,500. The Federal Network Agency has allocated more frequencies than ever before, but the initial target has not been met. Growth is slowing, and the pilot phase is complete.
- AGV and intralogistics are gaining momentum. Automated vehicles and driverless material handling are the use cases that have been validated. Sub-10ms latency and stable handover offer concrete operational benefits.
- General connectivity remains the domain of WiFi-6E. Without strict quality of service requirements, WiFi-6E is more economical. The effort to install private 5G only justifies itself for specific industrial applications.
- Three decision paths for IT departments. Expand (if AGV/robotics are involved), hybrid with WiFi-6E (mixed usage depending on the use case), or reduce (if no productive use cases have been identified).
- Budgetary implications for 2026. Validated pilot projects in 2024 now require a decision: migration to production or accounting write-off. Neither scenario is wrong, provided the figures are consistent.
What has actually been built among the 465 frequency allocations
Germany has quickly established itself as a global leader in private 5G industrial networks. Since 2019, the Federal Network Agency has been allocating frequencies in the 3.7 to 3.8 GHz band with flexible access conditions and reasonable fees. As a result, 465 frequency allocations in April 2025 represent a 23% increase compared to the previous year, with Germany capturing 31.6% of the European private 5G market.
This looks like a success. The nuance lies in the scale.
A frequency allocation does not guarantee an operational network. Between approval and productive operation, system integration, OT connection, terminal certification, and often a painful lesson about the actual maturity of IT/OT convergence intervene. Companies that requested a frequency in 2023 now either have a functional network or a lesson learned about the underestimation of complexity.
465
Frequency allocations (3.7-3.8 GHz) in Germany in April 2025. European market share: 31.6%.
Source: Federal Network Agency / Consultancy.eu 2025
Use Cases That Have Really Taken Off
Three years of pilot operations offer a clearer picture than any analyst’s forecast. Reality splits into two categories: use cases that have delivered results and those that remain stuck in the proof-of-concept (POC) phase.
What Works: AGV and Intralogistics. Automated guided vehicles (AGVs) and autonomous material handling have been validated as use cases for private 5G networks in production. A large distribution center replaced 120 WiFi access points with 8 private 5G antennas, eliminating radio frequency interference and enabling more than 60,000 daily handovers without interruption for 400 AGVs. A tile factory replaced 36 WiFi access points with 12 5G access points and resolved issues with high ceilings and heavy metal infrastructure that physically limited WiFi. Latency below 10 ms with 99.9999% availability is no longer just a marketing promise; these are measurable operational parameters.
Use Cases With Mixed Results: Predictive Maintenance and Remote Quality Control. Camera-based quality inspection and sensor fusion for predictive maintenance worked well in the lab. In productive environments, data volumes are higher than anticipated, edge architecture is more complex, and OT integration is more expensive than the 5G infrastructure itself. Those still in the POC phase here face an architectural decision, not a technical one.
Still in POC: General Office and Plant Connectivity. This was never the right use case. Those requesting a campus network for general connectivity underestimated WiFi6E and overestimated 5G.
Private 5G vs. WiFi6E: When Each System Becomes Viable
Private 5G: Makes Sense When…
- AGV/AMR fleets in operation or planned
- Latency requirements below 10 ms
- High-frequency handovers in large factory areas
- Strong electromagnetic interference from machinery
- Heavy metal infrastructure (WiFi signal attenuation)
- Production criticality with 99.999%+ reliability
WiFi6E: Makes Sense When…
- Office and administrative buildings within reach
- Existing WiFi infrastructure can be upgraded
- No strict QoS requirements (latency, reliability)
- Prioritization of lower installation and operational costs
- Mixed environments with consumer devices
The most common mistake in decision-making: treating 5G and WiFi as competitors rather than complementary layers. A hybrid operation, where 5G powers the production plant and WiFi6E serves the offices and social areas, is the technically and economically most appropriate solution for industrial facilities with mixed requirements.
What the DSI Decides Today After Three Years of Pilot Projects
Companies that launched pilot projects in 2023 or 2024 are now at a critical decision-making point. Here are the three viable paths:
Production Expansion: If AGV, robotic manufacturing, or critical Edge applications for production are integrated, the business case is clear. Investment in productive infrastructure pays off through operational stability and scalability. What matters most is the integration of OT, not just the technology itself.
Hybridization with WiFi6E: For facilities with mixed needs, the question is not 5G versus WiFi, but rather which network layer powers which application. This configuration reduces total cost of ownership (TCO) while still offering industrial performance where needed.
Orderly Discontinuation: If the pilot project demonstrates no productive use cases, discontinuation is the right decision. Not every proof of concept (POC) needs to go into production. An honest retrospective costs less than continuous investment in an infrastructure with no operational benefit.
The financial implication is direct: those without a clear commitment to one of the three paths by 2026 will pay for infrastructure with no ROI horizon. This is not a 5G issue, but a decision-making task.
Frequently Asked Questions
How many private 5G campus networks are actually active in Germany?
The Federal Network Agency has allocated a total of 465 frequencies by April 2025. The number of actually operational networks is lower, as frequency allocation and productive operation are separate stages. The target of 13,500 networks by the end of 2026 will not be reached.
When does it become cost-effective for SMEs to install a private 5G campus network?
Clearly in AGV/AMR fleets, critical intralogistics for production, and applications requiring latency below 10 ms. It is harder to justify for general building connectivity or applications that WiFi6E can also cover reliably. Installation costs are higher than WiFi6E, but they pay off quickly if the use case matches.
In what ways does private 5G differ from WiFi6E in industrial environments?
Private 5G uses licensed frequency bands with exclusive spectrum access, offering greater robustness against electromagnetic interference from machinery and more deterministic latency. WiFi6E operates in the unlicensed 6 GHz band, is more cost-effective, and suffices for most standard use cases. In industrial environments with heavy metal infrastructure and high reliability requirements, private 5G provides measurable advantages.
What is the most common cause of 5G campus project blockages during the POC phase?
The maturity of IT/OT convergence, not the radio technology itself. Integrating 5G into existing OT systems, production machinery, and legacy automation is often more complex and costly than the 5G infrastructure itself. Those who underestimated this effort now have technically functional networks but no productive connections.
What should the DSI prioritize on its agenda regarding 5G campus networks in 2026?
A clear decision on the path to follow: production expansion, hybridization with WiFi6E, or orderly discontinuation. Those who decide on none of the three options will pay for infrastructure with no ROI horizon. The decision should be guided primarily by the use case portfolio, not by technological preferences.
Additional Lectures
Three Corporate Spin-offs in 2026: Continental, Infineon, and ThyssenKrupp Reorganize
Professional Quantum Computing: What Will Be Truly Useful in 2026
More Content from MBF Media Platform
cloudmagazin: BSI-KRITIS and NIS2 in the Multi-Cloud Context for 2026
Cover Image Source: Pexels