The Infrastructure Problem Nobody Talks About
Most conversations about AI in telecom focus on the intelligence layer: the models, the agents, the automation workflows. But a report published today by Telecoms Tech News highlights a more fundamental challenge that Deutsche Telekom is actively solving: the infrastructure underneath AI-powered operations is just as important as the AI itself.
Deutsche Telekom is leading a shift from vertical network silos, where each technology stack operates independently with its own hardware, software, and management tools, to a horizontal cloud architecture that separates infrastructure into three distinct layers: infrastructure, application, and automation. The logic is simple: AI systems that need to optimize across an entire network can't function when data and control are trapped in isolated vertical stacks.
As Deutsche Telekom's Franz Seiser notes, optimizing within individual silos produces local improvements that miss the bigger picture. System-wide AI automation requires system-wide visibility and control. That's what horizontal cloud architecture delivers.
Why This Matters for AI-Driven Operations
This architectural shift directly addresses one of the most common failure points in telecom AI deployments. Many carriers have invested heavily in AI models and automation tools, only to find that their underlying infrastructure can't deliver the continuous, real-time data these systems need. Older monitoring systems might provide a data point every five minutes, which is useless for AI that needs to detect and respond to anomalies in seconds.
The horizontal model solves this by creating a unified data layer that AI systems can query across the entire network, regardless of which vendor supplied the underlying equipment. For operations teams building AI-powered automation workflows , this means the difference between automation that works in isolated pockets and automation that can reason about and act on the full network picture.
The Vendor Relationship Is Changing Too
This shift has a significant consequence for vendors like Nokia and Ericsson that often goes overlooked. In the vertical model, these suppliers provided complete end-to-end stacks. In the horizontal model, they need to deliver specific components that plug into a shared platform. It's a substantial business model change that's already reshaping procurement conversations across the industry.
For carriers evaluating automation platforms, this trend reinforces the importance of choosing tools that are vendor-agnostic and can integrate across heterogeneous infrastructure. Platforms like Symphona Resolve that connect to multiple data sources and can orchestrate workflows across different network domains are better positioned for this horizontal future than tools built to work within a single vendor ecosystem.
Meanwhile in Construction: AI Estimation Gets a Boost
In a similar development for the construction industry, TrueBuilt announced its acquisition of Capabuild on March 10, combining AI-powered estimation with field documentation capabilities including LiDAR scans and 360-degree site walkthroughs. The move targets the $210 billion restoration and insurance mitigation sector, aiming to connect what happens in the field directly to preconstruction planning.
This acquisition reflects a broader pattern across both telecom and construction: the most impactful AI automation doesn't come from adding intelligence to existing workflows. It comes from connecting previously siloed data sources, whether that's network telemetry across vendor stacks or field documentation linked to project estimates, so AI can operate with a complete picture rather than partial information.
The Takeaway for Operations Leaders
Both of these developments point to the same lesson: successful AI automation is as much an infrastructure and integration challenge as it is an AI challenge. The carriers and contractors seeing the best results are the ones investing in data connectivity and platform architecture, not just smarter models.
If your AI automation initiatives are producing underwhelming results, the bottleneck may not be the AI itself. It could be the fragmented data and siloed systems sitting underneath it. Addressing that foundation is what turns isolated automation wins into enterprise-wide results. Platforms designed for cross-domain telecom operations or construction workflow integration can accelerate this by providing the connective layer between AI intelligence and operational reality.
.svg)
.svg)
.avif)
.avif)
.avif)
.avif)
.svg){kind=icon}
.svg){kind=icon}
.svg)
.svg)
.svg)
.svg)
.svg)
