Pages

Sep 4, 2025

Telecoms, Transformers, and Transportation

There is a common challenge that affects telecommunications, transportation, and now, the application of transformer technologies in AI systems. Telecommunications systems consist of multiple mobile networks and service providers. In many countries, there can be a handful of competing network operators, their direct-to-user service-provider arms, and multiple indirect channels in the form of (mobile) virtual network operators (MVNOs). 

One aspect of the value in this arrangement is that users of all providers can communicate with one another. This was not always the case. In the past, each operator aimed to keep customers on their network. Examples include SMS messaging during the USA in the late ‘90s, and walled garden services in the mobile internet era. There is a similar dynamic in the AI domain with large language model (LLM) providers attempting to bond consumers to their offerings. 

As the telecommunication industry grapples with 5G business case challenges and the prospects for 6G, there is a recognition that a focus on interoperability, is more likely to drive user awareness and adoption of new network technology capabilities. Perhaps the best illustration of this is in the collective approach that mobile network operators and equipment vendors are taking to developing a market for network APIs. 

Cross system interoperability 

Cross system and interoperability ideas are connected to developments and organizational innovations from the transportation sector. Several years ago, I contributed to the early planning stages for the California Integrated Travel (Cal-ITP) project. It works with public transit providers, of which there are hundreds across multiple transit modes, in California. Cal-ITP set out to solve the problem of transit riders lacking a consistent experience. There are barriers to gaining new customers, complicated travel planning across different systems, and extra expenses for individual transit providers. 


Standards are fundamental to cross-system journeys. It helps, for example, if individual network operators publish data feeds for their schedule and vehicle-location data using standard formats and up-to-date data. 

The availability of schedule data stimulates the market for competing journey planning Apps. This also helps transit riders who need to plan for multiple providers (rail and light-rail, bus-metro, ferry-metro etc.) when completing their trips. 

Verkehrsverbund Lessons 

The roots of service provider coordination ideas go back to the concept of a Verkehrsverbund (linked transport system). This came about when central cities in in Austria, Germany, and Switzerland saw improvements in regional transit as key to reversing their decline. Hitherto, network managers had an incentive to keep passengers on their networks to drive ticket revenue and passenger-miles metrics. In 1967, Hamburg’s main transit operator (owned by the city) took the initiative to coordinate and explore ways to integrate fragmented services and fares. It convinced all transit operators in the region to collaborate in a Verkehrsverbund. 

Hamburg experienced extraordinary success. The effective organization and coordination of many disparate transit services encouraged VVs in other metropolitan areas seeking solutions to declining transit ridership. 

In Cal-ITPs case, there is a modern twist on integrating transport systems. This goes beyond standardized transit data and involves the addition of contactless payment systems. The intersection of payments and transportation also allowed the initiative to support welfare service agencies through discount eligibility verification and discounted fare offers. This is the kind of adjacent market opportunity that mobile network operators and AI solution providers can target with IoT propositions. 

Linked Eco-system for the Telecommunications Industry 

For the telecommunications and transformer-AI sectors, the question is now about a modern-day equivalent of the VV. The telecommunications industry’s consolidated approach to APIs is one move in this direction. There need to be many more in anticipation of capabilities available via newer network generations (e.g., 5G-SA, hybrid cellular-satellite, 6G). 

A critical element of the Cal-ITP initiative was the education and training assistance provided to individual transit agencies. The telecommunications industry faces a new challenge in doing the same, both with developer communities and with enterprises seeking higher performing and dedicated networks. It is instructive to observe the sheer volume of developer support that the transformative-AI sector is deploying to drive adoption, even to the point of pushing the vibe-coding narrative that attracts non-developers and business individuals. In a sign that individual AI providers see a common industry cause while competing simultaneously, there are early signs of alignment tests. These seem ad hoc and not yet at a level that compares with interoperability testing enabled by formal standards bodies such as 3GPP, for example. 

The question is when and how will the telecommunications industry promote a common and scalable approach for developers and users to capitalize on mobile network capabilities. Beyond connectivity, there is a great deal on the horizon as more IoT, reliability, and sensing capabilities materialize in the 6G era. 

 

Image Credit Daniel Sessles via unsplash.com 

1 comment:

  1. 29 September 2025 update

    Contrary to industry optimism, AI agents are not inherently robust. Every endpoint, integration and context channel can be a pivot point, and “MCP compliance” is no easy task.

    While MCP enhances AI capabilities, it also broadens the attack surface - the sum of points where an attacker could exploit a system. This “AI-native” attack surface arises from the deep integration of AI with external systems, introducing risks such as:

    🎯 Increased Access: AI systems connected via MCP can reach sensitive data and tools, amplifying the impact of a breach.

    🎯 Autonomous Actions: MCP enables AI agents to invoke tools independently, which could be hijacked for malicious purposes if compromised.

    🎯 Shadow AI: The ease of deploying MCP agents may lead to unsanctioned integrations, evading enterprise security oversight.

    https://zerodaysnotice.substack.com/p/mcp-and-the-new-ai-native-attack

    ReplyDelete