Are Telco Network APIs Primed for Rapid Adoption?

This blog post is the third in our series, “Unlocking the Programmable Network.” In Part 1: The State of Telco Network APIs, we summarized the landscape, presented a taxonomy, and summarized the predominant Telco Network API ecosystem—GSMA Open Gateway. Part 2 highlighted why operators are pursuing Telco Network APIs and how they will boost their entire value proposition in a Cloud- and AI-dominated world. This post, Part 3 of our series, challenges the status quo with a contrarian thesis on rigid standardization versus a flexible approach.

The global telecommunications industry is striving towards a high-stakes pivot from utility to programmable platforms. The central thesis is that Network APIs—standardized via GSMA Open Gateway and the Linux Foundation CAMARA open source project—will unlock billions by exposing 5G and wireline network capabilities to the enterprise developer ecosystem.

However, our analysis reveals a widening chasm between strategic optimism and commercial reality. While the industry has rallied, ~80% of global mobile connections under Open Gateway, commercial traction has been elusive, with the lion’s share of Telco API revenues concentrated in fraud prevention, customer validation, and CPaaS, which existed long before the emerging open API framework began to take root.

Our thesis is that the rigid standardization of RESTful network APIs, while necessary for basic interoperability, is structurally insufficient to enable the next wave of technology—Agentic AI.

Agentic AI—autonomous systems capable of reasoning and executing multi-step workflows—fundamentally alters how network resources are consumed. Unlike static applications that are hard-coded and inherently capability-constrained, AI agents utilizing protocols like the Model Context Protocol (MCP) require stateful, dynamic discovery and behavior that static REST APIs struggle to deliver. Agentic AI compels the need to shift from a human-centric model, to a machine-to-machine model.

The bottom line- despite rosy forecasts projecting a multi-billion market by 2030, the 2025 opportunity does not exceed $300 M USD, driven by low-margin, SMS and OTP substitutes, based on well-established APIs that are proprietary to each major mobile operator. For CSPs to enable Telco API, a fundamental evolution is required from "API-as-a-Product" toward an "Agent-first" architecture.

As we addressed in the first two blogs in this series, leading mobile operators are coalescing around an open Network API framework, driven by the GSMA Open Gateway project. Open Gateway-defined APIs are currently being implemented in the Linux Foundation CAMARA open source project, which is also addressing the API development environment. At the time of publication, GSMA reports that Open Gateway involves 80 operator groups representing 292 mobile networks representing 80% of the world’s mobile connections.

Conceptually, API standardization is particularly important to enterprises, who fundamentally need to operate over a diverse and complex set of networks. Without a rigorous and stable API standard, enterprise developers are forced to implement and re-implement application integrations using operator-specific APIs (among other integration mechanisms), adding not only cost, but extending delivery timeframes as well. In a world tempered by on-demand cloud providers, and supercharged by instantaneous AI responsiveness, expensive, time-consuming application development will not suffice.

Rather, what is needed is a more seamless way to leverage network details for end-to-end services 

As Telco AI unfolds, the predominant AI use cases are distinguished as AI for Networks (‘Internal’ to achieve operational excellence) and Networks for AI(‘External’ for commercial monetization). The former set of use cases are of greater importance for internal automation initiatives, which we’ll discuss at some future date. We’ll focus on the external API ecosystem, which is what CAMARA/Open Gateway is designed to help standardize and proliferate.

This domain focuses on exposing capabilities to third-party AI agents. The more popular APIs featured in discussions to date include:

The conventional approach is predicated upon the Enterprise AI Application being fully aware of the environment it is operating in, including the operators, network technologies, OSS and BSS applications, etc. Details about how to utilize each of the Open Gateway / CAMARA network APIs must be hard coded into the AI Application imposing significant burden on the Developer.

Operator, technology, and regional variations, coupled with the RESTful, stateless, fixed APIs shift the complexity from the network to the developer, inherently limiting the degree of intelligence and sophistication that can be realized.

Table stakes for a viable App supporting a multi-national financial services, logistics, or technology firm consist of seamless services delivered end-to-end across a hybrid network.

A more agile model is needed.

The Model Context Protocol (MCP) is an open-source standard for integrating AI applications (Agents) to a diverse range of external systems. MCP is widely supported by the leading Large Language Models (LLMs) along with a broad range of commercially available AI applications, workflow automation, vibe coding, video and image generation, and many others.

Telecommunications operators can leverage MCP to abstract operator-specific capabilities to dramatically speed up time to market. Instead of being compelled to hard code the Open Gateway API calls, the AI Application will select the capabilities (also referred to as AI Tools) that the MCP server advertises, abstracting the underlying details for each operator’s OSS, BSS, and other applications. Refer to Figure 1 for a high-level operational flow for enabling AI via MCP.

Unlike static, RESTful Telco APIs (standard or proprietary), MCP employs a stateful model that maintains context across a multi-step workflow, allowing a more sophisticated operational process, further freeing the Enterprise AI Application Developers from telecommunications details and expertise. By leveraging the inherent natural language capabilities of the AI environment, ease of use is also significantly improved.

Because of the degree of automation, security, privacy, and compliance considerations may also be integrated directly into the MCP server, in accordance with the operator’s domains and environment.

The high-level of intelligence offered by AI, not only benefits the enterprise (and their developers), but can also be leveraged by the operators, who can proactively manage both their business and network infrastructure through their own AI Applications.

The CAMARA project has recognized the importance of MCP and recently published a whitepaper on the topic: https://camaraproject.org/wp-content/uploads/sites/12/2026/01/camara_wp_mcp_011226.pdf. It shows how external systems attempting to utilize CAMARA Network APIs can leverage the MCP Protocols and MCP servers to interact with the network.

MCP isn’t a completely free lunch. The MCP protocol itself is undergoing evolution to improve its security and robustness. Plus, unlike RESTful APIs, using MCP means relying on the reasoning logic of the language model powering the agent making the appropriate MCP call. So determinism and reliability are factors that will need to be carefully watched as MCP-fronted CAMARA APIs take root.

Like many other vendors in the ecosystem working on agentic frameworks, AWS is piloting Network Language Models (NLM) trained on 3GPP specs and vendor manuals. An "Agent" sits atop the heterogeneous network, translating intent ("Give me a slice") into proprietary core commands. This potentially obviates the need for perfect global standardization by moving the intelligence from the Interface to the Agent. Nevertheless, the risks we touched on still apply. We will have more to say on this topic now that we’ve seen a wide range of agentic approaches at MWC 2026 in Barcelona.

Although an AI-centric approach offers significant commercial benefits- simplicity, time to market, enhanced intelligence, at the top of the list- the industry is (very) early in the journey. While the hyperscalers are aggressively investing hundreds of billions in attaining their AI-centric vision, operators must attend to their broad customer base with legacy and existing mobile services continuing to grow. 

Among the fundamental challenges that not only operators, but their enterprise customers as well must embrace, is a shift from human-centered operations, to intelligent, autonomous platforms. Because this impacts every facet of their business, this will not happen over the night. A more prudent strategy is an incremental transition, albeit an accelerated one driven by AI timescales, which exceed even those for the cloud.

Another major challenge is the need for even more robust cybersecurity, authentication, privacy, data sovereignty, and compliance. Automation and autonomy remain vulnerable to the ever-escalating arms race, necessitating continual innovation. Attack surfaces in automation platforms require new proactive validation mechanisms along with nimble support organizations capable of responding quickly to incidents, especially in the face of AI-originated attacks.

Because AI regulations are at an early stage, in-country rules and regulations will vary (widely), and require a hybrid response. Government imposed limits on content, data storage, privacy, and data protection are projected to increase as the AI revolution unfolds. Compliance recordkeeping and best practices will need to be overhauled as a result.

Rapid proliferation of AI applications calls into question the fundamentals of Telco Network APIs. If the details can be abstracted, do the premises driving the Open Telco Network API ecosystem remain valid? What is the benefit of Open Gateway / CAMARA standardized APIs if operators plan to expose them along with their proprietary APIs to intelligent MCP servers? Is the additional time required warranted in the face of the HyperScalers literally rushing to establish AI competitive advantage?

Clearly there are no obvious and easy answers, and we expect operators to proceed in unpredictable ways to advance their own interests. Only the timescales are compressed.

The telecommunications industry is currently fighting the last war. The obsession with standardizing REST APIs (CAMARA) mimics the Web 2.0 enterprise strategy (Twilio, Stripe). However, the consumer of the future is not a human developer; it is AI Agents. And they have already landed around the globe.

To effectively compete, the entire industry must pivot from "exposing APIs" to "enabling Agents." Adopting MCP alongside Open Gateway / CAMARA APIs allows Telcos to expose richer, dynamic capabilities that Agents can self-discover, bypassing the slow committee-based standardization of GSMA. Another compelling benefit is to greatly accelerate end-to-end services across multiple technologies, operators, and geographies. 

Services for the AI-world require additional domains and disciplines that are commensurate with the new challenges. Ability to provide more granularity with adjusting performance, operational constraints (e.g, defining a Geo-Fence for a defined subset of users), and a more robust approach to compliance to a range of policies is essential.

Operators must also once again rethink their organizations, operations, technologies, and systems to determine how the exploit the unbridled capabilities of enhanced intelligence over the long-term. Investing in platforms today capable of facilitating the transition in an increasingly complex environment that could be further disrupted by AI breakthroughs that seemingly are falling from the sky. 

Final Verdict: Telco Network APIs are critical, and MCP or other agentic-friendly approaches will be necessary. The industry must evolve from "Static APIs for Apps" to "Dynamic Protocols for Agents" to capture the multi-billion opportunity. 

To stay informed about the rapidly evolving Telco Network API landscape, stay tuned for subsequent posts in this series, which will address the most important topics especially in light of the AI revolution.

To discuss your organization’s specific needs or to participate in AvidThink’s Telco Network API Research Report, please contact us today at [email protected]