Ericsson's 6G: An "Intelligent Fabric" with Native AI

Ericsson has outlined its vision for 6G, presenting it as an "intelligent fabric" where artificial intelligence is intrinsically integrated into the network itself. This perspective, showcased at Mobile World Congress (MWC) Barcelona 2026, marks a significant evolution from current approaches, where AI tools are often added as separate systems. The Swedish company envisions a network that not only transports data but is also capable of processing and reacting to it in real-time, combining connectivity, compute capabilities, and AI into a single, cohesive structure.

This deep integration of AI, extending from radio access to the network core, across transport and management systems, aims to create more autonomous and responsive infrastructures. The goal is to meet the growing demands of AI workloads, which require faster response times and more efficient data handling. For companies evaluating on-premise or edge LLM deployments, a network with these characteristics can be a crucial enabler for ensuring data sovereignty and optimizing TCO.

Architecture and Benefits of Integrated AI

Ericsson's approach places AI directly within the network, unlike current configurations that often treat it as an external component. This architecture allows networks to self-adjust, improving performance and reducing energy consumption without the need for constant manual input. Ericsson's roadmap for 6G does not involve a sudden leap but builds upon the evolution of existing technologies such as 5G Standalone and 5G Advanced, which serve as foundations for the progressive integration of AI-driven features.

Over time, this evolution will support advanced use cases such as industrial data processing, extended reality (XR), and "physical AI." These scenarios demand networks capable of handling high data volumes with extremely low latency, requirements that a natively AI-enabled network infrastructure is designed to address. The ability to process data close to the source, typical of edge environments, is fundamental for these applications, reducing reliance on centralized data centers and improving overall system responsiveness.

Collaborations and Preliminary Progress

Ericsson's progress in 6G is closely tied to a broad network of partners. At MWC 2026, the company highlighted collaborations with industry leaders such as Intel, NVIDIA, Qualcomm, and MediaTek. With Intel, Ericsson is exploring the combination of compute, connectivity, and cloud systems for AI-driven networks. The partnership with NVIDIA aims to embed AI across radio, edge, and core systems, while Qualcomm and MediaTek are involved in testing early 6G device and radio capabilities.

Ericsson also contributes to open-source efforts through the Linux Foundation, where it supports the development of software Frameworks for radio access networks. These collaborative efforts translate into concrete tests and prototypes. For example, Ericsson and Qualcomm tested physical-layer features for 6G, exploring the 6–8 GHz range to improve uplink performance. Another demonstration with MediaTek completed a data call using a 6G testbed and a prototype device, highlighting latency reduction and support for data-heavy applications like XR. Furthermore, a collaboration with Apple showcased spectrum sharing between 5G and a simulated 6G system, illustrating how operators might manage the transition between generations without wasting network resources.

Deployment Prospects and Infrastructure Implications

Ericsson's roadmap, developed in coalition with Qualcomm and other partners, targets initial commercial systems around 2029 and broader deployment closer to 2030. This timeline aligns with the work of standards bodies such as the 3rd Generation Partnership Project (3GPP), which is expected to release early 6G specifications later in the decade. This process includes hardware testing and software refinement, ensuring interoperability between systems from different vendors.

Erik Ekudden, Group Chief Technology Officer at Ericsson, emphasized the immediacy of this transition: "We are already on the journey toward an intelligent fabric, and it is happening right now. With clear proof points across the entire network, we are proving that a fully AI-powered network is not a distant capability five years out. By bringing intelligence into every domain today, we are giving the industry the foundation it needs to scale the next generation of AI." For CTOs and infrastructure architects, this vision implies the need to prepare their networks to manage more complex systems, larger data volumes, and ensure the reliability of AI-driven processes. The transition to 6G will be a gradual evolution, not a sudden leap, requiring continuous upgrades and integrations to support an increasingly demanding AI ecosystem, with direct implications for on-premise and hybrid deployment strategies.