Beyond Speed: How AI, Sensing, and 6G Are Transforming Cellular Networks into Intelligence Platforms

Traditionally, every generation of cellular technology has been defined primarily by a singular, straightforward pursuit: moving data faster. From the early days of mobile data to the widespread rollout of 5G, the benchmark for success has always been higher peak download speeds, lower latency connections, and increased bandwidth to handle a growing volume of mobile internet traffic. However, the telecommunications industry’s ongoing transition toward 6G is shaping up to be a fundamentally different kind of upgrade. Instead of focusing exclusively on how fast a network can move data from point A to point B, the industry’s strategic focus is shifting toward what new services, capabilities, and real-time experiences the network can actually deliver, with artificial intelligence serving as the primary catalyst for this historic evolution.

Unlike previous generations where intelligence was largely an afterthought or layered on top of legacy architectures, 6G is being designed from the ground up to embed AI deeply into both the radio access network (RAN) and the core network infrastructure. This deep integration will enable future networks to predict demand peaks autonomously before network congestion even sets in, managing radio resources dynamically and in real time. Beyond internal optimization, that same underlying intelligence could enable the network to leverage its own physical infrastructure and ambient wireless signals to provide data services and gather rich information about the surrounding physical environment.

While artificial intelligence features already exist in modern 5G deployments, 6G is engineered to weave these capabilities into the fabric of the network architecture. The definition of 6G continues to take shape through the rigorous 3rd Generation Partnership Project (3GPP) standards process, with many foundational technologies currently sitting in the research and trial stages. Nonetheless, the long-term direction of the industry is becoming increasingly clear. Technologies such as artificial intelligence, distributed computing, and radio sensing are rapidly becoming just as integral to cellular planning as traditional metrics like new spectrum frequencies, lower latency targets, and higher raw data rates.

6G Standardization and the Evolving 3GPP Roadmap

The formal global standardization of cellular technologies relies heavily on the structured milestones established by the 3GPP. The organization officially began studying 6G requirements and feasibility as part of Release 19, which wrapped up in December 2025. Following that milestone, Release 20 is tasked with evaluating the foundational items required for 6G. This phase includes deep-dive studies into artificial intelligence integration, integrated sensing capabilities, and the completely redesigned architecture of the 6G network, with the entire release expected to be fully completed by 2027.

Looking further ahead, Release 21 is now officially open for industry contributions and represents a critical turning point as the first normative 6G specifications phase. Pending the finalization of its precise scope and timeline, Release 21 is expected to build directly upon the foundational studies conducted during Release 20, culminating in a formal protocol freeze targeted for 2030.

Despite the formal timeline stretching toward the end of the decade, commercial 6G systems are widely anticipated to arrive around 2030, though pre-commercial deployments, field trials, and early demonstrations will undoubtedly appear much sooner. Many of the advanced technologies currently being considered for 6G are already making their debut through interim 5G-Advanced deployments. This transitional phase gives equipment vendors and telecommunications operators a valuable window of opportunity to test complex capabilities—such as AI-driven network optimization and integrated sensing and communication (ISAC)—in live environments before committing massive capital to larger-scale 6G deployments.

At the same time, the momentum behind 6G development is not restricted to commercial consumer networks. Governments and defense departments worldwide are also actively exploring potential use cases, particularly in operational scenarios where high-speed communications, distributed computing, environmental sensing, and real-time decision-making must work seamlessly together. For instance, the U.S. Department of Defense is actively exploring 6G technologies and hardware prototypes for specialized military applications, adding an entirely separate, high-stakes source of demand for early 6G systems.

The U.S. government’s Mission 6G 28 initiative represents another significant push toward accelerated early testing and real-world validation. The initiative is actively encouraging industry-led demonstrations of cutting-edge technologies, including AI-driven networking and integrated environmental sensing, ahead of major global events like the 2028 Los Angeles Olympic Games. These high-profile demonstrations utilize pre-standard technology, providing researchers and engineers with an invaluable early look at which theoretical concepts can successfully transition from academic research laboratories into demanding, real-world operational environments.

Simultaneously, infrastructure vendors are rapidly developing specialized tools and testing platforms to support these early trials. These solutions give network operators and internal research and development teams a practical way to evaluate new capabilities long before global standards are officially finalized and frozen.

The structured standards process provides the telecommunications industry with critical milestones to monitor closely over the coming years. With Release 20 studies scheduled to conclude in 2027, the subsequent development of normative 6G specifications in Release 21 will lead toward a final protocol freeze expected in March 2029. Once those specifications are locked in, the immense pressure shifts directly to network operators and equipment manufacturers to translate complex technical documents into reliable, commercial-grade systems.

The Transformation of the Radio Access Network via AI-RAN

One of the clearest, most transformative examples of artificial intelligence moving deeper into cellular architecture is happening within the radio access network. The RAN encompasses the physical base stations, towers, and antennas that directly connect consumer devices and enterprise hardware to the wider cellular network. Traditionally, RAN equipment has been constructed around proprietary hardware architectures and custom-built silicon designed exclusively for telecommunications-specific functions, resulting in rigid, long upgrade cycles that made rapid innovation difficult.

Artificial intelligence is actively challenging this legacy model, though the broader industry is far from taking a single, unified approach to the transition. Major infrastructure players are charting distinctly different paths toward an AI-driven RAN.

Ericsson is pursuing a more evolutionary, pragmatic approach. The company is embedding advanced AI capabilities directly into the existing RAN infrastructure that telecommunications operators already have deployed in the field. The core objective of this strategy is to unlock meaningful performance improvements and efficiency gains without forcing operators to undertake costly, large-scale hardware replacements across their entire footprints. Early field trial results for this evolutionary path point to tangible gains in overall network efficiency and data throughput, particularly in complex operational areas like traffic scheduling and real-time radio resource management.

In contrast, Nokia and Nvidia are pushing for a far more disruptive and transformative model. Their collaborative AI-RAN approach utilizes graphics processing unit (GPU)-based computing running alongside traditional telecommunications workloads, effectively turning standard base station hardware into a much more flexible, multipurpose computing platform.

Nvidia strongly backed this ambitious vision with a substantial $1 billion strategic investment in Nokia, and the two corporate partners already have live field trials up and running in cooperation with major carrier T-Mobile. However, packing powerful GPUs directly into the RAN introduces notable engineering challenges. It brings additional computing capacity to the network edge, but it also triggers significantly higher power consumption, stricter cooling requirements, and elevated infrastructure deployment costs.

Ultimately, network operators will have to weigh these operational burdens carefully. They must decide whether the new revenue streams generated by running distributed AI workloads at the network edge will be sufficient to justify the increased capital and operational expenditures. If an operator’s primary goal is simply to achieve better baseline network performance and spectral efficiency, adding lighter AI layers to existing RAN infrastructure may prove more than sufficient. However, the commercial business case for a heavy, GPU-based AI-RAN becomes considerably stronger if operators identify lucrative opportunities to monetize that extra edge-compute capability by offering external services.

Diverging Paths in the 6G Core

While the radio access network handles the wireless connection to devices, artificial intelligence is simultaneously reshaping the network core—the central brain responsible for routing traffic, managing user sessions, and allocating overarching network resources. Recognizing this profound shift, the 3GPP advanced two competing architectural approaches for integrating AI natively into the 6G core, with both options currently being studied in parallel by international working groups.

The first approach, designated as Solution Variant #18.1, embeds artificial intelligence directly into the core network architecture through a series of new, agent-based network functions. In this model, these intelligent agents would possess the autonomy to interpret high-level intent directly from a user or an enterprise application and then orchestrate existing underlying network functions automatically to carry out the desired outcome.

The second approach, known as Solution Variant #18.3, takes a more modular route by keeping artificial intelligence structurally separate from the core itself. Under this architecture, a dedicated AI domain interacts with existing core network functions through a specialized translator function. This allows an external or parallel AI layer to optimize, manage, and orchestrate network behavior intelligently without fundamentally rewriting or altering the underlying core network functions.

Geopolitical and industry alignments have naturally emerged around these two competing technical visions. Variant #18.1 has been championed primarily by Chinese vendors and network operators, including major industry heavyweights like Huawei, ZTE, and China’s largest national carriers. Meanwhile, Variant #18.3 has secured robust backing from a coalition of Western vendors and telecom operators, including Nokia, Ericsson, AT&T, T-Mobile, Qualcomm, and Google. Interestingly, South Korea’s SK Telecom is actively involved in supporting both competing paths, reflecting the reality that the global telecommunications industry has not yet settled on a single standard architecture.

For the time being, the 3GPP has chosen to keep both structural options open for continued study. Where artificial intelligence ultimately ends up sitting within the 6G core will profoundly shape future network architecture, technical complexity, and vendor influence for decades to come.

Integrated Sensing and Communication

As artificial intelligence equips cellular networks with advanced decision-making and self-management capabilities, the addition of radio sensing provides the system with continuous, high-resolution information about the physical environment surrounding it. Integrated Sensing and Communication (ISAC) utilizes existing wireless cellular signals simultaneously for traditional communication and environmental sensing. This dual-purpose methodology allows standard cellular infrastructure to detect, locate, and track physical objects and motion without requiring the deployment of expensive, dedicated radar hardware.

Recent field trials are beginning to demonstrate what ISAC can achieve in real-world scenarios. In July 2026, telecommunications provider AT&T and infrastructure vendor Ericsson collaborated to utilize existing commercial 5G infrastructure installed outside the AT&T Stadium in Texas to successfully detect, locate, and track multiple drones flying at altitudes between 300 and 400 feet. The experimental setup relied entirely on massive multiple-input/multiple-output (MIMO) antennas, advanced signal processing algorithms, and AI-enabled sensing software, completely bypassing the need for a separate radar deployment.

While this high-profile demonstration serves as a powerful technical proof point, industry experts note that ISAC is not yet ready to completely replace dedicated radar systems. Instead, it is far more likely to serve in the near term as a valuable, cost-effective additional layer of environmental sensing, particularly in dense urban or industrial environments where cellular infrastructure is already widely deployed. Potential early use cases include automated drone detection, enhanced perimeter security, industrial automation monitoring, smart ports, and complex logistics facilities.

The ultimate test for ISAC will be whether the technology can successfully transition out of controlled trial environments and prove its reliability, accuracy, and performance across diverse geographical conditions, adverse weather, varying distances, and different object types. If ISAC successfully clears these operational hurdles, it could provide AI-driven networks with a vital new source of real-world data, enabling systems to make automated decisions based dynamically on what is happening simultaneously inside the network and out in the physical world.

Looking Ahead to the 6G Era

The coming years will definitively determine whether 6G can successfully deliver on its ambitious promise: transforming cellular infrastructure from a mechanism for faster consumer connections into a comprehensive, distributed intelligence platform. The clearest checkpoints for this transformation are marked clearly on the global standards calendar, beginning with the conclusion of Release 20 studies in 2027, followed by architectural finalization and the freezing of the first normative 6G specifications in March 2029.

These upcoming milestones will establish the technical baselines, but the ultimate success of 6G will depend on whether network operators can prove that embedding artificial intelligence, distributed edge computing, and environmental sensing deeper into the network creates enough commercial value to justify the associated financial costs and engineering complexity. This fundamental shift is what truly separates 6G from all previous cellular generations. By evolving beyond the traditional role of simply connecting mobile devices and instead becoming an active participant in computing infrastructure—processing AI workloads, optimizing resources, and interpreting the physical world—the cellular network is entering an entirely new era of intelligent connectivity.