Image © 2026 GSMA / MWC
Mobile World Congress Barcelona is the perennial event for the mobility industry. From its start exactly twenty years ago, MWC has showcased the latest in cellular infrastructure and device advancements dating back to the second “G.” Two decades is an eternity in any industry, and this year’s show highlighted the rapid advancements in cellular infrastructure and mobile devices. From my perspective, I walked away (after logging thirty miles dashing between halls at the Fira) with three big takeaways – so let’s dive in!
AI Everywhere
Not surprisingly, an overwhelming theme of AI everywhere punctuated MWC this year. From my perspective, AI washing has evolved into agentic washing, and credit NVIDIA for much of the noise. I call it noise because it is difficult to discern fact from fiction and outcome, given the claims from both established incumbents and start-ups all wanting to cash in on the AI gold rush. There is plenty to examine here with the news cycle at MWC, ranging from AI-infused Radio Access Networks to integrating AI into core infrastructure and the exploration of personal applications of AI.
NVIDIA has designs to reimagine radios and gobble up another opportunity to leverage its market leadership position in GPUs, coupled with software tech stacks, to penetrate the telecom market. By all measures, it is succeeding in its endeavors, but companies including Spectrum Effect, Opanga Networks, and Cohere Technologies were well ahead of NVIDIA’s plans in leveraging machine learning and AI to improve visibility into contention, resolve interference, and enable spectrum efficiency. Consequently, AI RAN is not necessarily a new concept, except for the development of agentic frameworks today. With that said, I spent time at MWC with Ericsson, Nokia, and Samsung Networks, and they all shared the dramatic improvements they are seeing in their RAN infrastructure with AI.
From an AI core infrastructure integration perspective, T-Mobile is one of the first mobile network operators to demonstrate the power of 5G Standalone and AI to unlock potentially disruptive new services. I recently wrote about the Uncarrier’s launch of a live language translation service following its recent Capital Markets Day. What makes the voice service compelling is that it eliminates the need for an application and the associated user friction. In speaking with T-Mobile’s chief technology officer, John Saw, he views the service as a demonstration of what is possible with a highly programmable network that embraces an AI-native architecture. I wholeheartedly agree.
Finally, there is a lot of talk lately about physical AI, but a logical interim step to the manifestation of Star Wars-esque droids in our homes and businesses performing tasks is personal AI. Early pin solutions were failures, possibly because they were ahead of their time in terms of AI maturity and the wrong form factor, but Qualcomm’s announcement at MWC of the Snapdragon Wear Elite platform could change things. It marries an NPU with 5G Redcap and provides a platform that developers can leverage and scale more efficiently. Qualcomm has a demonstrated history of success in mobile devices, and it is a safe bet that the company can facilitate the category’s success.
The Path to 6G
I expected 6G to be a hot topic at MWC this year, and it certainly was. We are three to four years away from the GSMA defining its 6G standards, but suffice it to say it, 6G will be highly software-defined and AI-infused. RAN infrastructure will need to support new spectral profiles, and refreshed hardware will be required, but the million-dollar (or euro) question is whether the core network infrastructure will need a refresh. With the continued disaggregation of telecom hardware and the use of general-purpose compute that meets necessary certifications for mobility workloads, 5G core infrastructure might only require a new software stack.
According to my research, 5G core infrastructure deployments still lag, with a 25-30% global penetration, as operators navigate what I characterize as a disastrous NSA-to-SA migration. I am sure the traditional infrastructure providers would disagree with my hypothesis, but it begs further scrutiny. What I like about this path to 6G is that it leverages existing and ongoing operator investments in 5G core. Furthermore, a new software layer over the 5G core complements the efforts of Ericsson’s Vonage business unit and the consortium it leads, as well as Nokia’s API exposure efforts, to facilitate developer innovation and the birth of new applications and mobile network services. Mobile network programmability, in combination with reduced capital spend, could dramatically improve ARPA and ARPU, with commensurate new monetization opportunities.
The Reimagination of OSS and BSS
Agentic AI clearly has the power to reimagine traditional operations support systems and business support systems for operators. OSS and BSS are essential in managing both network infrastructure and commercial operations. OSS focuses on provisioning and fault management, clearly a ripe use case for agent-to-agent interactions to deliver on the promise of autonomous networks. On the other hand, BSS functions that impact customer experience, such as ordering and billing, can also benefit from agentic workflow automation.
At MWC, I spent time with a host of OSS and BSS solution providers who extolled the virtues of integrating an agentic framework, including AWS. Claims of reducing provisioning and network infrastructure deployment times by 50% are powerful, and the same exists for improving customer interactions. I also validated those efficiency statistics with a handful of mobile network operator executives at MWC. Bottom line, the ability to boost operator efficiency, mitigate subscriber revenue leakage, and improve subscriber experiences is real.
Final Thoughts
Mobile network operators must find new ways to improve operational efficiency, delight subscribers, and monetize services beyond access. It is a dilemma that has existed for decades, but modern AI could become a force multiplier for the telecom industry. Its value has already been proven within organizations and enterprises of all sizes, and the potential to finally allow operators to move from telcos to techcos with AI is within reach. It will be interesting to see what unfolds and if the telecom industry rises to the challenge.
