//php echo do_shortcode(‘[responsivevoice_button voice=”US English Male” buttontext=”Listen to Post”]’) ?>
The Mobile World Congress (MWC) 2026 closed last week, underscoring the telecom industry’s pivotal shift from legacy hardware to AI-driven infrastructure.
Marking its 20th year in Barcelona, the event drew 105,000 attendees from 207 nations, reinforcing its status as the leading global connectivity gathering despite ongoing travel disruptions in the Middle East.
The GSM Association (GSMA), the organizing body, designated this year’s theme as “The IQ Era.” The theme reflects a departure from the industry’s historical focus on bandwidth expansion, shifting attention to dynamically programmable networks and the impending 6G standard.
In an opening address, GSMA director general Vivek Badrinath articulated the sector’s current mandate. “If we want to unlock the full promise of 5G, harness AI responsibly, and protect people from escalating digital threats, we must act with urgency,” Badrinath stated.
By Bluetooth SIG 03.10.2026
By GigaDevice 03.10.2026
By GigaDevice 03.10.2026
John Hoffman, CEO of GSMA, reflected on the location’s role over the past two decades, noting in a statement that “Barcelona had the infrastructure, the ecosystem and the commitment to grow with us.”
Hardware infrastructure divide
Reporting from the exhibition floor, EE Times detailed a dividing line regarding the physical foundations for AI and next-generation networks.
Telecommunications operators are debating whether to process AI workloads using distributed graphics processing units (GPUs) or through central processing units (CPUs) that feature integrated AI acceleration. This architectural choice carries significant implications for capital expenditures, power grid stress, and equipment deployment at the network edge.
Nvidia and Nokia advance distributed GPUs
Nvidia has aggressively pursued the telecommunications market, advocating for GPUs as the standard for radio access network (RAN) operations. The company presented its Aerial RAN Computer (ARC), promoting the network as a computational platform that concurrently handles standard telecommunications functions alongside external AI applications.
Ronnie Vasishta, SVP of telecom at Nvidia, told EE Times that the company views cellular access points as the future connectivity fabric for AI inference. Vasishta argued that operators must move away from fixed hardware systems. “It’s like you don’t wait for a TV because the next model is better… to watch TV,” Vasishta said. “Buy something that can be software upgradable, and that’s what the AI native platform is.” To support this, Nvidia released an open-source telecommunications model based on its Nemotron 3 architecture to manage network fault detection.
Nokia aligned its strategy with Nvidia, supported by a $1 billion investment. CEO Justin Hotard argued legacy silicon no longer suffices.
“AI is the new workload reshaping networks. That shift requires architectural change across every layer, including the radio,” Hotard said during a pre-MWC media event. By adopting the computing architecture used in data centers, Nokia aims to separate hardware from software to process complex workloads at the radio edge.
Ericsson and Arm target CPU efficiency
In contrast, Ericsson, Intel, and Arm maintain that CPU-based architecture is more practical for telecommunications. This contingent asserts that deploying GPUs for edge inference adds unwarranted complexity and risks overloading the utility power limits of cell sites.
Arm positioned its Neoverse computing platform as a power-efficient alternative for operators integrating AI. Eddie Ramírez, VP of Arm’s infrastructure business, told EE Times that energy limitations strictly dictate facility capacity. “When you build a data center, the amount of compute is dictated by power,” Ramírez stated. He observed that operators are pivoting from GPU rental to “inference as a service,” seeking air-cooled hardware for edge networks.
Ramírez also noted that autonomous AI agents will fundamentally alter data patterns. “What is going to change is when it’s not you making the queries, it’s an AI agent,” he explained, warning operators to prepare for sustained data traffic.
Ericsson maintains software independence by utilizing purpose-built silicon, a decision intended to avoid vendor lock-in with GPU providers. In an EE Times interview, Marie Hogan, head of 6G portfolio strategy at Ericsson, described a mixed approach to computing distribution.
“Some models that require massive compute will likely be centralized if they don’t have strict timing pressure. Smaller models and compute may be localized and integrated into radio or base station hardware,” Hogan said.
Ericsson’s CTO, Erik Ekudden, emphasized the need for immediate deployment. “There is no need to wait for 6G when it comes to AI. AI is here, both in the network and as a business opportunity,” he noted.
Qualcomm focuses on network sensing
During MWC 2026, Qualcomm outlined a 6G vision that expands the network into an environmental sensing system capable of spatial mapping. By integrating sensing and communications capabilities, the network can transmit information while operating as a radar system to detect objects.
Cristiano Amon, president and CEO of Qualcomm, described the long-term impact of this technology. “6G is more than the next step in wireless evolution. It is the foundation for an AI-native future that distributes intelligence across devices, the edge, and the cloud, and transforms network providers into AI-driven enterprises,” Amon stated.
John Smee, Qualcomm’s global head of wireless engineering, cautioned against relying solely on centralized computing in a briefing with EE Times. “We think it’s very important that there’s a holistic approach to energy consumption and to be part of the value creation,” Smee said.
Qualcomm proposes distributing AI tasks across wearables, the network edge, and data centers to extend battery life for consumer devices such as smart glasses.
Robotics, physical AI, and monetization
During the show, companies conducted live demonstrations of robotics and physical AI. Nokia and Nvidia demonstrated this capability by operating a robotic dog via an AI-powered 5G connection with Indosat Ooredoo Hutchison in Indonesia.
Operators are aggressively targeting robotics as a new profit engine for current infrastructure. T-Mobile US unveiled ‘Kinetic Tokens’ to convert physical AI movements directly into revenue.
The GSMA also reported commercial progress on its Open Gateway initiative, a series of application programming interfaces that enable enterprise developers to set network performance parameters.
Providers now use these interfaces to secure drones and drive industrial automation, leaving behind routine consumer broadband. A European operator consortium also announced the ‘Edge Continuum’—a federated edge computing network for autonomous vehicles that cross borders.
Preparing for the 6G transition
The technical debates at the 2026 Mobile World Congress established the foundation for 6G standardizations, which the 3rd Generation Partnership Project (3GPP) is currently formalizing for an expected commercial release in 2029.
Hardware executives insist the shift to software-defined networks must precede the standard’s completion to meet current data needs.
As utility costs climb, the future of telecom hinges on executing AI within strict energy bounds. Strategies revealed in Barcelona will define the next decade of network economics and connectivity.
See also:
Nvidia Advances AI-Native Strategy at MWC
At MWC, Intel Outlines CPU-Driven Case for Telecom Networks