AI native networks represent a fundamental structural shift in how global telecommunications infrastructure must be designed, Nokia President and CEO Justin Hotard said at MWC26 on March 4, 2026. Hotard argued that existing network topologies are inadequate to handle the demands of artificial intelligence traffic.
Hotard described traditional network traffic — voice, data, and video — as predictable and linear. He contrasted this with AI-driven traffic, which he characterized as fundamentally different in behavior and structure.
Token-Driven Traffic Replaces Stream-Driven Models
“The traffic we’re modelling in networks today, and the traffic we’re capturing out of networks today is completely different. It’s bursty, it’s dynamic and sometimes it’s uplink constrained,” Hotard said. He noted that chatbot engagements and machine-to-machine (M2M) communications are driving this shift, moving away from stream-driven models toward token-driven traffic patterns.
“It’s not about voice, it’s not about data, it’s not about video, it’s about something new, which is the token,” Hotard stated. This change in the fundamental unit of telecommunications traffic has direct implications for how networks must be engineered going forward.
Deterministic Connectivity as the New Standard
Hotard introduced the concept of “deterministic connectivity” as the target architecture for next-generation networks. He defined this as delivering the right latency, performance, quality, security, and trust to ensure individual devices receive the data tokens they need to execute actions and decisions.
“With the right latency, the right performance, the right quality, the right security and ultimately the right level of trust we have what we call deterministic connectivity. This is a structural shift.”
Justin Hotard, President and CEO, Nokia
AI Native Networks and the End of the Silo Model
Hotard stated that AI native networks must carry the performance, capability, and bandwidth to support video, data, and voice — but that this requires breaking the existing silo model of network design. He described the current approach as incompatible with the architectural demands of AI workloads.
“The concept of scale up to scale out to scale across has accelerated and ultimately the unit of networking is changing to meet the demands of the workload,” he said. The remarks reflect a broader industry debate about whether current telecom infrastructure can support the exponential growth of AI-generated traffic without significant redesign.
Implications for Network Design and Investment
The Nokia CEO’s comments carry significant weight for network operators and equipment vendors globally. As AI adoption accelerates across industries, the pressure on telecommunications infrastructure to evolve is intensifying. Hotard’s keynote at MWC26 signals that Nokia views this architectural transition as both urgent and unavoidable.
Moreover, the shift toward token-driven traffic models could require operators to invest in new hardware, software, and network management systems. Consequently, the timeline and cost of this transition remain key questions for the industry as AI workloads continue to scale globally.