In an AI-native architecture, AI is not a feature — it’s fundamental
While the standardization of 6G is still years away, the next-generation of cellular technology is often described as AI-native, and as telcos and their partners look to the future, they are rallying around the promise that soon, AI will be used throughout their operations, enabling things like internal optimizations and customer-facing use cases and generating new service revenues.
But what does it mean to be AI-native? After speaking with the companies investing in and facilitating this transition, it quickly became clear what an AI-native approach is not — it’s not an add-on, an afterthought, or secondary. It’s not merely a feature or a tool. Instead, it describes an architecture in which AI is “fundamental,” and “core” and something that’s there from “end-to-end.”
On a more technical level — and in an ideal world — AI-native also means that telcos will begin to see significant productivity gains enabled by cloud computing, large language models (LLMs), natural language processing-based user interfaces and generative text/image generation capabilities.
AI for optimization
According to Blue Planet VP’s Kailem Anderson, the benefits of AI will touch every aspect of the 6G environment. “It will be built into the chipsets, the hardware protocols, the software stack and various abstraction layers so that the network is truly intelligent,” he told RCR Wireless. “What does that mean? It means 6G will truly embrace principles around self-healing, self-optimizing, self-organizing, so that the network operates in a truly declarative or intent-driven way.”
But it’s just about optimization, clarified Ronnie Vasishta, who is responsible for the telecom business, strategy and products at Nvidia; it’s also about reoptimizing. “AI-native also means that you’re creating optimizations for the environment you’re in,” he said, adding that things like signal reflections off buildings and weather patterns mean the network metrics are always changing, and therefore, the network must be constantly “reoptimized” for the best performance. More connections are expected in a 6G world and the denser the network, the more important these continuous network adjustments become. As a result, Vasishta believes that AI is “the one tool” available to support the needs of 6G under current spectrum and power conditions.
Anderson explained that in 6G, AI will also be leveraged for network planning, fulfillment and assurance in closed-loop way, commenting that in 5G, for instance, AI is practically nonexistent in the telco planning and fulfillment cycle.
AI for monetization
The concept of network slicing — or providing individual chucks of network for specific users or applications — was introduced in 5G and, for the most part, has remained just that: a concept. It’s certainly done today, but as Anderson explained, it’s been a pretty “clunky” introduction. He’s confident though that in 6G, network slicing will be possible on a “very granular” level, where the network can be sliced down to an individual user level. And more importantly, the monetization of this capability will “get figured out in 6G.” Largely, this will be a function of the network disaggregation established in 5G as this will allow telcos to drive cost and operation efficiency out of the RAN environment, he said.
In addition, AI-RAN Alliance’s AI-and-RAN working group is exploring how using the same infrastructure to run both RAN workload and also AI workload simultaneously will open up new revenue streams for telcos. “This is about making the most of what we have … to support AI applications while still managing our networks core functions,” Alex Jinsung Choi, chair of the AI-RAN Alliance and principal fellow of SoftBank Corp.’s Research Institute of Advanced Technology told RCR Wireless News. “The outcome from this group is to show us how to increase resource utilization and open up new revenue streams by hosting various AI applications on the same platforms that runs our network functions.”
Achieving this AI-native vision — where AI is end-to-end, where it’s behind the scenes enhancing the network and at the forefront being sold as a product — is a team effort: “Vendors in the automation space to help stitch this together to leverage the various embedded AI capabilities that span across the hardware and the chipsets and various parts of the stack to bring that together,” Anderson said.
