AI-Native 6G: Investing In The "Sense-And-Inference" Telecom Network

Photo by Nastya Dulhiier on Unsplash

When designs for 5G began, it was all about speed. How fast could the network carry data traffic? However, while 6G will be the largest infrastructure buildout in human history, according to Nvidia founder and CEO Jensen Huang, it won’t just be a fatter pipe. Tomorrow’s 6G mobile networks will be built on AI-native architecture that will integrate sensing and communication into the mobile telecom network. And investors want to know what the return will be on AI-native 6G — and whether telecom companies can afford the massive capex involved.

AI: From 5G to the 6G Mobile Network

With 5G, we’re already getting a glimpse of what an AI-integrated mobile telecom network will look like, but AI is simply bolted onto existing 5G architectures and networks. When 5G was designed, AI usage to the degree expected today wasn’t even in the picture. With 6G though, AI will become a native part of the infrastructure, built right in from the very beginning.

The 5G networks that span the globe today utilize self-optimizing network technology, which actually dates back to 4G LTE. However, according to the Wireless Infrastructure Association, a key goal for 6G is fully closed-loop, real-time network automation. In other words, AI agents will observe the behavior of networks on a continuous basis.

They will automatically make decisions and complete enforcement actions, and very minimal human intervention is needed for performance and turning. As a result, the radio access network (RAN) will evolve from a passive layer capable of transmitting and receiving data between end-user devices into a “context-aware system driven by AI.”

A hallmark of this context-aware system will be integrating sensing and communication (ISAC), which enables the entire network to act like a sensor. In other words, it will be able to detect movement, drones, or traffic patterns without the use of a camera. To do this, the network will tap radio signals and multimodal fusion, with 6G intelligent networks supporting totally new classes of business services, digital twins and other new use cases, physical AI apps, and much more.

The Return on Intelligence

Investors already know there’s money to be made in intelligence and the 6G revolution, but what’s unclear is the size of the returns we can expect on such investments. One analysis from Nvidia and Softbank indicates that telco operators can earn about $5 in AI inference revenue (see below) for every $1 invested in new AI-RAN infrastructure over five years. Thus, while investors are largely focused on picks-and-shovels plays for AI, I would argue that they should be looking at telecom operators due to this significant opportunity.

For example, while AT&T shares are off a bit from recent peaks, the P/E multiple has plummeted 47%, according to data from TradingView. Verizon is doing better, with its P/E multiple up 4%, but it hardly reflects the opportunities related to AI-native 6G.

For accredited investors set on the picks-and-shovels play, the private markets are filled with players that provide equipment to operators like AT&T and Verizon.

Can Telcom Companies Really Afford Another Enormous Capex Cycle?

It’s understandable that investors are skeptical about AI-native 6G because many think operators can’t afford another capex cycle of the size needed to roll out AI-native 6G. After all, McKinsey reports that telecom companies’ revenue growth has been essentially stagnant over the last 10 years. However, the firm also outlines four different opportunities telecom companies have to monetize AI-native 6G, supporting their build-out.

1. Laying fiber to connect all those new data centers – Fiber connections will be needed between the tens of thousands of data centers expected to be built in order to scale AI workloads. McKinsey’s research suggests a $30 billion to $50 billion global market opportunity for this fiber buildout and for leasing or buying existing fiber infrastructure from telcos and other providers.

2. Providing intelligent network services to support access to the high-performance cloud – Enterprise workload needs will become more and more complex, meaning they’ll need intelligent network services that provide more flexibility and control in the way they manage that network. McKinsey estimates that AI inferencing will make up most AI workloads by 2030, versus only 15% to 30% in 2023.

3. Transforming unused space and power into revenue – For example, Verizon’s mobile-edge computing partnership with Amazon Web Services utilizes a hybrid strategy, enabling businesses to secure compute power near to the end user for real-time inferencing.

4. Constructing a new GPU-as-a-service division – Companies can use GPUaaS offerings to secure remote access to high-performance GPUs inside data centers without an expensive upfront investment. McKinsey’s analysis indicates that the total addressable market for GPUaaS for telecom companies could be between $35 billion and $70 billion by 2030.

Approaching the telecom business from an investor’s standpoint requires a whole new perspective on the industry. New revenue opportunities look set to support all that capex they’ll have to pour into the AI-native 6G infrastructure, and they create new long-term revenue streams for telcos.  

Depending on the source, predictions for 6G call for it to start being rolled out in the early 2030s, so it will be here sooner than you think. Is your investment portfolio ready?