With 6G networks of the future, every millisecond will matter. 

Mobile networks are likely to be under pressure to meet the growing demands of multimodal applications as well as expanding requirements of AI over the coming years. Latency plays a key role in meeting these new demands and as operators migrate from non-standalone 5G (NSA) to standalone 5G (SA) and 5G advanced, they are engineering their networks to make them more responsive.

During a panel hosted by Ookla at Mobile World Congress in Barcelona on March 3, Luke Kehoe, industry analyst at Ookla, moderated a discussion on latency and network responsiveness featuring Ankur Kapoor, EVP and  chief network officer at T-Mobile US and Tibor Rathonyi, senior advisor at Ookla. 

Kapoor noted that because NSA networks still rely on a 4G core, these two network technologies have to talk back-and-forth with each other, which keeps latency on the high side. Because SA networks eliminate that extra communications, latency becomes much more consistent from the device to the network core. Add Low Latency, Low Loss, Scalable Throughput  (L4S) to 5G SA and you have a network that is much more amenable to latency-sensitive applications such as video calls and gaming.

T-Mobile launched the first nationwide SA network in the U.S. in 2020 and then expanded SA to its 2.5 GHz spectrum in late 2022. The company launched nationwide 5G-Advanced in April 2025 and added L4S capability  in July 2025.

“You want to have is consistent latency. You don’t want to have any peaks and valleys,” Kapoor said. 

Rathonyi noted that Ookla’s network measurement tools are able to measure latency, both at idle and under load, and the quality of experience for different types of services such as gaming and web browsing. 

“We see in our data that latency has a strong correlation to quality of experience,” Rathonyl said, adding that once networks get to single-digital latencies, it’s nearly an instantaneous experience for the end user. 

However, this is just the beginning. Kapoor noted that he believes that 6G will be an “architectural shift” for wireless networks and will have to be AI native. “That’s where the industry is headed,” he noted. 

“Training Wheels for 6G”

In fact, Kapoor describes the SA networks of today as the “training wheels” for the 6G networks of the future because 6G networks will not just be processing bits and bytes like 4G and 5G networks but will act as the “connective tissue” for physical AI.  

What does he mean by “connective tissue?”  Kapoor said he believes 6G will be processing “tokens.” In the AI world, a token is a basic unit of information that a model processes. AI models break down data into chunks, or tokens. 

And when networks start processing AI tokens, uplink and latency become more important, Kapoor said. “Now every millisecond matters,” he said.