AT&T frames network sensing as a platform, starting with drones

AT&T wants its mobile network to do more than carry traffic: it wants the radios to detect nearby objects and help machines understand their surroundings. Speaking at TIP FYUZ 2026 in Seattle, product development director Amy Gerrie set out the operator's ambition for integrated sensing and communication, or ISAC, and insisted the effort is bigger than a single feature.

"What we're creating is not a feature. It's a platform," Gerrie said, adding that she expects developers to build on top of it.

What has AT&T actually tested?

Gerrie described a trial run in the US state of Texas several months ago. Radio signals bounced off objects across several cell sites at once, and AI classified the returning signals so the system could detect, track and locate each object. Sensing comes first, she explained, with communications to be integrated later in the ISAC progression.

The test focused on drones, a deliberately hard target. Drones are small, fly low and move fast. They may try to avoid detection, and unlike phones, they are not registered on the network. Radar and cameras struggle to pick them up.

"It's the perfect use case because it's the hardest use case," Gerrie said.

When does a commercial platform arrive?

Joe Constantine, CTO of Ericsson Americas, shared the stage and the timeline. Ericsson will release its ISAC platform for mobile operators in 2027, with scale following in 2028. The vendor has built the product for 5G networks. Gerrie said AT&T is building to 5G standards as they exist today, with an eye towards 6G.

The rollout plan starts narrow and widens. AT&T will begin at defence and critical infrastructure sites before expanding nationally. The focus stays on the sky for now; sensing on the ground, among people, cars and buildings, will follow.

Who buys it?

Defence is already among AT&T's most important customers, Gerrie noted, and drones have changed the threats those customers face. She cited Ukraine, US borders, critical infrastructure and crowded venues such as stadiums as the settings that now demand better detection.

Drone delivery adds another driver. Amazon expects deliveries in nearly 500 cities and towns by the end of the year, and Gerrie said someone will need to know who is flying where and whether they should be there.

First responders form a second key use case. AT&T already provides their communications through FirstNet, and during a flood, Gerrie suggested, the same network might help locate people who need rescuing.

Enterprise customers will be a target market, but AT&T wants developers to build their own applications rather than the operator defining every use. Gerrie sees scope beyond enterprise as well.

Why use the mobile network for sensing?

Gerrie's argument rests on physics and economics. A self-driving car's lidar cannot see around a corner, but the network can. It can also detect objects that are not connected to it — the whole point of the drone trial. Network sensing would complement radar, cameras and other sensors rather than replace them.

The cost case matters as much as the capability. Sensing would reuse existing towers, power and spectrum, which changes the economics compared with building separate sensor sites. Gerrie added that investment in AI makes this the right time, since classifying bounced radio signals is what makes the technique practical.

Her forward view extends to robots, phones and autonomous systems using the network to understand what is around them, and longer term to digital twins of cities, crowd control and traffic management. The network would serve machines as well as people.

As for the timeline pressure, Gerrie was blunt about the drones and robots themselves: "They're coming, whether we want them to, whether we're ready for them to or not."

With Ericsson's platform due in 2027 and national rollout planned from defence and critical infrastructure sites outward, the next two years will show whether network sensing can move from Texas trials to a developer platform.