How Does Agentic AI Change the Search for Extraterrestrial Signals at the Allen Telescope Array?

What the Agents Actually Do at the Telescope

The Allen Telescope Array sits in Northern California. Scientists there have integrated the NVIDIA Holoscan platform into daily operations. Breakthrough Listen and the SETI Institute now run agentic AI as part of the standard search for extraterrestrial intelligence.

An agent can steer the instrument. It can enhance and correct incoming data. It can apply AI models to the incoming stream — all from a single natural language request. The result is a faster, more adaptable search for potential signals from extraterrestrial life and technologies.

Andrew Siemion chairs SETI research at the SETI Institute and leads Breakthrough Listen. He said the new agentic observatories may find things that have more in common with the agents themselves than with the humans who created them. He called it a breakthrough in translation rather than something to fear. Some scientists have suggested that extraterrestrial civilizations could be billions of years older than humanity. Such civilizations might have developed technologies far beyond humanity's, including forms of AI.

How the Data Pipeline Moves Terabits Per Second

Radio telescopes generate enormous volumes of data. Scientists need to identify potentially interesting signals as observations unfold. At the Allen Telescope Array, DAQIRI pulls terabits per second of sensor data straight into GPU memory. DAQIRI is a Holoscan platform library. Observations are processed in flight rather than filling hard drives.

CyberEther does the math using NVIDIA CUDA-X libraries including cuFFT and cuBLAS. CyberEther is a signal-processing framework developed at the SETI Institute. Nexus acts as the agent harness. Nexus is a web-based observatory orchestrator. It gives agents a single source of truth for instrument state and compute resources; pipelines are also tracked. Researchers can monitor packet loss. They can read live spectrograms. They can rewrite signal-processing blocks by prompting in natural language.

The system handles 2.4 terabits per second on NVIDIA RTX PRO 6000 Blackwell GPUs. Inference runs on premises, next to the instrument. An agent can steer a telescope's antennae. It can calibrate and focus the instrument for a clean signal. It can catch and fix data errors. It can plot the result. Agents compress what would take humans weeks or years into hours. This speed changes how researchers allocate observing time.

First Real-Time Detection and the Road Ahead

The pipeline has already proven itself. It achieved the first real-time AI detection of a pulsar in the Crab Nebula. A pulsar is a rapidly spinning stellar remnant that pulses like a cosmic lighthouse.

Next, the team is evaluating open models including NVIDIA Nemotron. These models matter for cases where data locality and turnaround matter most. They also matter for observatories that may someday run offline on the far side of the moon. How many such observatories will exist remains unknown. The announcement does not say when a far-side deployment might happen.