Imagine satellites that can watch, interpret and respond in real time. Firefighters get immediate notification as wildfires ignite along a remote woodland border. Engineers in a bustling city are warned to redirect traffic as a violent summer storm approaches. A ship is alerted as ice drifts into shipping lanes. Seconds and minutes can be critical for response times, but for decades such events slipped by unseen, buried in mountains of archived satellite data and discovered too late to make a difference.
Today, society stands at the threshold of a radical transformation: an era of artificial intelligence (AI) where satellites aren’t passive eyes in the sky but dynamic partners to help us see and understand events as they unfold. Among those at the heart of this revolution are two pioneering technologies — Ubotica’s Space:AI Layer and NASA’s Dynamic Targeting — each reshaping our ability to observe, analyze and act on events as they occur. The term “Space:AI Layer” refers to two emerging concepts in AI and technology. One is a physical intelligence layer for environments where fused sensor data is used to allow physical AI tools (robots, autonomous vehicles) to perceive and cooperate with their surroundings. A second AI-enabled layer is on orbit, composed of data centers and communications infrastructure to provide computing power and connectivity from space.
Intelligent orbital networks
Under the old paradigm, Earth observation was simple: Satellites mapped the globe in preset patterns, regardless of whether what they captured held value. Every day, fleets of spacecraft amassed petabytes of imagery, much of it rendered useless by clouds or empty landscapes — or containing no events of interest at all.
NASA’s Jet Propulsion Laboratory (JPL) and Ireland-based Ubotica have developed a technology backed by AI that shifts that dynamic. Instead of using satellites to simply collect raw data, they would interpret it, share insights and adapt their missions to what matters most on the ground. JPL’s vision, called Dynamic Targeting, builds on this idea, imagining constellations of intelligent spacecraft functioning like a planetary nervous system, making autonomous decisions and collaborating as a space-based sensing network.
“The idea is to make the spacecraft act more like a human: Instead of just seeing data, it’s thinking about what the data shows and how to respond,” Steve Chien, a technical fellow in AI at JPL and principal investigator for the Dynamic Targeting project, said on NASA’s website. “When a human sees a picture of trees burning, they understand it may indicate a forest fire, not just a collection of red and orange pixels. We’re trying to make the spacecraft have the ability to say, ‘That’s a fire,’ and then focus its sensors on the fire.”
How it works
The Space:AI Layer enables satellites to process data onboard to shrink the time between observation and understanding. Instead of sending terabytes of raw imagery to Earth for slow analysis, the satellites extract key insights on orbit and deliver actionable intelligence within minutes.
Primary instrument
observation
Look-ahead sensor:
current observation
Look-ahead sensor:
previous observations
Primary
instrument range
Nadir
At the core of this capability is Dynamic Targeting, co-developed by JPL and Ubotica. Breaking away from static imaging routines, satellites equipped with Dynamic Targeting use “look-ahead imaging” to scan areas 40 to 60 degrees ahead of their orbital path — giving themselves a preview of what’s coming up, Ubotica’s website explains. In just five seconds, onboard AI analyzes the images for anomalies, identifying clouds, thermal hot spots or other events of interest. If something valuable is detected — the glow of a wildfire or a suspicious vessel at sea — the satellite can autonomously reorient itself, adjust its sensors and focus on the emerging threat or opportunity.
This process unfolds in less than 90 seconds. Commands are generated, sensors are reconfigured and high-value data is captured and processed — all before the event slips out of view.
CogniSAT’s mission
The leap from concept to reality occurred in July 2025, when Ubotica and JPL tested Dynamic Targeting aboard a CogniSAT-6 CubeSat. The satellite — the size of a suitcase and designed and built by United Kingdom-based Open Cosmos — hosted a payload developed by Ubotica featuring a commercially available AI processor. This wasn’t a simulation. It was a live mission, with the satellite processing real data and making real-time decisions on orbit.
During the demonstration, CogniSAT-6 didn’t hunt for specific phenomena like fires. That will come later. Instead, its mission focused on an ever-present phenomenon — clouds. Most science instruments on orbiting spacecraft look down at whatever is beneath them. However, for Earth-observing satellites with optical sensors, clouds can get in the way as much as two-thirds of the time, blocking views of the surface. To overcome this, Dynamic Targeting looks 500 kilometers ahead and has the ability to distinguish between clouds and clear sky, Ubotica said. If the scene is clear, the spacecraft captures images of the surface when passing overhead. If it’s cloudy, the spacecraft cancels the imaging activity to save data storage for another target.
The impact of intelligent satellites
The potential of Space:AI and Dynamic Targeting is immense. In disaster response, the ability to detect and analyze events like wildfires or hurricanes in real time means emergency teams can work from annotated maps highlighting collapsed infrastructure and isolated populations within minutes, not days, helping focus resources where most needed.
In agriculture, satellites would differentiate between normal seasonal variations and early warning signs of drought, pest outbreaks or disease. Farmers and policymakers would be empowered to make decisions proactively, mitigating crises before they escalate.
Environmental protection could enter a new era. Instead of combing through endless archives for evidence, authorities would reference timely, irrefutable intelligence on illegal mining, logging or fishing operations and track the events as they happen, allowing for effective enforcement and conservation.
Maritime security also would benefit. Dark vessel detection enhanced with AI-driven Dynamic Targeting would allow satellites to scan vast ocean expanses, focusing on suspicious ships approaching critical infrastructure instead of wasting time imaging empty water.
What’s next
With the cloud-avoidance capability now proven, NASA said CogniSAT-6’s next test will be to hunt for severe weather — essentially targeting clouds instead of avoiding them. Other potential targets are thermal anomalies like wildfires and volcanic eruptions. The JPL team has developed unique algorithms for each application.
“This initial deployment of Dynamic Targeting is a hugely important step,” Chien said. “The end goal is operational use on a science mission, making for a very agile instrument taking novel measurements.”
There are several approaches to realizing this goal, including deployment on spacecraft engaged in space exploration. Chien and his colleagues at JPL drew inspiration from their prior work with the European Space Agency’s Rosetta orbiter, where they demonstrated autonomous detection and imaging of plumes emanating from comet 67P/Churyumov-Gerasimenko.
On Earth, adapting Dynamic Targeting for radar applications would facilitate the study of hazardous deep convective ice storms, extreme winter weather events that are infrequent and transient — making them challenging to observe with current technologies, NASA said.
By employing specialized algorithms, satellites equipped with look-ahead instruments would identify these dense storm systems, enabling focused radar observation as the spacecraft passes overhead, collecting comprehensive data during brief intervals of six to eight minutes.
Additional concepts include implementing Dynamic Targeting across multiple spacecraft. Image analysis results from a leading satellite could be promptly transmitted to a trailing satellite, which then would concentrate on specific phenomena. This approach could extend to constellations comprising dozens of orbiting spacecraft. A test of this strategy, named Federated Autonomous Measurement, occurred in late 2025.