WASHINGTON — The rapid growth of commercial Earth observation satellites and artificial intelligence tools is giving defense and intelligence agencies access to a new kind of intelligence product: fused data from multiple sensors designed to deliver insights about what’s happening in the world.
But turning that concept into something operational remains difficult.
At the Satellite Conference earlier this week, executives described a gap between what military users want — seamless integration of data from different sources — and how the commercial market actually operates.
“Processing and analysis now can be done almost instantaneously but the new challenge is now tasking and optimization,” said David Gauthier, chief strategy officer of GXO Inc. and a former National Geospatial Intelligence Agency official.
Defense agencies are increasingly focused on combining different sensing methods — optical imagery, radio-frequency signals, radar and other data — into a single operational picture, a concept known as sensor fusion. Closely tied to that is “tipping and cueing,” where one sensor detects an anomaly and directs another system to collect more detailed data on the same target.
Those workflows have long existed inside classified government systems. The shift now is toward using commercial data, as companies deploy large constellations capable of revisiting targets frequently and generating data across multiple sensing modes.
But the commercial ecosystem has not caught up.
There is a structural gap between what defense users want, seamless ‘multi-int fusion,’ and how the commercial market actually operates, Gauthier said.
Most companies operate vertically integrated platforms, collecting, processing and delivering data through proprietary systems. There is no widely adopted, standardized interface that allows different providers to exchange data or task each other’s satellites in real time. Differences in formats, metadata and latency mean that combining data from multiple vendors often requires manual integration or custom engineering.
Barrier not just technical, but economic
“There’s no incentive for us to build a tip and cue construct across the industry,” said Todd Probert, president of HawkEye 360’s government business.
Even when systems can be linked, timing remains a constraint. Gauthier described scenarios where one system detects a signal and cues another, only for delays in collection or downlink to break the chain.
“By the time those space-to-ground and ground-to-space delays have been introduced into your architecture, you’ve lost the opportunity,” he said.
Jared Newton, senior technology strategist at Planet Federal, said the challenge is not just connecting systems but how commercial constellations are designed and sold.
Actual tipping and cueing is “very difficult unless you have dedicated capacity or ubiquitous sensing,” Newton said.
Most commercial providers allocate satellite time in advance, guaranteeing customers a specific collection window. That model leaves little flexibility to respond dynamically when another system detects something worth investigating.
“We can pull it off every now and then,” Newton said. “But if we really were to do this seriously, we’d have to design a system that optimizes downlink, optimizes process, optimizes the access over the regions that a user cares about.”
In practice, that would require a different architecture, one that sets aside capacity specifically for rapid re-tasking and integrates collection, processing and delivery in a way that can respond in near real time.
‘Machine to machine’
What is ultimately needed, said Gauthier, is a system where sensors across companies can interact automatically.
“And what we want is to be able to task at machine speed, with machine to machine APIs,” he said. “If an RF collection sees something over here, then I want capacity in the next 30 minutes on somebody else’s satellite to take the image.”
That level of integration does not yet exist at scale.
In what he described as an ideal scenario, a military operator would see multiple data sources converge within minutes, automatically corroborating a potential threat. “That doesn’t exist yet. But that’s what I want,” Gauthier said.
The integration challenge comes as the market itself is shifting beyond raw data.
In a new report titled “Golden Insights: High-Quality Products Derived from Earth Observation Data,” published by the Center for Strategic and International Studies, Gauthier argues that the value of commercial geospatial data is increasingly tied not to collection, but to the ability to generate “decision-ready” insights from that data.
The report outlines how companies are moving up the value chain, offering analytic products that combine multiple data sources and are intended to support real-time decisions. But it also warns that the market for these products is still immature, with limited transparency around how they are generated or how to assess their quality.
The report breaks down product categories and metrics for evaluating commercial geospatial data products, “giving customers a bit of a buying guide for how to get to insights,” said Gauthier.