For several years, the space-based geospatial intelligence industry has been chasing a logical vision for AI: use it to make our existing systems faster and smarter. Train models to detect objects. Automate change detection. Speed up analysis.
These capabilities have delivered operational benefits. But they’ve also kept us focused on a specific paradigm — collect an image from space, process it, apply AI and deliver insights via dedicated access platforms. We’ve built impressive AI tools on top of a traditional architecture that was designed before the onset of our increasingly AI-powered world.
In short, we’ve been using AI to improve the value of our data, when the real opportunity is to use our data to improve the value of AI. I’m talking about giving AI spatial intelligence.
Spatial intelligence: the future of AI
The full potential of AI for GEOINT is to use it to understand where things are, how places connect and what’s happening across the physical world.
AI shouldn’t just identify adversarial threats by labeling and counting them. It must also be able to accurately predict what those threats are doing, identify where they’re headed next and provide recommendations on how to respond. In other words, AI should not only tell me where a carrier is docked, it should also be able to draw on a historical archive of data to tell me what it means when that carrier is at those exact coordinates and to identify all the possible threat scenarios to come.
For AI to do that, it needs spatial intelligence: the ability to segment the world with pixel-level accuracy in a sensor-agnostic fashion, allowing it to instantly process a current and historical reference of objects and patterns.
It needs a living, continuously updated digital twin of Earth — a computable Earth. Not disconnected data layers, but an integrated, real-time spatial model that pulls from electro-optical imagery, synthetic aperture radar (SAR) imagery and even drone imagery to provide accurate context at global scale.
The pressure for spatially intelligent AI will grow in 2026
The pressure to usher in spatially intelligent AI will reach a tipping point in 2026.
First, governments will continue to invest billions in sovereign defense and intelligence capabilities — and expect immediate returns. From Europe’s re-armament to modernization across the Middle East and Asia, we’ll continue to see heavy investment in sovereign capabilities. The technology they’re building from scratch will take years to deploy, leaving these organizations drowning in data and starving for insights in the meantime. They are investing in commercial intelligence capabilities that can be brought online today. Vantor has seen that first-hand, with double digit growth across our international business in 2025. But expectations are high, and many of these systems are inherently disconnected. These customers will expect fast and meaningful improvements in unified intelligence and near real-time insights.
Second, the AI industry will continue to roll out innovative models that set higher expectations. Foundation models trained on satellite imagery will keep improving. Some are already astonishingly good. These releases will demonstrate impressive capabilities, including automated feature extraction, change detection and spatial reasoning. The problem: Most models will operate in isolation, optimized for specific tasks rather than integrated spatial understanding. They’ll excel at demos and benchmarks but struggle to deliver the contextual and predictive insights AI promises. The gap between hype and reality will become impossible to ignore, and the pressure to fuel more powerful AI will grow.
Third, the AI industry itself will attempt to solve the spatial intelligence problem. Spatial intelligence is already a hot concept in the broader AI world. Companies such as NVIDIA and World Labs have been focused on building world models that can navigate real-world 3D environments and build hyper-realistic synthetic environments from scratch.
These models still need an accurate digital representation of our world. We can build the abstraction layer between sensors and models, bridging the physical and digital worlds.
How GEOINT solves the spatial intelligence problem
Space-based GEOINT can provide AI with what no other industry can: persistent, global coverage at the scale and consistency needed to serve as Earth’s authoritative baseline.
We can fuse that foundation with intelligence across every domain — from space, air and ground — to create a unified, spatially accurate world model that can empower AI like never before. To build this AI-ready foundation, we must work together as an industry to:
Architect for multi-domain fusion from the start. The most promising developments today combine satellite optical imagery with SAR, visual positioning systems and ground truth data. We need to build integrated tasking and processing systems that make it easier to collect, organize and process data. Vantor has been working toward that through our virtual constellation with partners such as Umbra and Satellogic. More of that across the industry is critical.
Establish common standards. Initiatives like the Overture Maps Foundation show how industry can align around protocols that make integration seamless and make it easier to fuse different datasets. We need to accelerate this, making interoperability the default, not the exception.
Build easily deployable software. A unified digital twin only delivers value if it’s easily accessible to AI systems in classified networks, commercial clouds and the tactical edge. You want to build the computable Earth once and then deploy it everywhere it needs to go, not recreate it for every individual system.
We are facing a generational opportunity. The GEOINT industry has spent decades perfecting observation, and we’ve deployed AI to achieve that same goal.
Now we have the chance to reimagine our view of AI and focus on solving for an even bigger goal: delivering the spatial intelligence infrastructure that powers the next generation of AI to transform our industry and the world.
Let’s build it.
Dan Smoot is CEO of Vantor.
This article first appeared in the January 2026 issue of SpaceNews Magazine.
SpaceNews is committed to publishing our community’s diverse perspectives. Whether you’re an academic, executive, engineer or even just a concerned citizen of the cosmos, send your arguments and viewpoints to opinion (at) spacenews.com to be considered for publication online or in our next magazine. If you have something to submit, read some of our recent opinion articles and our submission guidelines to get a sense of what we’re looking for. The perspectives shared in these opinion articles are solely those of the authors and do not necessarily represent their employers or professional affiliations.
