Spica Space
EO / Boulder
Mission nominal
Mission
Spica Space was founded on a single premise: that a satellite pass is only useful when it arrives as a decision, not as an image. Our founders spent a decade building imaging payloads at Planet Labs and Airbus Defence and Space, and watched the industry ship more pixels to fewer operators every year. The pixels are not the product. The verified event is the product — the illegal dredging operation, the methane plume, the levee breach, the crop loss that becomes a claim.
Everything we have built since 2018 has been aimed at shrinking the time between acquisition and action to under ninety minutes, anywhere on the planet's land surface. We operate one of the densest commercial constellations dedicated to sub-meter revisit. Every pass is filtered through a proprietary AI stack before it reaches a customer. Operators get change-detection they can defend in a courtroom, an underwriting meeting, or a boardroom — and they get it on the same day the satellite flies overhead.
— The Spica Space founding team, Boulder, 2018
Origin
Three engineers who had spent the previous decade inside the two largest commercial imaging programs in Europe and North America left in 2018 to build the constellation they wished they had been allowed to operate.
2018 · Founding
Spica Space was founded in 2018 by three engineers who had previously led imaging payload programs at Planet Labs and Airbus Defence and Space. They had a shared view: the commercial Earth-observation industry was optimizing for pixel throughput when operators actually needed verified events. The company was incorporated in Delaware with its first office in Boulder, Colorado.
Three offices · One team
Headquartered in Boulder, Colorado, with engineering offices in Toulouse, France and Singapore. The team of 142 includes 41 PhDs in remote sensing, computer vision, and atmospheric science. Mission operations run 24 hours a day across the three time zones, which is also why the median tasking-to-delivery latency holds at 78 minutes anywhere on the land surface.
Capital · 2024
In March 2024 the company closed an $84 million Series B led by Lux Capital, with participation from Founders Fund and the International Finance Corporation. Capital is allocated to constellation expansion, onboard AI inference, and the open-source Spica-Change PyTorch library, which has been cited in more than 140 peer-reviewed papers since 2023.
By the numbers
PhDs across remote sensing, computer vision, and atmospheric science on a 142-person team.
Led by Lux Capital, with Founders Fund and the IFC. Applied to constellation, AI stack, and open source.
Operating in agriculture, energy, maritime, insurance, and government. 128% net revenue retention reported in 2024.
Supported by a 24/7 mission operations center. Median tasking-to-delivery latency: 78 minutes.
Credentials
The sole Earth-observation provider to hold FedRAMP Moderate authorization as of November 2024, alongside SOC 2 Type II and ISO 27001 certifications. Compliance documentation is available under NDA on request, with public summaries published to the platform trust center.
Read about platform compliance →Named Earth Observation Innovator of the Year at the 2024 Geospatial World Awards, selected ahead of three larger incumbents. The jury cited the onboard inference architecture and the open-sourcing of the Spica-Change library as the deciding contributions to the field.
Independently benchmarked by the Open Geospatial Consortium in Q1 2025. Median tasking-to-delivery latency of 78 minutes. Change-detection accuracy of 97.3% on the SpaceNet-7 flood and building-damage benchmark, outperforming the next-best commercial model by 11.6 points. The full methodology and dataset splits are linked from the platform page.
See the full benchmark methodology →What we believe
We hold three beliefs tightly enough that we will lose deals over them. The first is that pixels are not the product — verified events are, which is why every tile that leaves our ground station has been through change-detection before downlink. The second is that latency is a feature on the same axis as accuracy; a perfect map of last week's flood is less useful than a good map of today's. The third is that our customer is the operator holding the radio or the underwriting file, not the dashboard they happen to be looking at. We design backwards from that person.
We publish the Spica-Change library because the field moves faster when the underlying science is open. We partnered with Microsoft Planetary Computer to release 3.8 million km² of tropical-forest baseline, with ESA's Phi-Sat program and NASA JPL on on-orbit foundation models, because the next decade of Earth observation will be written by people who can inspect the tools. We would rather you understand exactly what we are selling than feel impressed by it.
Request a demo
Thirty minutes, screen-shared. We will walk through the same tasking, change-detection, and export tooling used by 380+ organizations, using a recent acquisition over a region you choose. No slideware.