Seven years, $225M raised, and a L3Harris sensor deal later, a Spanish startup wants to sell AI companies Earth data 100x more precise than what exists today. The catch: the satellites that would deliver it don not yet exist.
image from Gemini Imagen 4
Spanish startup Xoople raised $130M Series B (total $225M, unicorn-valued) to build a satellite constellation delivering Earth observation data 100x superior to existing systems for AI training. The company currently has no satellites in orbit but has secured critical distribution by embedding its platform into Microsoft Azure and Esri's GIS ecosystem before its own data supply is operational. A partnership with L3Harris to build custom sensors positions commercial deployment for Q1 2026, but the company faces intense competition from Planet, BlackSky, and Maxar, which already operate constellations and hold multi-billion dollar NRO contracts.
A Spanish startup called Xoople has raised a $130 million Series B to launch a satellite constellation purpose-built for AI training data. The pitch: Earth observation data precise enough to let machine learning models see the planet at a resolution and freshness that commercial imagery can't match. CEO Fabrizio Pirondini told TechCrunch the sensors will collect "a stream of data that is going to be two orders of magnitude better than existing monitoring systems." That's a large claim from a company that doesn't yet have satellites in orbit and currently runs on publicly available data from Europe's Sentinel-2 spacecraft.
The funding, led by Nazca Capital with participation from MCH Private Equity, the Spanish government's CDTI, Buenavista Equity Partners, and Endeavor Catalyst, brings Xoople's total raised to $225 million since its founding in 2019. Pirondini declined to share a precise valuation but confirmed the company is "in unicorn territory." Xoople also announced a deal with L3Harris Technologies, the U.S. space and defense contractor, to build sensors for the spacecraft it plans to launch. Commercial deployment is expected this quarter after seven years of development.
The distribution strategy is the most concrete thing Xoople has going for it. The company has embedded its data platform into Microsoft and Esri, the two dominant platforms where enterprise, government, and GIS buyers already work. Aravind Ravichandran, CEO of Earth observation consultancy TerraWatch Space, told TechCrunch that Xoople built its distribution network before securing its own data supply — a bet that AI demand will pull customers to the platform before Xoople's own constellation is operational.
"Their distribution strategy is the only real asset they have right now," Ravichandran said. "Neither Microsoft nor Esri has proprietary Earth observation data. Xoople is trying to be the layer underneath both."
The competitive landscape is crowded. Planet, BlackSky, Maxar, and Airbus all operate satellites today and are developing AI-focused datasets. Planet runs the largest commercial constellation in orbit. BlackSky and Maxar hold multi-billion dollar contracts with the U.S. National Reconnaissance Office, the spy agency that in 2022 pumped billions into commercial imagery to reduce dependence on its own classified systems. Those contracts are the floor that Xoople needs to crack.
Google's AlphaEarth, published on July 30, 2025, is the benchmark Xoople will be measured against. Alphabet's AI model functions as a virtual satellite, integrating petabytes of existing Earth observation data to produce analysis at scale. It validates that major AI labs consider geospatial data a serious training substrate — but it also shows that Google, with essentially unlimited compute, is playing the same game Xoople is attempting at smaller scale.
The precision claim is where the story needs a caveat. Pirondini's "two orders of magnitude" characterization is a CEO quote from a press release context. No independent benchmark or third-party assessment of Xoople's data quality has been published. The L3Harris sensor deal is real; the satellites those sensors go into do not yet exist. Xoople has seven years of engineering work and $225 million in financing behind it, but the product that would justify the claim is not yet verifiable.
What's clear is that the market Xoople is targeting is real. AI developers want ground truth about the physical world, updated continuously, at resolution that public satellite data historically hasn't provided. Whether Xoople's architecture delivers on that promise — and whether its distribution-first strategy can survive once the established players move — is the question the next funding round will start to answer.
Story entered the newsroom
Assigned to reporter
Research completed — 3 sources registered. Xoople raised $130M Series B (total $225M). Founded 2019, CEO Fabrizio Pirondini (ex-DEIMOS IMAGING). L3Harris sensor deal for proprietary satellites.
Draft (581 words)
Reporter revised draft (574 words)
Reporter revised draft (574 words)
Reporter revised draft (3 words)
Approved for publication
Published (574 words)
Get the best frontier systems analysis delivered weekly. No spam, no fluff.
Artificial Intelligence · 3h 21m ago · 4 min read
Artificial Intelligence · 5h 20m ago · 3 min read
