Skip to content
Setcoin Group
Request access

Setcoin Intelligence · Space Infrastructure · Earth Observation · Updated October 2026

Earth Observation and AI Analytics: What a Square Kilometre Is Worth

What is satellite data worth per square kilometre and per revisit, who pays for it, and does AI change who captures the value? Ninety-eight chapters on sensors, fleets, prices, government and commercial buyers, the operators, constellation economics, foundation models, sovereign deals, regulation and risk, built on company filings, contracts, public price lists and Setcoin's satellite catalogue analysis.

Open to everyone: the first 3 chapters in full, with exhibits and models; a free account adds their Excel models. Space subscription: all 101 chapters, every model and Excel file, 7-day free trial, then €199 a month. Already a reader? Sign in.

revisitevery ~4 h16 agile satellites
Chapters
98 + 3 appendices · 180 min
Interactive Models
5, with Excel
Data
Filings, contracts, price lists, satellite catalogue to 20 Oct 2026
Figures As Of
22 October 2026
$6/km²
SkySat archive list price
57
ICEYE SAR satellites
~2/3
EO bought by governments
180 min
Read, 98 chapters

In short

Satellite imagery is getting cheap and the money is moving elsewhere. Planet lists SkySat archive at $6 per square kilometre and Satellogic at $4; constellations collect far more than they sell. Value sits in guaranteed access (assured tasking at $40 and sovereign capacity deals) and in answers sold per asset or decision. Governments buy about two-thirds, and growth has moved from US intelligence, whose commercial imagery budget was cut in the 2026 request, to European and Asian defence. SAR leads: ICEYE reported over €250 million of 2025 revenue and raised at over €10 billion. Free data and AI foundation models make analytics cheaper, shifting margin to whoever owns the customer's decision.

01 · The answer

Pictures are cheap. Access and answers are not.

Imagery prices are falling at the low end. Planet lists SkySat archive (about 50 cm) at $6 per square kilometre and Satellogic its archive at $4; resellers quote 30 cm archive at about $17–26. On Setcoin's constellation model, a 30-satellite high-resolution fleet must sell about a fifth of what it collects just to cover its costs at those prices.

The money is in guaranteed access and in answers. Assured tasking lists at $40 per square kilometre, and sovereign deals, in which a government buys capacity or whole satellites, drove 2025–2026 growth at Planet ($116.1 million in one quarter, +58%), BlackSky (+50%, international +200%) and Satellogic (+259%).

Governments buy about two-thirds of commercial EO. Growth has moved from US intelligence, where the 2026 request cut the NRO's commercial imagery line from about $300 million to $200 million a year, to European and Asian defence: Germany's roughly €1.7 billion SAR constellation with Rheinmetall and ICEYE, Sweden, Poland, Japan.

Radar is the growth engine and AI the leveller. ICEYE has 57 SAR satellites in orbit and reported over €250 million of 2025 revenue. Free Sentinel data and foundation models such as AlphaEarth, Prithvi and Clay make analytics cheaper, so value moves to labelled history, accuracy and the customer's workflow.

The study, chapter by chapter

98 chapters + 3 appendices · 180 min · updated 2026-10-23

  1. Executive Summary and Reading Guide
  2. Method: from Pixels to Price per Square Kilometre and Value per Decision
  3. What Earth Observation Is: the Sensors
  4. A Short History of Commercial Earth Observation
  5. Resolution, Revisit and Latency: the Three Axes of Value
  6. Light, Clouds and the Physics of Optical Imaging
  7. The Fleet in Orbit: Setcoin's Satellite Catalogue Analysis
  8. Reading Fleet Claims: What "N Satellites" Really Means
  9. Revisit: the Constellation Arithmetic
  10. Collection Capacity: How Much a Satellite Can Image in a Day
  11. Price per Square Kilometre: the Price Ladder
  12. Case Study: From List Price to What a Buyer Really Pays
  13. Open Data as a Price Floor: Sentinel and Landsat
  14. Archives and the Value of History
  15. Market Size: Reconciling the Forecasts
  16. The Market Matrix: Sensors by Buyers
  17. Government Demand I: US Intelligence (NRO and NGA)
  18. US Civil Government Demand: NASA, NOAA, USDA, FEMA and Others
  19. Government Demand II: Defence Beyond the United States
  20. NATO and Multinational Buying
  21. Europe: Copernicus, IRIDE and the Push for Sovereign EO
  22. Canada, Australia, the United Kingdom and Other Allies
  23. Asia: China, Japan and India
  24. The Middle East, Africa and Latin America
  25. Commercial Demand: Agriculture
  26. Commercial Demand in Europe and the United States Compared
  27. Commercial Demand: Insurance and Finance
  28. Case Study: Flood Mapping in Hours
  29. Commercial Demand: Energy, Infrastructure and Mining
  30. Mapping, 3D, Construction and Real Estate
  31. Commercial Demand: Climate, Methane and Carbon
  32. Disasters and Humanitarian Response
  33. Maritime Domain Awareness and Dark Vessels
  34. Forestry, Land Use and Nature
  35. Water, Coasts and Oceans
  36. Ukraine as a Demand Shock
  37. Open-Source Intelligence, Journalism and Accountability
  38. Planet Labs: from Daily Pictures to Satellite Services
  39. Planet's Next Fleet: Pelican and Tanager
  40. BlackSky: Rapid Revisit and Sovereign Gen-3
  41. The Economics of Daily Global Imaging
  42. Vantor (Maxar Intelligence): the Incumbent
  43. Airbus and the European Incumbents
  44. ICEYE and the SAR Boom
  45. SAR for Buyers: Modes, Resolution and What Radar Can and Cannot Do
  46. The Other SAR Operators: Capella, Umbra, iQPS, Synspective and the Incumbents
  47. Commercial Weather and Atmospheric Data
  48. Other Optical Operators: Satellogic, Airbus, EarthDaily and the Field
  49. Aerial Imagery, Drones and Other Competitors
  50. Constellation Economics: What a Square Kilometre Costs to Produce
  51. Ground Truth, Labels and Accuracy
  52. The Analytics Value Chain: Who Captures the Money
  53. Change Detection: the Core Analytic
  54. AI Foundation Models for Earth Observation
  55. Case Study: Building an EO Analytics Product
  56. On-Board Processing and Edge AI
  57. Cloud, Compute and Storage Economics
  58. Data Platforms and Marketplaces
  59. A Setcoin Price Index for Satellite Imagery
  60. Pricing Models: Per Square Kilometre, Subscription, Capacity and Outcome
  61. Government-Built Proliferated Constellations: Competitor or Customer?
  62. Satellites-as-a-Service: Selling Constellations to Governments
  63. Contract Terms Buyers Should Negotiate
  64. Case Study: A Ministry Buys a Sovereign SAR Capability
  65. What Monitoring Costs a Buyer
  66. How Governments Evaluate EO Suppliers
  67. Government Share Stress Test
  68. Listed EO Companies Compared: Operating Metrics
  69. Regulation I: US Licensing, Shutter Control and Export
  70. Licensing Imagery for AI Training
  71. Regulation II: Europe, Privacy and Security Rules
  72. Spectrum for Earth Observation
  73. The Ground Segment: Downlink, Relays and Latency
  74. Orbital Congestion and Debris Risk for EO Fleets
  75. Launch and Supply Chain for EO Operators
  76. Consolidation and M&A
  77. Security of EO Systems: Cyber, Jamming and Counterspace Threats
  78. Capital Flows into Earth Observation
  79. The Technology Giants and Earth Observation
  80. Hyperspectral and Thermal: the New Sensors
  81. Europe's EO Start-up Ecosystem
  82. Very High Resolution and Very Low Orbit
  83. The Arctic: a Growing Theatre for Earth Observation
  84. RF Sensing and Signals Intelligence from Commercial Satellites
  85. Imaging Space from Space: Space Domain Awareness
  86. Scenarios to 2035
  87. Earth Observation in 2026 in Numbers
  88. Implications by Reader
  89. Earth Observation for Infrastructure Due Diligence
  90. Due-Diligence Checklist for an EO Company
  91. A Buyer's Procurement Playbook
  92. Risk Register
  93. The Defence-Led Path to 2030: a Narrative Scenario
  94. Signals to Watch and What Would Change Our View
  95. Setcoin's View in One Page
  96. Twenty Questions This Study Answers
  97. How Earth Observation Connects to the Rest of the Space Economy
  98. Methodology and Sources
  99. Appendix A: Operator Reference Table
  100. Appendix B: Glossary
  101. Appendix C: Model Reference Values

02 · The evidence

Four Numbers Behind the Market

Artist's concept of a train of Earth-observing satellites (NASA)

$4B

NRO commercial imagery programme

EOCL over ten years with options; the 2026 request cut it by about a third

First radar image from the NASA-ISRO NISAR satellite (NASA)

€1.7B

Germany's sovereign SAR constellation

Rheinmetall–ICEYE joint venture; production from Q3 2026

New Orleans and Lake Pontchartrain seen by Landsat (NASA)

$815M

Planet's backlog

About half due within twelve months; satellite services drive growth

Irrigated crop circles in Al Jowf, Saudi Arabia, in false colour (NASA)

Free

Sentinel-2 and Landsat imagery

10–30 m every few days: the price floor every paid product must beat

03 · The price ladder

What a Square Kilometre Costs

List prices per square kilometre in 2026, from free government data to assured tasking. Large buyers negotiate lower; minimum orders and cloud raise the effective price for small sites.

List price per km², 2026

$ per km²

Sentinel-2, Landsat (10–30 m)Free
Satellogic archive (~70 cm)$4
SkySat archive (~50 cm)$6
SkySat flexible tasking$12
30 cm archive (resellers)~$17–26
30 cm tasking (resellers)~$25–55
SkySat assured tasking$40

Planet and Satellogic list prices as reported (Sep 2026); reseller guides for 30 cm.

Try the model

How often can a fleet see a place?

Sixteen agile optical satellites can look at a mid-latitude point about every four to five hours; one satellite about every three days; a 56-satellite SAR fleet about every hour. Add cloud to see how long you really wait for a clear optical image.

Interactive model

Revisit calculator: how often can a fleet see a place?

How many hours pass, on average, between opportunities to image a mid-latitude point, for a given fleet?

Scenarios
4.4 hAverage revisit
5.5Looks a day
70Satellites for hourly revisit
3Satellites for daily revisit
One VHR optical2.9 days
16 agile optical4.4 h
30 agile optical2.3 h
Dove-class (102, nadir)15.9 h
ICEYE-class SAR (56)54 min
Your fleet4.4 h
Average revisit in hours at mid-latitudes (log scale): reference fleets and yours
Show the calculation table
ItemValue
Area within reach per satellite per day (M km²)350.9
Daylight factor0.5
Average revisit (hours)4.36
Expected wait for a clear look (hours)4.36
Sensitivity: which input moves the answer most

What moves “Average revisit” most: each input 20% lower (left of a pair) and 20% higher

Field of regard (pointing width)+1.1 / −0.721
Satellites+1.0 / −0.688
Area within reach per satellite per day = 605,000 km of ground track × field of regard.Average revisit (hours) = 24 × 510M km² ÷ (satellites × area within reach × daylight factor).Mid-latitude approximation: real revisit is shorter near the poles and longer at the equator. Cloud applies to optical: expected time to a clear look = revisit ÷ (1 − cloud).

This is one of 5 interactive models in the report. The first 3 chapters of the report are open to everyone; a Space subscription, with a 7-day free trial, opens every model and its Excel file.

04 · Three findings

Where the Money Moves

From company results, government contracts and Setcoin's models, October 2026.

Retreat of Greenland's Jakobshavn glacier seen by satellite (NASA)
Sovereign

Governments buy satellites, not images

After a week in March 2025 when Ukraine lost access to US-licensed imagery, defence ministries bought capacity they control: Germany, Sweden, Poland, Japan and others.

Hurricane Harvey flooding seen in a NASA satellite image (NASA)
SAR

Radar sees through cloud and night

ICEYE's 57 satellites, flood maps in hours for insurers, dark-ship detection for navies: SAR is the fastest-growing part of commercial EO.

City lights along the Nile at night (NASA)
Answers

AI moves value to the decision

On Setcoin's model, analytics margins are about 80% on free data and about 40% on $10/km² imagery. Owning the customer's workflow matters more than owning pixels.

Try the model

What must a square kilometre sell for?

A 30-satellite high-resolution fleet costs about $4.60 per square kilometre it collects and about $23 per square kilometre it sells at 20% utilisation. Switch to a sovereign deal, a SAR fleet or a monitoring fleet.

Interactive model

Constellation cost: what must a square kilometre sell for?

What does a fleet spend per square kilometre it sells, and how much of its collection must it sell to cover costs?

Scenarios
$22.83Cost per km² sold
$4.57Cost per km² collected
38.1%Share to sell at $12/km²
$100MAnnual cost
5% sold Below cost$91.32
10% sold Below cost$45.66
20% sold Below cost$22.83
40% sold Covers cost$11.42
60% sold Covers cost$7.61
80% sold Covers cost$5.71
Your fleet (20% sold)$22.83
Cost per km² sold by share of collection sold (log scale) · compare: $12/km²
Show the calculation table
ItemValue
Annual cost ($M)100
Area collected a year (km²)21,900,000
Area sold a year (km²)4,380,000
Cost per km² collected ($)4.566
Cost per km² sold ($)22.831
Break-even share at the compared price38.1%
Sensitivity: which input moves the answer most

What moves “Cost per km² sold” most: each input 20% lower (left of a pair) and 20% higher

Collection per satellite+5.7 / −3.8
Share of collection sold+5.7 / −3.8
Satellite life+3.4 / −2.3
Satellites+2.3 / −1.5
Satellite cost−2.2 / +2.2
Operations, ground and staff−1.8 / +1.8
Launch per satellite−0.548 / +0.548
Annual cost = satellites × (satellite + launch) ÷ life + operations.Area sold a year = satellites × km² a day × 365 × share sold. Cost per km² sold = annual cost ÷ area sold.Break-even share at a price = cost per km² collected ÷ price. Excludes development, financing, failures and sales costs.

This is one of 5 interactive models in the report. The first 3 chapters of the report are open to everyone; a Space subscription, with a 7-day free trial, opens every model and its Excel file.

05 · Who operates it

Where Earth Observation Is Run

Operators, government programmes and a key polar ground station. Ringed: the largest fleet in its class or a sovereign programme. Switch between world, North America, Europe and Asia-Pacific.

Sovereign or largest fleet · hover or tap a site for detail

All 18 sites, by category

Optical operators

  • Planet (San Francisco) ◎ — 115 Doves, 14 SkySat, 9 Pelican, 2 Tanager in orbit
  • Vantor (Westminster, CO) — 30 cm WorldView and Legion (9)
  • BlackSky (Herndon, VA) — 16 satellites; Gen-3 sovereign
  • Airbus Defence and Space (Toulouse) — Pléiades Neo 30 cm, SPOT
  • Satellogic (Montevideo / Buenos Aires) — ~70 cm, lowest list prices
  • Chang Guang (Changchun) ◎ — Jilin-1, China's largest commercial fleet
  • Pixxel (Bengaluru) — Hyperspectral

SAR and RF operators

  • Capella (IonQ) (San Francisco) — SAR, 9 in orbit
  • Umbra (Santa Barbara) — Very high resolution SAR, 5 in orbit
  • HawkEye 360 (Herndon, VA) — RF geolocation, 39 in orbit
  • ICEYE (Espoo) ◎ — Largest commercial SAR fleet, 57 in orbit
  • Rheinmetall ICEYE plant (Neuss) ◎ — German sovereign SAR production from Q3 2026
  • iQPS (Fukuoka) — SAR, 11 in orbit
  • Synspective (Tokyo) — StriX SAR, 11 in orbit
  • Spire (Vienna, VA / Luxembourg) — AIS, weather, RF; 96 in orbit

Government programmes and ground stations

  • IRIDE / e-GEOS (Rome) ◎ — Italian constellation, 31 in orbit
  • Copernicus (ESA ESRIN, Frascati) — Free Sentinel data
  • Svalbard ground station (KSAT) — Downloads polar orbits on most passes

Company and agency statements; fleet counts from Setcoin's catalogue analysis, 20 Oct 2026.

Try the model

What if US intelligence buys less?

An illustrative $440 million company with 60% government revenue, 40% of it from US intelligence, still grows 17% after a 30% US cut if sovereign and commercial sales grow. Lose the US line with no growth elsewhere and revenue falls 24%.

Interactive model

Government share stress test

If US intelligence buys less, how much revenue does an EO company keep, given growth elsewhere?

Scenarios
$516MRevenue next year
+17.2%Change
$106MUS intelligence revenue this year
24.0%Revenue at risk if the US line ends
US intelligence: this year This year$106M
US intelligence: next year Next year$74M
Other government: this year This year$158M
Other government: next year Next year$222M
Commercial: this year This year$176M
Commercial: next year Next year$220M
$M: this year and next, by customer group
Show the calculation table
Customer groupThis year ($M)Next year ($M)
US intelligence105.673.9
Other government158.4221.8
Commercial176220
Total440515.7
Sensitivity: which input moves the answer most

What moves “Revenue next year” most: each input 20% lower (left of a pair) and 20% higher

Revenue this year−103 / +103
US intelligence share of government revenue+14.8 / −14.8
Growth of other government revenue−12.7 / +12.7
Growth of commercial revenue−8.8 / +8.8
Government share of revenue+6.9 / −6.9
Cut to US intelligence revenue+6.3 / −6.3
US intelligence = revenue × government share × US share. Other government = revenue × government share − US intelligence. Commercial = revenue × (1 − government share).Next year = US × (1 − cut) + other government × (1 + growth) + commercial × (1 + growth). A negative cut means growth.An illustration, not an estimate for any named company.

This is one of 5 interactive models in the report. The first 3 chapters of the report are open to everyone; a Space subscription, with a 7-day free trial, opens every model and its Excel file.

06 · Signals to watch

The Next 24 Months

  1. 2026–2027US intelligence budget for commercial imageryWhether EOCL stabilises or keeps shrinking.
  2. From Q3 2026Germany's SAR satellites from NeussProof of the sovereign manufacturing model.
  3. 2027–2028Europe's EOGS: build or buyA federated service buying from industry, or government-owned satellites.
  4. 2026–2028Archive licences to AI developersThe first large deals would turn archives into a revenue line.

07 · Check it yourself

Test This Study Against the Live Record

Every figure here comes from filings, contracts and the public satellite catalogue. These Setcoin services let you check companies and projects yourself.

Setcoin Verify

Check a fleet claim

Satellites in the public catalogue, contracts, ownership, sanctions listings and court records on an EO company, in one cited report.

See what Verify checks →
Setcoin Atlas

See what the ground shows

Infrastructure projects worldwide, with satellite change detection to check construction where registers are silent.

Open the Atlas →
Capital Map

Who funds space and defence tech

Capital groups by sector, stage, structure and ticket size.

Explore Capital Map →

08 · The full study

What's Inside

What EO is and costs

  • Sensors, physics, history, the fleet and fleet claims
  • Revisit, collection capacity and the price ladder
  • Open data, archives, market size and the market matrix

Who buys and who sells

  • US intelligence and civil agencies, sovereign defence, NATO, Europe, Asia, other regions
  • Agriculture, insurance, energy, mapping, climate, disasters, maritime, forests, water
  • Planet, BlackSky, Vantor, Airbus, ICEYE, other SAR and optical operators

Economics, AI and risk

  • Constellation economics, the value chain, change detection and foundation models
  • Pricing models, satellites-as-a-service, contract terms, buyer's cost, stress test
  • Regulation, ground segment, supply, security, consolidation, scenarios, risks, signals
Coverage

Revisit calculator

How often a fleet can see a mid-latitude point, from satellites, pointing width and day or night sensing.

Economics

Constellation cost per km²

What a fleet spends per square kilometre it sells, and the share of collection it must sell at a given price.

Risk

Government share stress test

How an EO company's revenue responds to a cut in US intelligence purchases and growth elsewhere.

Free data tools

Check the live record yourself

The study's findings move with launches, contracts and filings. These free Setcoin tools show the current public record behind them, refreshed from the catalogues and procurement registers.

Free tool

Space Contracts & Tender Radar

607 new US federal space awards and European tenders from the last twelve months, by part or service bought, buyer, country and stage, with the tenders still open.

See which agencies are buying EO data and analytics, what they publish as value, and the tenders still open.

Open the radar →
Free tool

Space Claim Check

Type an operator’s name and see what the public satellite catalogues record: satellites launched, in orbit, operating and re-entered. The full check adds US contracts, SBIR awards, UK and NOAA licences and list screening.

Check what the catalogues record for an EO operator’s fleet: launched, operating and re-entered satellites.

Check an operator →

Questions

Questions About the Study, Answered

How much does satellite imagery cost per square kilometre?

Sentinel-2 and Landsat imagery is free. In 2026 Planet lists SkySat archive (about 50 cm) at $6 per square kilometre, flexible tasking at $12 and assured tasking at $40; Satellogic lists archive at $4. Resellers quote 30 cm archive at about $17–26 and tasking at about $25–55. Minimum orders, often 25 km², raise the effective price for small sites.

How big is the Earth observation market?

Published estimates for 2025 range from about $3.8 billion to $8 billion depending on what is counted. Setcoin's bottom-up estimate of satellite data plus first-level analytics is about $4.5–6 billion, of which governments buy about two-thirds. Setcoin's base case reaches about $13–17 billion by 2035.

Is the US cutting its purchases of commercial satellite imagery?

The FY2026 budget request proposed cutting the NRO's Electro-Optical Commercial Layer from about $300 million to about $200 million a year and zeroing a commercial radar line. Congress's intelligence committees moved to restore funding; final figures are classified.

Why are governments buying their own satellites?

For control in a crisis, speed and cost. In March 2025 Ukraine's access to US-licensed imagery was suspended for about a week. Since then Germany (about €1.7 billion with Rheinmetall and ICEYE), Sweden, Poland, Japan and others have bought satellites or dedicated capacity.

What is SAR and why is it growing?

Synthetic aperture radar images the ground with microwaves, through cloud and at night. It suits defence, maritime surveillance, flood mapping and ground-movement monitoring. ICEYE, the largest commercial SAR operator with 57 satellites in orbit, reported over €250 million of revenue in 2025.

How often can satellites image the same place?

On Setcoin's approximation, one agile optical satellite can look at a mid-latitude point about every three days, sixteen about every four to five hours, and a 56-satellite SAR fleet about every hour. Cloud lengthens the wait for a clear optical image.

How is AI changing satellite analytics?

Foundation models such as Google's AlphaEarth Foundations, NASA and IBM's Prithvi and Clay are pre-trained on large archives, so new tasks need far fewer labelled examples. That makes analytics cheaper to build and moves value to labelled history, accuracy and integration into customers' decisions.

What does it cost to monitor an area with satellites?

On Setcoin's model, monitoring 1,000 km² weekly costs about $104,000 a year with free Sentinel data plus analytics, about $172,000 with 3 m monitoring imagery, and about $915,000 with 50 cm flexible tasking.

How are Planet and BlackSky doing?

Planet reported $116.1 million of revenue in the quarter to 31 July 2026 (+58%), positive adjusted EBITDA and a backlog of about $815 million, guiding to $430–441 million for its fiscal year. BlackSky reported $33.3 million in the June 2026 quarter (+50%), with international revenue up 200%.

How do I read the full study?

The first three chapters are open to everyone in full, with no account; the rest can be previewed. A space sector subscription, with a free trial, opens all 98 chapters, the five interactive models and their Excel files.

Setcoin Research analyses public markets, filings and programmes for information only. It is not investment advice and does not recommend any security.

Read Earth Observation

1. Read the first 3 chapters in full, no account needed. 2. Create a free account (two minutes, no card) to download their Excel models. 3. Subscribe to Space for every chapter, model and Excel download: 7-day free trial, then €199 a month or €1,490 a year.