There has been a lot of talk lately about whether commercial Earth observation (EO) companies could replace parts of NASA’s Earth science mission portfolio. With a new Administrator coming in, that debate is heating up.
As someone who works in EO across strategy, policy and business, I wanted to lay out my perspectives on why the roles of public and private sectors are not interchangeable.
The catalyst for writing this was public statements and broader industry conversations, not any official NASA position (at least, not yet). The debate is real, the stakes are high, and I wanted to put my thinking down while these decisions are still being made.
Jared Isaacman, President Trump's nominee for NASA Administrator has articulated a compelling vision: "NASA needs to constantly be recalibrating to do the near impossible, what no one else is doing - and the things they figured out, they hand off to industry."
While excerpts from his manifesto were leaked recently, Isaacman clarified with a lengthy post on X. On Earth observation specifically, he said: "I know the “science-as-a-service” concept got people fired up, but that was specifically called out in the plan for Earth observation, from companies that already have constellations like Planet, BlackSky, etc. Why build bespoke satellites at greater cost and delay when you could pay for the data as needed from existing providers and repurpose the funds for more planetary science missions (as an example)?"
This is not theoretical. NASA had already issued an RFI for Landsat continuity, asking whether the private sector can deliver Landsat-class performance at roughly $130 million per year versus the billion-dollar missions of the past. With budgets constrained and commercial constellations proliferating, the "science-as-a-service" model seems logical.
The space industry's transformation validates this thinking. SpaceX revolutionized launch. Commercial crew missions ferry astronauts to the ISS. If it worked for rockets, why not Earth science?
Because Earth science is fundamentally different. The budget pressures are real, but the solution is not wholesale privatization. Conflating launch services with Earth science risks dismantling infrastructure that enables not just climate research, but weather forecasting, disaster response, agricultural monitoring, and the commercial Earth observation (EO) industry itself.
Why Commercial EO and Public Science Missions Are Not Comparable
First up, EO is not monolithic - it is not just about taking pictures from space. Commercial constellations optimize for what sells: high-resolution optical imagery for defense, precision agriculture, and infrastructure monitoring. Companies like Planet, BlackSky, and Vantor (previously Maxar) excel at this.
But Earth science missions optimize for entirely different parameters. They measure atmospheric chemistry, ocean color for phytoplankton tracking, ice sheet thickness, greenhouse gas concentrations, evapotranspiration, and sea level rise. These measurements require precise spectral bands, rigorous calibration, and long-term stability - not the highest resolution or fastest revisit.
Most critically, these measurements have limited commercial value. Who pays for daily atmospheric CO2 soundings? Who buys ice sheet velocity data? The market for these products is researchers, climate modelers, and policymakers - not customers with procurement budgets. Companies optimize for revenue. Science missions optimize for measurements that matter scientifically but don't generate profit.
ESA's Copernicus program demonstrates this distinction clearly. Sentinel satellites provide systematic global coverage of land, ocean, and atmosphere, measurements of which are designed for science and policy, not commercial sales. The program costs over €7 billion because these observations require dedicated infrastructure that commercial providers have no incentive to build.
Much of the cost of Earth science missions is not the spacecraft itself but the calibration infrastructure, validation networks, science teams, metrology labs, retrieval algorithms, and long-term data stewardship. These components are essential for turning raw measurements into stable, comparable scientific records, and they are often the first things reduced in commercial models to control costs.
Commercial constellations optimize for what sells. Science missions optimize for what matters scientifically, more often measurements with limited, direct commercial value.
The Category Error: EO Is Not a Normal Market
EO is often compared to other infrastructure that successfully transitioned to private operation: telecommunications, launch services, energy. But this analogy breaks down under scrutiny.
Those industries have clear revenue models and customer demand. Telecom companies charge customers for connectivity. Launch providers charge for payload delivery. Energy utilities sell power. In each case, the service has direct paying customers and can generate sustainable returns.
Earth science measurements don't work this way. The benefits are diffuse and public: better climate models, early warning systems for droughts, understanding of ocean acidification, tracking of deforestation. Society needs this information, but no single entity will pay for it at the scale required.
Consider ice sheet measurements - radar altimetry and interferometry to track thickness, velocity, and mass balance in Greenland and Antarctica. This data is essential for projecting sea level rise and understanding asset and societal risk for coastal communities. No commercial provider operates systematic polar ice monitoring missions - there's no commercial market for continuous polar coverage at the calibration standards science requires.
Insurance companies won't fund it. Agriculture won't fund it. Defense won't fund it. Yet this information is critical for coastal planning and climate projections globally. This is not a temporary market inefficiency, it is a structural public goods problem.
The comparison to launch services is particularly interesting. SpaceX succeeded because NASA guaranteed demand through Commercial Crew contracts and paid for development milestones. The government created a market. But for ocean color or ice dynamics measurements, there is no latent commercial demand to unlock.
The government is not just an early customer, for many of these measurements; it is effectively the only customer.
Decades-Long Continuity Can't Be Contracted
Understanding Earth systems requires more than data - it requires comparable measurements across decades. Landsat's value isn't the 30-meter resolution; it is that a pixel from 1984 can be directly compared to a pixel today through consistent calibration chains, stable sensor performance, and institutional commitment to data quality.
Commercial companies optimize for different time horizons. They pivot to higher-margin opportunities, get acquired, or fold when business models change. A commercial provider might deliver excellent Landsat-class data for five years, but who guarantees the next forty-five?
Commercial EO satellites are designed for rapid refresh: 3–5 year lifetimes for SmallSats and 7–10 years for larger buses. Earth data records require continuity across 30–50 years. Short-lived platforms introduce instrument drift, orbit change, and calibration breaks that make long-term comparison impossible. The scientific requirement is stability, not speed of refresh.
Reanalysis datasets such as ERA5, MERRA-2, and JRA-55, which are crucial for both weather prediction and climate modeling, depend on uninterrupted, calibration-stable measurements across decades. Even small gaps or instrument inconsistencies degrade the skill of these climate datasets for years. Commercial data sources are not designed to meet the multi-decadal stability requirements these systems depend on.
Operational weather forecasting illustrates this starkly. NOAA's GOES and JPSS satellites provide continuous atmospheric observations that feed into forecast models globally. These systems require 24/7 operational uptime and multi-decadal continuity. A gap of even days would degrade forecast skill worldwide. Could a commercial provider guarantee that level of reliability if a more profitable defense contract materialized? Would they maintain obsolete sensors to preserve calibration continuity when upgrading would cut costs? It does not make business sense.
The Landsat RFI demonstrates this tension by capping the program at $130 million annually, far below what commercial providers would need to guarantee multi-decadal continuity while maintaining profit margins. The budget reality suggests that "commercial Landsat" might mean accepting reduced continuity or quality, not achieving the same outcomes more efficiently.
The Foundation Can't Be Privatized
Here is an underappreciated irony: commercial EO companies already depend on NASA's public infrastructure. Commercial EO companies like Planet calibrate against Landsat. They use the same validation sites, like Railroad Valley in Nevada, which NASA fought to preserve from lithium mining specifically for satellite calibration. The private sector also relies on methodologies developed by the USGS Earth Resources Observation and Science (EROS) Calibration and Validation Center of Excellence (ECCOE).
This is not a bug, it is essential. Without a trusted reference standard, commercial data lacks scientific credibility. Landsat, Sentinels, and other scientific EO missions provide that anchor, and commercial providers build on this foundation.
The dependency runs deeper. NASA and ESA (along with other space agencies) develop new sensor technologies, validate retrieval algorithms, and take risks on experimental measurements. The 2017 Decadal Survey's systematic R&D pipeline identifies measurements not yet ready for operational implementation - like planetary boundary layer dynamics or surface topography and vegetation observations that require years of foundational research to mature.
If these missions are privatized, this pipeline would be at risk. Commercial providers have no incentives to invest in these measurements, especially without a clear pool of customers. The innovation that enables future commercial capabilities depends on public investment in foundational science.
You cannot privatize the foundation without undermining much of what is built on top of it. Think of it like Jenga: remove the base, and everything falls.
Public Goods Must Stay Public
Landsat data has been free and open since 2008, a policy shift that unlocked billions of dollars in downstream economic value. The 2023 Landsat Economic Valuation study (which I had the honor of co-authoring) found that Landsat alone generated $25 billion in economic benefits within the United States, supporting industries from agriculture to insurance to mining.
What is notable is that this value creation is not a given, it depends on zero-friction access. Researchers anywhere can download decades of data, develop algorithms, build applications, and publish results. The Copernicus program follows the same model, making Sentinel data freely available globally.
A "science-as-a-service" model would likely introduce paywalls or licensing restrictions. Commercial providers need returns on investment. Even if the U.S. government subsidizes data access for American researchers, what about scientists in Kenya studying regional drought patterns? In Bangladesh monitoring flood risk? In Brazil tracking deforestation? The global research enterprise that depends on open EO data would fragment.
This is not hypothetical. NASA's Commercial Smallsat Data Acquisition (CSDA) program, explicitly supplements - not replaces - flagship missions. The data comes with licensing restrictions and is not freely redistributable. It is useful for specific projects but cannot substitute for open baseline observations.
The implications extend beyond research. Consider NASA's FIRMS (Fire Information for Resource Management System), which provides near real-time wildfire data globally. Insurance companies use it to model risk. Utilities use it to protect assets. Emergency managers use it to coordinate response. Likewise, the EU's Global Wildfire Information System (GWIS) provides similar coverage.
Commercial providers typically would not build and maintain such tools without a revenue model, especially as providing free global access including for competitors and international users does not align with their business objectives. The value of FIRMS and GWIS comes from its neutrality and reliability - attributes that stem from public stewardship.
Who Bears the Risk When Systems Fail?
The fundamental governance question is: who is accountable when critical EO systems fail or degrade?
Public agencies can absorb scientific, financial, and political risk for missions with no commercial return. Private companies cannot: their fiduciary duty is to shareholders, not long-term climate stewardship. Structural misalignment, not capability, is the real barrier.
Public Earth science missions answer to society. NASA and NOAA optimize for accuracy, stability, and equitable access. They serve everyone equally - American researchers and international partners, well-funded labs and graduate students in developing nations.
This distinction matters immensely for systemic infrastructure. If foundational EO records become proprietary:
- Climate models could break. These models require consistent calibration across decades to detect subtle trends and separate signal from noise. A gap in sea surface temperature measurements or a calibration shift in atmospheric soundings would immediately degrade model skill, and the effects would linger for years. Multi-decadal climate records are not just long datasets, they are carefully maintained chains of cross-calibrated observations that can't be easily reconstructed if broken.
- International agreements might weaken. The Paris Agreement, GCOS Essential Climate Variables and other global frameworks depend on transparent, trusted EO data for verification and accountability. If baseline measurements come from commercial providers with proprietary calibration methods and potential conflicts of interest, the scientific foundation becomes contestable. Countries could dispute findings by questioning commercial data quality or neutrality.
- Developing nations would lose access. Countries without procurement budgets would lose the free data they currently use for disaster preparedness, agricultural planning, and resource management.
- Global risk systems could fragment. When major disasters hit, Landsat and Sentinel openly licensed imagery supports recovery efforts globally. Commercial EO providers do contribute data through voluntary programs, but it typically comes with licensing constraints that limit redistribution or require activation. The difference is guaranteed access versus conditional access.
Commercial providers cannot bear this responsibility under normal market incentives. Their business models optimize for profit and shareholder returns, not multi-decadal public stewardship – quarterly earnings and multi-decadal climate records operate on incompatible timescales.
Where Commercial EO Does Fit
I am not trying to make an argument against commercial space - far from it. There are legitimate and valuable roles for private providers in EO. I interact and work with almost all the companies in the sector. I am hoping they are nodding their heads along to most of what I say here.
High-resolution optical and SAR imagery for specific applications works well commercially. Defense, precision agriculture, insurance, and infrastructure monitoring all have clear customers willing to pay. Companies like Maxar, Planet, Airbus, Iceye, BlackSky, Spire, Satellogic, Capella Space etc. have demonstrated sustainable business models for higher resolution, rapid-revisit use cases - but sustainability for defense and commercial use cases is different from the multi-decadal calibration continuity that baseline science missions require.
NASA's CSDA program shows how this can work: the government purchases commercial data to supplement flagship missions for targeted research needs. Scientists gain access to high-revisit, high-resolution imagery that complements systematic public observations. Similarly, NOAA purchases commercial weather data to integrate into operational forecasting models, supplementing data from public missions, while not replacing baseline observations.
This model works precisely because it preserves the public baseline while leveraging commercial capabilities where they add value. NASA and NOAA should always strive to improve procurement efficiency, but efficiency reforms are not substitutes for scientific needs. The existence of waste does not make public-good measurements commercially replaceable.
The Contracting Problem
Making this actually work at scale requires addressing the contracting problem. Commercial providers need long-term revenue certainty to invest in continuity. The National Reconnaissance Office's Electro-Optical Commercial Layer (EOCL) program awards multi-year contracts worth billions - $5 billion over ten years split among providers. Civilian EO hasn't operated this way historically. If policymakers want greater commercial involvement, they need to provide NRO-style long-term commitments, not year-to-year appropriations.
A functional public-private architecture would look like this:
- Public missions maintain the baseline observational infrastructure - the reference standards, long-term climate records, and measurements with no commercial market. These programs guarantee open data access and multi-decadal continuity.
- Commercial providers supplement with higher resolution, higher revisit, or specialized measurements for specific applications. The government anchors commercial investment with predictable, multi-year contracts for defined capabilities.
This division of labor leverages the strengths of both sectors without expecting commercial providers to serve public missions they are not structured to deliver.
Emerging Role for Philanthropies
Beyond governments, philanthropic funding represents another emerging model. Organizations like MethaneSAT, Carbon Mapper, and Earth Fire Alliance demonstrate how philanthropic capital can support public-good missions - filling gaps that traditional government programs or commercial markets address. These hybrid models show there is room for innovation in how EO is funded and delivered.
These philanthropic missions offer distinct advantages: they can move quickly on emerging priorities (like methane monitoring), operate without quarterly earnings pressures, and bridge the gap between pure research and operational applications. They are particularly valuable for targeted problems where government programs have not yet scaled and commercial markets remain nascent.
However, these models also highlight the importance of maintaining public infrastructure. When MethaneSAT failed in 2025, researchers working with data from the mission pivoted to using data from ESA's Sentinel-5 mission, demonstrating that even well-intentioned philanthropic efforts often depend on the continuity of public baseline missions.
Innovation in funding models is valuable, but it cannot replace the foundational infrastructure that all these approaches ultimately rely on.
Earth Science Data Is Infrastructure, Not a Service
For me, the fundamental error is in framing EO as a service that can be procured on-demand, like cloud computing or satellite bandwidth or launch. Earth science is infrastructure - the foundational layer that enables everything built on top of it.
Infrastructure requires institutional commitment that transcends market cycles and political administrations. It requires transparency, neutrality, and guaranteed long-term access. It requires optimization for societal benefit rather than profit margins.
Roads, water systems, and electrical grids are obvious infrastructure. EO data is less tangible but equally foundational. Climate models, weather forecasts, drought early warning systems, agricultural planning, disaster response, and the commercial EO industry itself all depend on this baseline.
NASA and agencies like ESA don't operate Earth science missions because commercial providers can't build satellites. They operate them because some observations are too important to depend on business models, shareholder returns, or market conditions.
Isaacman is right that NASA should focus on what no one else can do. But for Earth science, the unique capability is not only building advanced sensors, it is also guaranteeing decades of consistent, open, trustworthy observations of our changing planet. That is something public institutions are structured to deliver.
The challenge is recognizing where the line falls between missions that can be handed off and those that must remain in public hands. For Earth science, that line is clear: the baseline must stay public, even as commercial providers play expanding roles around it.
The real question is not whether NASA can save money by buying data from commercial EO providers. It is whether we are willing to risk the continuity, openness, and reliability of the observations that underpin our understanding of the Earth. The answer should be obvious.
Until next time,
Aravind
