Science-as-a-Service for Earth Science: What Would It Actually Take?

My previous essay examined why science-as-a-service does not work for Earth science. Commercial constellations optimize for what sells; science missions optimize for measurements that matter scientifically but have limited commercial value. The requirements for decades-long continuity, absolute calibration, and institutional commitment create structural demands that differ from commercial business models.

Since then, two things have happened.

First, the ASCEND Act passed the Senate. It codifies NASA's commercial data purchasing program with explicit language that commercial is to "complement" NASA's own missions, not replace them.

Second, Jared Isaacman's confirmation hearing and written responses gave us a clearer picture of his vision for NASA's approach to Earth science. His nomination cleared the Senate Commerce Committee on December 8 and awaits full confirmation, potentially in the coming weeks. In his testimony, he acknowledged that "there will be Earth observation missions that NASA must retain for technical or strategic reasons" while emphasizing that NASA should "take advantage of affordable alternatives where they exist." How NASA's positions and policies will evolve under new leadership remains to be seen.

Meanwhile, NASA's FY2026 budget proposal cuts NASA's Earth Science division by 52 percent - raising questions about whether NASA can sustain the baseline that commercial is meant to complement, and whether it has the budget to purchase commercial data at scale.

This essay examines the structural requirements - funding, contracting, continuity, data ownership - that would need to be addressed before commercial could take on more than a supplementary role in Earth observation.

The efficiency arguments for science-as-a-service sound compelling. Commercial providers can build and launch satellites faster. They iterate quickly and they operate leaner. If NASA can buy data instead of building bespoke missions, why wouldn't it?

Baseline Earth observation (EO) has specific requirements that distinguish it from other domains where commercial procurement has succeeded. This is not about whether commercial providers are capable - many are. It is not about whether NASA should buy commercial data - the ASCEND affirms that it should. The question is narrower: "what structural conditions shape how far commercial's role can extend beyond supplementary data purchases?"

Isaacman's testimony and written responses framed the choice as not binary - some missions NASA retains, others shift to commercial alternatives. That framing is reasonable. But it leaves open the question of what distinguishes one category from the other, and what structural conditions would need to be true for commercial to take on more than supplementary roles. That is what this essay examines.

What Is "Baseline" Anyway?

EO missions fall into two broad categories, and the distinction matters for understanding where commercial can and cannot substitute for public investment.

Exploratory missions test new sensors, prove concepts, and demonstrate capabilities. They are designed to be short-lived - two to five years - and to iterate quickly. They target specific questions or technology validations rather than systematic global coverage. Commercial providers can absolutely contribute here, and often do so faster and cheaper than traditional government procurement. ESA's Scout missions follow this model: small, rapid, focused on innovation.

Baseline continuity missions are structurally different. They provide systematic, long-term measurement records - often spanning thirty years or more. They require global coverage and consistent methodology so that measurements from 1984 can be compared to measurements from 2024.

Landsat is the canonical example: a continuous record of land surface change stretching back to 1972. GRACE and its successor GRACE-FO measure changes in Earth's gravity field to track groundwater depletion and ice sheet mass loss. Systematic SAR missions like Sentinel-1 provide consistent radar observations regardless of cloud cover or daylight.

These are not innovative missions in the commercial sense. They can’t iterate freely the way commercial products do, because every upgrade imposes a continuity cost: overlap, cross-calibration, algorithm stewardship, and often reprocessing of the archive. The value comes from managed continuity—keeping measurement definitions stable enough that 1984 can be compared to 2024, and changing only in ways that preserve traceability across decades. What matters here is that these requirements create structural demands that differ fundamentally from what commercial business models optimize for.

The Coverage Factor

Beyond continuity, baseline missions differ from exploratory ones in another fundamental way: coverage requirements.

Commercial business models optimize for regions where customers will pay. Agricultural areas, urban centers, infrastructure corridors, defense-relevant locations. This is rational - companies need revenue, and revenue comes from customers with purchasing power.

Baseline EO requires something different. Planetary processes do not respect commercial geography. Arctic methane emissions matter for the global carbon budget even though no one lives there to purchase imagery. Antarctic ice sheet dynamics drive sea level projections even though there is no commercial market for glacier monitoring. Southern Ocean carbon uptake affects atmospheric CO2 trajectories even though ships rarely transit those waters.

Systematic global coverage means exactly that - coverage driven by scientific requirements rather than market demand. When people say we can use commercial providers that "already have constellations," those constellations are optimized for commercial demand patterns, not systematic planetary observation. A provider might offer Arctic coverage if NASA pays for it specifically, but that is different from the standing requirement to observe everywhere, all the time, regardless of who is paying for any particular image.

Structural Conditions for Earth Science-as-a-Service

Moving commercial EO from supplement to baseline isn’t blocked by capability, it is blocked by institutional plumbing.

Baseline measurements require – (1) multi-year funding and contracting that can span administrations, (2) continuity guarantees when a provider fails, pivots, or upgrades technology, and (3) open, global data access without fragmenting the downstream ecosystem.

If those three conditions are not deliberately designed into procurement and governance, “science-as-a-service” will either revert to narrow, task-order data buys that don’t produce systematic records, or require contracts so large and durable that the promised budget savings disappear.

Structural Constraint #1: Funding and Contracting

NASA's Earth Science budget is approximately $2.2 billion per year at current levels. The FY2026 proposal cuts that by 52 percent. The science-as-a-service model is being proposed not as an expansion but in the context of dramatic budget reduction.

The Contracting Problem

Commercial providers need revenue certainty to commit capital to satellite development and operations. They cannot build constellations on the promise that Congress might appropriate funds next year. NASA operates on annual Congressional appropriations, which means the agency cannot commit to decade-long contracts even if it wanted to. A new administration can shift priorities. A budget cut can eliminate planned purchases. Commercial providers know this. They price that uncertainty into any arrangement, or they decline to participate in bids that require long-term capital commitment without matching revenue guarantees.

The ASCEND Act, having passed the Senate, would codify the NASA's Commercial Satellite Data Acquisition (CSDA) program but does not change this fundamental dynamic. The Act ensures scientific publication rights, allows NASA to negotiate licensing terms for broad data access, and requires annual reporting on how commercial acquisitions advance Decadal Survey priorities.

But ASCEND authorizes NASA to purchase commercial data; it does not provide multi-year contracting authority or insulate funding from annual appropriations cycles. The program it codifies remains subject to whatever budget Congress provides each year.

The National Reconnaissance Office has historically offered a contrast. NRO's Electro-Optical Commercial Layer program, awarded in 2022, is worth more than $4.4 billion over ten years across three providers - Maxar alone holds a $3.4 billion contract, with BlackSky at approximately $1 billion and Planet's undisclosed but substantial. That contract structure gives commercial providers the certainty needed to invest in capacity.

But NRO operates under a different appropriations structure, with national security urgency and bipartisan protection that civilian science programs do not command. And even this model is now under pressure: the FY2026 budget proposal reportedly includes cuts of up to 30 percent to EOCL, and funding for a planned commercial SAR program has been removed entirely. Commercial providers have petitioned Congress to reverse these cuts, arguing they would undercut both national security capabilities and the commercial remote sensing industry's financial stability.

If even EO for national security - historically the most protected category - faces funding uncertainty, civilian science programs cannot assume greater stability.

The Scale Problem

The scale required compounds the contracting challenge. Baseline EO spans multiple measurement types: multispectral (like Landsat), synthetic aperture radar, hyperspectral, gravity measurements, atmospheric composition, ocean color. Expanding commercial's role substantially would require contracts across multiple providers, each needing sufficient scale to justify investment. A rough estimate - my own, for illustration - might put the annual contract value required at $500 million to $1 billion for significant commercial contribution beyond current CSDA levels. That figure approaches or exceeds Planet's entire FY2025 annual revenue of approximately $244 million.

ESA's Copernicus program offers a different model worth examining. The Copernicus Contributing Missions framework integrates commercial and national missions alongside the Sentinel baseline - but the Sentinels remain publicly funded and operated. Commercial contributes and supplements, but does not replace the foundational infrastructure. That distinction reflects the structural reality that commercial investment requires returns that baseline EO cannot generate through market mechanisms alone.

The Budget Contradiction

The fundamental tension is this: science-as-a-service is discussed as a way to reduce costs, but commercial providers need sustained contracts to commit to continuity, and sustained contracts require funding that NASA's budget may not have.

The savings from not building government satellites do not materialize if the alternative is contracts large enough and long enough to give commercial providers investment certainty.

Administration change compounds this. Even if one administration commits to commercial EO contracts, the next administration may have different priorities. Commercial providers understand this. Either the contracts span administrations - which requires legislative authority NASA does not currently have - or commercial providers face stranded assets when priorities shift.

Structural Constraint #2: Continuity and Business Risk

Commercial companies operate on business viability, and that creates structural tensions with baseline EO requirements.

Business Volatility

When customer demand shifts, commercial providers adjust their offerings. When technology improves, they upgrade to the new capability even if it breaks continuity with previous measurements. When better returns appear elsewhere - a defense contract, a different market segment - they pivot. This is normal business logic, not a criticism. But it is incompatible with the requirement that baseline measurements continue regardless of market conditions.

The EO industry has seen significant business volatility. Planet and Maxar (now Vantor) have pivoted strategies multiple times. Google acquired Skybox Imaging for $500 million in 2014, renamed it Terra Bella, then sold it to Planet in 2017 when EO no longer fit its strategic priorities - a major tech company deciding the business wasn't worth continuing. BlackSky, Spire, and other providers have faced financial pressures that forced strategic changes. Acquisitions consolidate the market but introduce uncertainty about long-term data policies. None of this is unusual for a maturing commercial sector. But it raises a concrete question for the scenarios this essay examines: if commercial providers were to take on baseline roles, what happens when one fails, pivots, or gets acquired by an entity with different priorities?

When Things Go Wrong

The Sentinel-1B failure in December 2021 illustrates what happens when baseline observations degrade. ESA scrambled to fill coverage gaps with commercial and partner data. The response was ad-hoc - purchasing imagery here, coordinating with national missions there, accepting reduced coverage in some regions. It worked partially because the public baseline infrastructure (Sentinel-1A, other national missions) still existed to anchor the response. The gap persisted until Sentinel-1C launched in late 2024 and became operational in early 2025.

Now consider the counterfactual: if commercial were the baseline, what would the backup be? There would be no fallback public asset. Rebuilding capability from scratch would take years and billions of dollars. The measurement record would simply break.

Governance and accountability fragment under commercial provision in ways that matter for continuity. Currently, when Landsat or Sentinel has issues, there is a clear chain of responsibility: NASA or ESA, with public accountability to scientific users, international partners, and legislative oversight. Under a commercial model, accountability splits: NASA handles contracting and oversight, the vendor handles operations while serving profit motives, and Congress controls appropriations that neither NASA nor the vendor fully controls. When continuity breaks, who is accountable? Diffuse responsibility typically means no one is accountable in practice.

Structural Constraint #3: Openness, Licensing, and Ownership

The third structural constraint concerns data access, licensing terms, and the question of who owns what when government funds commercial capability development.

The Open Data Trade-off

NASA can require open data access in contracts - and commercial providers can deliver it if the price is right. Umbra's decision to release SAR data under CC BY 4.0 licensing demonstrates that commercial providers can support open science. The question is not whether open licensing is possible, but what it costs at baseline scale.

The trade-off is concrete.

Option A: NASA pays enough that open data becomes the provider's business model for baseline imagery, effectively becoming the anchor customer that subsidizes free access for everyone else.

Option B: NASA accepts restrictions - delayed release, limited coverage, usage constraints - that preserve the provider's ability to monetize the same imagery commercially.

Option B is cheaper but defeats the purpose of baseline observations, which require immediate, unrestricted, global availability. Option A is possible but expensive, tying directly back to the funding constraints raised earlier.

Baseline EO has historically operated on a different model. Landsat data has been free and open since 2008. Copernicus Sentinel data is free and open from day one. This is foundational to the entire downstream ecosystem. Commercial analytics companies, agricultural technology firms, insurance risk modelers, academic researchers - they all build on free baseline data. If baseline becomes a commercial subscription service, then the EO ecosystem fragments.

Capability and Ownership

There is a deeper issue around capability development. If NASA funds a commercial provider to develop a new sensor, validate a platform, or prove a processing capability, what happens once that capability is proven? Can the provider monetize it exclusively? Does NASA retain guaranteed access, or does it compete with other customers? Government contracting often transfers intellectual property to contractors, but the terms vary and the implications for long-term data access are not always clear.

Commercial providers optimize for paying customers. They allocate satellite tasking based on who is willing to pay and how much. Government baseline requirements include systematic coverage of low-commercial-value regions - the Arctic, the open ocean, developing nations that cannot afford commercial imagery subscriptions. If NASA is one customer among many, it is not clear that systematic global coverage results. Providers will serve NASA's specific orders but may not maintain the spare capacity or operational patterns that systematic coverage requires.

Where Science-as-a-Service Does Fit

The structural constraints examined above apply specifically to baseline substitution. Commercial EO has a significant and growing role in science - but one that operates on top of public baseline infrastructure rather than replacing it.

The clearest examples involve supplementing baseline observations for specific research needs:

• Scientists studying volcanic activity need daily imagery of an eruption, commercial high-revisit constellations can provide what Landsat or Sentinel cannot.
• Researchers need SAR interferometry at higher temporal frequency than Sentinel-1 provides - commercial providers like Iceye or Umbra can fill gaps.
• A research campaign requires hyperspectral data that no current NASA mission collects - commercial sources may be the only option.

These are genuine scientific use cases where commercial data adds capability.

The Copernicus Contributing Missions framework - funded by the European Commission and implemented by ESA - offers a structural model. Commercial and national missions contribute observations that complement the Sentinel baseline. The program now includes over 20 commercial providers delivering very high resolution, high frequency, multisensor datasets. But the Sentinels remain publicly funded and operated. The distinction is maintained: public baseline plus commercial augmentation.

The boundary that matters is between additive and substitutional. Purchasing commercial data to enhance research capabilities is different from depending on commercial providers for the measurements that anchor long-term records, calibrate other instruments, and enable the downstream ecosystem. The structural constraints examined in this essay emerge when commercial provision crosses that line - when it moves from extending what public missions can do to replacing what public missions must do.

What a Functional Ecosystem Would Require

A science-as-a-service model that works would need contract structures that align incentives across the measurement lifecycle.

A functional science-as-a-service ecosystem would connect the three phases: R&D where commercial providers develop and test new capabilities, demonstration where the agency validates and builds the user base, and operations where successful demos transition to sustained procurement. Each side contributes what it does best - commercial provides speed and iteration, the agency provides scientific credibility and institutional commitment.

What is missing today, I think, is the pathway connecting these stages. Demonstration missions happen, but no mechanism exists that says: if this measurement proves valuable, here is how it transitions to operational status with committed multi-year funding. Building that pathway would require more than goodwill - it would require legislative authority for long-term contracts, appropriations that span administrations, and procurement structures designed for continuity rather than annual competition. Those are solvable problems, but they require deliberate institutional design that does not currently exist.

Conclusion

The conditions examined here - sustained funding, long-term contracting authority, continuity mechanisms, open data policies and licensing frameworks - have not been met. Some would require legislative changes. Some would require sustained political commitment that civilian Earth science has historically not commanded.

There is a harder question underneath this analysis. The structural constraints examined here assume a choice between public and commercial provision.

But, in a budget environment where NASA's Earth Science division faces dramatic cuts, the realistic alternative to commercial baseline may not be robust public baseline - it may be degraded or discontinued observations altogether. That changes the calculus, but it does not change the structural requirements. Commercial provision would still need the funding, contracting, and continuity conditions examined here. Those conditions become harder to meet, not easier, when the overall budget is shrinking.

The structural constraints examined here are not arguments against commercial EO. They are the factors that, I hope EO companies would agree, need to be addressed before science-as-a-service could extend to the foundational measurements that everything else depends on.

Whether those factors can be addressed - and whether the political will exists to address them are questions I can't answer. What this essay has tried to do is put them on the table.

Until next time,

Aravind