NASA published a reference architecture for lunar science data on September 11, extending its Planetary Data System standard to all 71 Artemis Accords nations. The two-part workshop series, which concluded September 8, gave signatory governments a concrete, AI-ready pipeline — the PDS4 information model — for sharing images, measurements, and samples from future missions.
China and Russia, who are building a rival lunar research station, were not included. Their data will flow into a separate framework, creating the conditions for two partial, non-interoperable scientific records of the Moon unless cross-bloc agreements are negotiated.
The Moon is about to be mapped twice — once by a coalition of 71 nations using a shared, AI-compatible language, and once by China and Russia, working from their own script. NASA spent the summer converting a six-year-old diplomatic pledge into operational infrastructure. The workshops, ending September 8, walked technical staff from signatory nations through the agency’s Planetary Data System, a curated archive that has been refined into the gold standard for planetary data over decades since its 1989 establishment. The promise was global access to the scientific record of the Artemis era. The architecture delivered this week works for those inside the tent. For those outside, it raises a question no one is publicly answering yet: what happens to humanity’s understanding of the Moon when the raw evidence sits in two rooms that do not share a door?
The blueprint is an ultimatum wrapped in XML
The reference architecture NASA distributed is the Planetary Data System’s PDS4 information model — an XML-based metadata schema designed to describe any planetary science dataset so that it can be searched, validated, and combined across institutions and decades. Every data field, unit of measurement, and contextual parameter gets a machine-readable label tied to controlled vocabularies. The system was built because NASA was losing science: a 1982 National Academy of Sciences review found early mission data had become irretrievable, killed by incompatible formats. PDS4 was the fix — and it now doubles as a gatekeeper for the Artemis scientific record.
The standard’s power is quiet but absolute. PDS4 mandates transparent, non-proprietary storage formats. Every dataset submitted passes through a peer review by one of seven discipline nodes — atmospheres, geosciences, imaging, and others — staffed by scientists who check correctness, standards compliance, and long-term usability before release. A partner nation can adopt the same model and feed its data into the same AI-ready pipeline. According to Andrew Mitchell, NASA’s deputy chief science data officer, the workshops gave Artemis Accords partners practical tools and a shared foundation to build on as the coalition moved forward together.
The Chinese model works differently. China’s Lunar and Planetary Data System divides mission output into proprietary, protection, and public periods. The China National Space Administration decides which Chang’e datasets become publicly available and when. The commitment to openness is stated. The discretion over release timing is centralised. The two systems are not technically incompatible — PDS4 can ingest calibrated Chang’e data, and already has — but the governance around them pulls in opposite directions.
According to Jacob Bleacher, NASA’s chief exploration scientist, returning humans to the Moon through Artemis will unlock scientific discovery through transparency, collaboration, and accessibility. Kevin Murphy, the agency’s chief science data officer, argues that NASA must make data easier for scientists to explore and use, pointing to the NASA-IBM Lunar Foundation Model, released September 10, as a demonstration of the pipeline’s value. The open-source AI was trained on two million image tiles from lunar missions, all drawn from PDS-formatted archives with consistent metadata. When data is structured right, you can fold it into a model that maps craters, identifies volcanic features, and estimates polar ice stability. You cannot do that with siloed national archives.
The architecture that enables the AI model is the same architecture the workshops taught. The sequence below shows how the components connect. The system is not speculative — it has been ingesting and harmonising international datasets for years, including reprocessed Chang’e microwave radiometer data.
The workshops themselves followed a deliberate arc. In May, the Indian Space Research Organisation led discussions among signatories that established a shared vocabulary for open data practices. NASA then split its follow-on into two virtual sessions — first principles, then tools — so working-level data engineers and archive specialists could participate across time zones. The second session gave participants a concrete template, not aspirational guidelines. The unspoken message: this is the standard. Build to it, or your data will not talk to ours.
What the architecture cannot do is pull Chinese or Russian data into the system without an agreement that does not exist. The Artemis Accords are bilateral, non-binding commitments with the United States, not a multilateral treaty. China and Russia have declined to sign, arguing the Accords’ resource-extraction and safety-zone provisions risk creating de facto territorial claims that conflict with the 1967 Outer Space Treaty. Their alternative, the International Lunar Research Station, launched via a March 2021 memorandum, has its own data-sharing arrangements, scheduled to begin construction between 2026 and 2035. No formal ILRS data standard has been published yet. But the path separates at the governance level, not the technical one — and governance is what decides who can read the Moon’s receipts.
The Moon is becoming a patchwork of scientific norms
Space governance analysts at institutions such as the Secure World Foundation and national strategic studies institutes have begun describing the Artemis and ILRS frameworks as rival but overlapping regimes. According to these analysts, both emphasise data sharing and present themselves as collaborative, but they embed different political priorities and safety-zone concepts. Absent a UN-backed lunar framework, the two coalitions risk creating fragmented norms for scientific data, resource use, and heritage protection rather than a single global standard.
The commercial layer complicates the picture further. Section 8 of the Artemis Accords commits signatories to open sharing of scientific data from government missions — but the commitment does not apply to private-sector operations unless conducted on behalf of a signatory. Data from companies like Intuitive Machines or Astrobotic is not covered unless the flight is a contracted government payload. The most commercially sensitive data may sit in proprietary silos even as headline images are released. The frameworks for openness are genuine but narrow, and the narrow part is where the money is.
For Western governments and firms, a bifurcated record carries indirect but concrete costs. Defence and space agencies rely on global datasets to calibrate navigation and space situational awareness models. If ILRS partners keep key Chang’e and Luna measurements in non-aligned archives, Western missions may need to repeat observations — raising costs and delaying site selection for infrastructure. Commercial ventures could face asymmetric access if Chinese hardware partners receive richer ILRS data than Western competitors, complicating joint projects and insurance underwriting.
The workshops did not create the split. They made it visible. NASA has built a 71-nation scientific coalition with a shared data standard — a historically significant achievement that will produce a partial record of humanity’s return to the Moon unless separate data-sharing arrangements with China and Russia are negotiated. The next twelve to twenty-four months will show whether the ILRS codifies a strictly separate archive model or signals a door for cross-bloc integration. The Moon will keep both sets of data. The question is whether anyone can read them together.
Beyond the headline
The bigger picture
The workshop series quietly pushed lunar governance toward a contest over infrastructure, not launch tonnage. The formats and metadata rules that determine how lunar observations are combined and searched will also decide whose data flows most easily into global models, textbooks, and risk assessments. The technical decisions about archive schemas are becoming instruments of long-term scientific influence.
What isn’t being said
Official statements emphasise transparency and shared discovery. They leave unanswered how commercially valuable data fits into these regimes. Resource grade readings, infrastructure siting surveys, and payload diagnostics may remain in proprietary or export-controlled channels even as scientific datasets are released. The gap matters because it is often the industrial data, not the press-release images, that determines who finances and builds lunar industries.
The regional split
US allies embedded in Artemis gain access to future human missions and AI-ready archives, but may find deep technical cooperation with ILRS partners more difficult. States closer to Beijing and Moscow receive hardware and financing through ILRS, but risk reduced visibility in tools and models built around PDS4. The divergence is less about picking a side than about recognising which ecosystem will best serve each country’s scientific and industrial ambitions over the next decade.
The data architecture is now everyone’s problem
With the reference architecture published and the division between Artemis and ILRS data systems hardening, the following groups face decisions in the coming months.
- Western space agency data manager
Assess your mission data pipelines against NASA’s PDS4 reference architecture now. The workshops provided a working template for structuring, curating, and publishing lunar datasets — adopting the same standard ensures interoperability with the 71-nation coalition and compliance with Artemis Accords commitments. NASA’s Science Mission Directorate Scientific Information Policy (SPD‑41a) and its open-science guidance are published on the agency’s site.
- Aerospace and defense investor
Evaluate how the data bifurcation affects the competitive landscape for lunar ventures. Companies aligned with the Artemis framework gain an AI-ready data ecosystem and a large user base, but face barriers to data from ILRS-aligned missions. Ventures tied to ILRS may have privileged access to Chinese and Russian datasets but limited interoperability with Western models. Track ILRS standards announcements over the next 12–24 months — if a formal data policy emerges, it will reshape the risk calculus.
- Lunar mission hardware and software developer
Incorporate PDS4 compatibility into your product roadmaps. With 71 nations now anchored to the standard, instruments and software designed for Artemis-aligned missions must output data in PDS4-compliant formats. The reference implementation is documented on NASA’s Planetary Data System site, and the peer-reviewed curation pipeline defines the minimum metadata your payloads will need to generate.
- Planetary science researcher in an Artemis Accords nation
Familiarise yourself with PDS4 and the NASA-IBM Lunar Foundation Model on Hugging Face. The harmonised datasets from multiple lunar missions are now accessible in a unified format, and the foundation model provides a pre-trained tool for crater mapping, volcanic feature identification, and polar ice stability estimation. Future proposals that leverage the shared data architecture will be positioned ahead of those that rely on fragmented sources.
Explainer
- PDS4
- The fourth-generation information model of NASA’s Planetary Data System, an XML-based metadata schema that describes planetary science data for cross-mission search, validation, and reuse. It mandates non-proprietary storage formats and ties every data field to controlled vocabularies and discipline dictionaries. The model was designed to prevent the kind of data loss identified in a 1982 National Academy of Sciences review, and its structured labels enable AI models to ingest and harmonise data from entirely different instruments and decades.
- Artemis Accords
- A set of non-binding bilateral commitments between the United States and 71 partner nations, launched in 2020, that establish principles for civil space exploration — including transparent operations, interoperability, and open sharing of scientific data from government activities. They are not a multilateral treaty, and they explicitly exclude China and Russia. The Accords’ data-sharing provisions do not extend to private-sector operations unless conducted on behalf of a signatory government.
- International Lunar Research Station
- A joint lunar base program led by China and Russia, formalised by a March 2021 memorandum of understanding, with construction targeted between 2026 and 2035. The initiative operates outside the Artemis Accords framework and maintains its own data-sharing arrangements through China’s Lunar and Planetary Data System. Its governance approach emphasises sovereignty and partner integration independent of US-led norms, creating a separate technical and political architecture for lunar science.
- China’s Lunar and Planetary Data System
- The central archive for Chinese lunar and deep-space mission data, operating under the China National Space Administration. It divides mission datasets into proprietary, protection, and public periods, with CNSA retaining discretion over what becomes publicly available and when. While the system states openness and sharing as a principle, its phased access model differs from NASA’s mandate of fee-free, full public release at publication or funding closeout under SPD‑41a.
- FAIR principles
- A set of data management standards — Findable, Accessible, Interoperable, and Reusable — published in 2016 by researchers and publishers to guide public research data stewardship. NASA’s Science Mission Directorate requires SMD-funded mission data and software to meet FAIR requirements, and the PDS4 information model serves as a concrete institutional-scale implementation of the principles, ensuring lunar data remains usable decades after collection.
RELATED STORIES
