Browse
All nodes
312 nodes — click any to read the body, follow edges, and see backlinks.
QuestionQuestion · 8
- Q-0001Is research data loss architectural or operational?
- Q-0002Under what architectural conditions does research data survive long horizons?
- Q-0003What does research data loss cost institutions and science?
- Q-0004What is the economic cost structure of preservation across architectural tiers?
- Q-0005What architecture produces verification evidence as a byproduct of operation?
- Q-0006What becomes possible if Tier 3 is the operating standard?
- Q-0007How does Tier 3 relate to AI-era research data requirements?
- Q-0008What concrete actions should institutions, funders, and researchers take?
ClaimClaim · 52
- C-0001Research data loss is architectural, not operational
- C-000273 to 93 percent of published research cannot produce its underlying data on request
- C-0003Funder mandates are shifting from self-reported plans to programmatic verification
- C-0004Tier 2 coordinated preservation works but does not extend to most research
- C-0005A representative R1 carries ~$1.1B/year in unverifiable research output as latent liability
- C-0006Marginal cost of adding Tier 3 to existing institutional infrastructure is effectively zero
- C-0007Tier 3 is the only architecture that generates verifiable compliance evidence as a byproduct
- C-0008Tier 3 produces the AI-ready data substrate as a structural byproduct of preservation
- C-0009Three architectural properties jointly determine long-horizon survival
- C-0010Tier 0 produces the documented year-over-year decline in dataset survival by default
- C-0011Tier 1 is one organizational decision away from the same outcome as Tier 0
- C-0012Tier 2 redundancy is contingent on continued coordination of 3-4 organizations
- C-0013Tier 3 produces redundancy as a structural byproduct of use
- C-0014Content addressing operates on any byte sequence regardless of semantics
- C-0015Permissioned Tier 3 configurations restrict access while preserving redundancy and integrity
- C-0016Architectural tier is determined by the deployment, not by the underlying software
- C-0017Personnel turnover is a structural preservation event for research data
- C-0018Single-domain backups are not backups
- C-0019Grant termination terminates data maintenance
- C-0020Platforms discontinue or restrict access on timelines researchers do not control
- C-0021Every loss mode reduces to single-copy dependency
- C-0022Reproducibility failure is the downstream signature of data loss
- C-0023The scholarly record itself is decaying through reference rot
- C-0024Per-paper liability application — Agh 2009 yields $725K-$1.88M with effectively-infinite B
- C-0025Three independent vectors are converging on programmatic verification of research data compliance
- C-0026Faculty flight and failed recruiting compound continuously, independent of liability surfacing
- C-0027The institutional position is an unhedged billion-dollar tail exposure against a rounding-error fix
- C-0028Tier 2 fragility takes three operational forms
- C-0029Coverage extension to the 73-93% requires architecture below domain-specific governance
- C-0030Tier 2 economics produce the surviving-consortia pattern
- C-0031ROI on research data infrastructure is positive in every documented study, ranging 5x to 800x
- C-0032Open access multiplies the return on preserved data infrastructure
- C-0033Data management plans and infrastructure are structurally disconnected
- C-0034Content addressing makes "my data is lost" a testable rather than unfalsifiable defense
- C-0035Universal coverage becomes possible at Tier 3
- C-0036Preservation horizon decouples from project budget
- C-0037Reanalysis becomes a first-class research activity
- C-0038AI is the institutional priority of 2026
- C-0039Institutions that deploy Tier 3 in 2026 hold three positions simultaneously
- C-0040R1 — Conduct an architectural audit of existing data infrastructure
- C-0041R2 — Deploy at least one protocol-level preservation node
- C-0042R3 — Integrate compliance evidence generation into the data deposit workflow
- C-0043R4 — Funders should require verifiable evidence of preservation, not self-reported plans
- C-0044R5 — Fund preservation through facilities and administrative cost recovery
- C-0045R6 — Maintain local clones and content-addressed copies of all research data
- C-0046R7 — Publish reference deployments, audit templates, and cost models
- C-0047The most expensive data infrastructure is the infrastructure built after the disaster
- C-0048FOIA and disclosure regimes can surface architectural data unavailability without an enforcement action
- C-0049Domain-specific Tier 2 carries hidden fragility — bespoke = smaller community = SPOF in resilience disguise
- C-0050Tier 1 cost-affordability is deceptive — a single surfacing event exceeds a century of repository fees
- C-0051Verification shifts from forensic reconstruction to inspection query under Tier 3
- C-0052Useful life of preserved data extends to horizon of scientific interest under Tier 3
EvidenceEvidence · 112
- E-0001Vines 2014 — 19% of 516 ecology dataset requests delivered
- E-0002Gabelica 2022 — 93% non-compliance among 1,792 biomedical authors who had committed to share
- E-0003Wicherts 2006 — 73% non-compliance, 141 APA psychology papers
- E-0004Acciai 2023 — 86% non-sharing, 1,634 PNAS and Nature-portfolio papers from 2017-2021
- E-0005Conditional dataset survival decays ~17% per year after publication
- E-0006191 research data repositories shut down 2012-2023; median age 12; 47% no migration
- E-0007U.S. preclinical research consumes ~$28B/year on irreproducible work
- E-0008Long-running protocol systems have survived multi-decade horizons without organizational continuity
- E-0009GitHub blocked developers in 5 sanctioned regions in July 2019
- E-0010kernel.org compromised 2011; Linux source not at risk because of distributed Git copies
- E-0011Median PhD time 7.3 years; postdoc avg 4.5 years; 15-23% tenure-track placement
- E-001265% of popular GitHub projects have bus factor ≤ 2
- E-0013HLRS Stuttgart — 57 of 262 user accounts de-registered, ~619 TB orphaned
- E-0014Agh 2009 Artemia — laptop theft destroyed only copy of irreplaceable dataset
- E-0015Kyoto University December 2021 — buggy backup script deleted 77 TB across 14 research groups
- E-0016Brazil National Museum fire 2018 — ~18.4M of 20M items destroyed
- E-0017NIH 2,291 grants ($2.45B) and NSF 1,752 grants ($1.4B) terminated Feb-Aug 2025
- E-0018NOAA Billion-Dollar Disasters database stopped updating; Mauna Loa CO₂ record proposed for defunding
- E-0019~3,400 datasets removed from Data.gov and 14 NOAA datasets decommissioned by Feb 2025
- E-0020191 of 411 long-term mammal studies terminated; 63-year marmot record cut
- E-0021NASA Astronomical Data Center terminated 2002 with no formal successor
- E-0022UK Arts and Humanities Data Service closed 2008 at 12-year median
- E-0023BIIACS social-sciences repository went dark December 2023 despite "perpetuity" pledge
- E-0024Twitter eliminated free academic API Feb 2023; 33,306 studies built on Twitter data
- E-0025GISAID suspended researcher accounts in 2023 and revoked Open Access designation
- E-0026China National Knowledge Infrastructure cut off foreign subscribers April 2023
- E-0027CERN terminated Russia/Belarus cooperation Nov 2024; ~500 scientists expelled
- E-0028UK Biobank moved to metered cloud-only access in 2023-24, ~doubling project costs
- E-0029Elsevier acquired Bepress 2017; >500 universities' OA infrastructure under publisher ownership
- E-0030Mendeley Desktop EOL 2022; 2018 update lost user PDFs and annotations
- E-0031Academia.edu paywall escalated $99→$498/year; 40% of users in developing nations
- E-0032Elsevier and Springer route TDM through gated APIs; institutional contract violations reported
- E-0033Nature survey of 1,576 researchers — >70% failed to reproduce others' experiments
- E-0034Amgen — 11% replication success of 53 landmark cancer biology studies
- E-0035Molecular Brain — 97% of authors couldn't produce raw data; 21 papers withdrawn
- E-0036Stern et al. 2014 — NIH-funded retracted papers carry $392,582 mean attributable cost
- E-003774% of published R analysis files fail to execute without error
- E-0038Pew Research — 25% of webpages 2013-2023 are gone; 38% at 10 years
- E-0039One in five scientific articles suffers reference rot; 7-in-10 web-citing have compromised context
- E-0040Harvard Law Review — >70% of URL citations 1996-2012 no longer resolve
- E-0041Duke University $112.5M FCA settlement (March 2019) — largest university FCA payment
- E-0042Harvard-Brigham/Anversa $10M FCA settlement (April 2017) — 31 retractions
- E-0043Dana-Farber $15M FCA settlement (December 2025) — most recent precedent
- E-0044NIH DMS Policy effective Jan 2023 + simpler standardized format effective May 2026
- E-0045Gates Foundation contracted OA.Works for programmatic compliance review (Jan 2025)
- E-0046Adjacent industries already cross verification threshold — SEC, HIPAA, FDA Part 11
- E-0047Nature 2025 poll — 75% of US researchers considering leaving the country
- E-00488% declared, 2% actual — 2.1M-article compliance gap
- E-0049INSDC — 53.9 trillion bases across 3 continents, continuously operational since 1980s
- E-0050wwPDB — 4 sites, 50+ years, $23B replacement, 100% of 34 cancer drugs 2019-2023
- E-0051WLCG — 1.5 EB, 170+ sites, 42 countries, 2M+ tasks/day
- E-0052NOAA NCEI — 60+ PB across 4 US locations; data confirmed safe after Hurricane Helene 2024
- E-0053Astronomical archives — STScI/CADC/ESAC 30-year IVOA partnership
- E-0054Digital Preservation Network dissolved 2018; integrity layer dissolved with coordinator
- E-0055MetaArchive Cooperative sunset 2025; silent under-replication, manual Stanford rebuild
- E-0056Self-hosted repository costs — 58-96% staff; MIT DSpace ~$260K/yr; Southampton ePrints £116K/yr
- E-0057Universities running protocol nodes — TU Dortmund, TU Dresden, MIT SIPB, Tor relays at 45+
- E-0058Protocol-node baseline cost ~$46/yr; protocol-daemon memory footprints near zero
- E-0059EMBL-EBI ROI 20-26x on £50M/yr (£1B-£1.3B annual user value)
- E-0060PDB ROI 800x — $5.5B economic impact / $6.1M federal funding/yr
- E-0061Landsat free-access — 53 scenes/day → 5,700/day; $25.6B/yr economic value
- E-0062COVID-19 vaccines — $1.9T BioNTech / $5.2T total enabled by open-genome publication
- E-0063Smeesters/Erasmus 2014 — "computer crash" defense in misconduct case
- E-0064~$3.3B/yr federal nondefense AI R&D + $2B+ DARPA + DOE Genesis Mission $320M
- E-0065UK Archaeology Data Service — £13M/yr efficiency gains, ~5:1 return
- E-0066Australia NCRIS — $7 returned per $1 invested
- E-0067XSEDE cyberinfrastructure — $4.7B-$22.7B value on $257.5M; 18:1 to 88:1
- E-0068Apon et al. — $14.3M HERD per $100K research-computing salary; $1.3M per 100 TFLOPs
- E-0069Human Genome Project — $14.5B federal investment 1988-2012 → $965B economic impact
- E-0072European Commission — minimum €10.2B/yr cost of not having FAIR research data
- E-0073Server idle capacity — only 40% of data centers measure utilization; 25% of physical servers comatose; 12-18% typical
- E-0074Networks run at ~26% average utilization globally; Internet2 maintains 50% headroom by policy
- E-0075Internet2 bandwidth contracts are flat-rate; 50% headroom maintained by policy; augment trigger at 40% utilization
- E-0076Median U.S. higher-ed central IT spend ~$10.6M/yr; compensation ~half, nearly entirely fixed
- E-0077Cabling is 61% of campus networking infrastructure cost; physical plant amortizes over 10-15+ years
- E-0078Hosted institutional repository pricing — DSpaceDirect $4K-$9K/yr; Digital Commons $10K-$12K/yr
- E-0079Cloud archival storage — AWS S3 Glacier Deep Archive ~$12/TB/yr; GCS Archive ~$14/TB/yr
- E-0080CLOCKSS — $550-$18,350/yr scaled to library budget; 20 years operational
- E-0081LOCKSS Global Network — $2,642-$13,222/yr + ~$700/yr per node; 27 years operational
- E-0082HathiTrust — $6,600-$13,000/yr scaled to total library expenditure; 18 years operational
- E-0083APTrust — $20,000/yr + $420/TB/yr (first 10 TB included)
- E-0084Portico — $1,500-$25,462/yr scaled; 1,076 publishers and 1,288 supporting libraries
- E-0085U.S. academic R&D ~$108.8B FY2023 across 914 institutions; top 30 capture 42%
- E-0086Mid-sized R1 reference profile — ~$200M R&D, 1,000-1,500 TT faculty, ~3,000 publications/yr
- E-0087NIDDK extramural human-subjects share ~40% sustained FY2014-FY2023
- E-00882025 Carnegie R1 threshold — $50M R&D and 70 doctorates; 187 institutions designated
- E-0089Horizon Europe — FAIR data mandatory, no opt-out; Article 17 ties compliance to payments
- E-0090Wellcome Trust — may decline new applications from non-compliant researchers; may suspend institutional funding
- E-0091NSF — proposals submitted without a Data Management Plan are returned without review
- E-0092OSTP Nelson Memo (Aug 2022) — immediate open access to federally funded research and data; >$90B/yr
- E-0093NSPM-33 implementation — research-integrity disclosure as the procedural template for verifiable evidence
- E-0094NASEM 2020 — current research-funding regime is structurally misaligned with multi-decade preservation
- E-0095EDUCAUSE 2025 Top 10 IT Issues — AI/data ranked
- E-0096NSF NAIRR Pilot — launched Jan 2024 with 10 fed agencies + 25 partners; expanded by 2026 to 14 + 28
- E-0097National AI Initiative Act 2020 + CREATE AI Act pending — legal foundation under which NAIRR operates
- E-009829 NSF National AI Research Institutes funded at ~$20M each over 5 years
- E-0099DOD Chief Digital and AI Office + DARPA AI Next/Forward — $2B+ since 2018 in defense AI
- E-0100AI faculty market — 2-3x academic-vs-industry compensation differential drives retention pressure
- E-0101European Open Science Cloud — EU Node operational since Oct 2024; EOSC Federation in 2025-26 build-up
- E-0102EuroHPC Federation Platform — first release April 2026; AI Factories program operational
- E-0103China Science and Technology Cloud — CAS operates 11 of 20 national scientific data centers
- E-0104Software Heritage — 27B source files from 421M projects under content-addressed Merkle DAG; SWHID is ISO/IEC 18670
- E-0105Westbrook & Burley 2019 — 88% of 210 new FDA molecular entities 2010-2016 facilitated by PDB structures
- E-0106Choose Europe for Science — €500M launched May 2025, expanded to ~€900M; recruiting US researchers
- E-0107Only 44% of U.S. faculty report adequate institutional IT support for grant-funded projects
- E-0108Purdue research computing — 49x return; 55% of $443.5M FY2020 R&D attributable to RCAC HPC users
- E-0109GitHub scale — 630 million repositories, 180+ million developers
- E-0110Colavizza 2020 — papers with data-availability links to repository data receive up to 25.36% more citations
- E-0111Asem et al. 2019 — >90% A. urmiana population loss 1994-2004 with associated genetic erosion
- E-0112Piwowar & Vision 2013 — 9% open-data citation advantage; ~150 reuse papers per 100 deposited datasets
- E-0113SARS-CoV-2 genome publicly shared Jan 11 2020 via GISAID; BioNTech Project Lightspeed launched Jan 27 2020
- E-0114Piwowar & Vision 2013 — 9% open-data citation advantage; 150 reuse papers per 100 deposited datasets in 5 years
MethodMethod · 6
SourceSource · 134
- S-0001Vines et al. 2014 — The Availability of Research Data Declines Rapidly with Article Age
- S-0002Gabelica et al. 2022 — Many researchers were not compliant with their published data sharing statement
- S-0003Wicherts et al. 2006 — The poor availability of psychological research data for reanalysis
- S-0004Acciai, Schneider & Nielsen 2023 — Estimating social bias in data sharing behaviours
- S-0005Strecker et al. 2023 — Disappearing repositories
- S-0006Server utilization studies (Uptime Institute, NRDC, LBNL, Anthesis/Stanford)
- S-0007TeleGeography State of the Network 2023-2025
- S-0008Internet2 fee model and capacity-augment policy
- S-0009Freedman, Cockburn & Simcoe 2015 — The Economics of Reproducibility in Preclinical Research
- S-0010RFC 6920 — Naming Things with Hashes
- S-0011Software Heritage Activity Report 2025
- S-0012INSDC — Arita et al. 2025
- S-0013wwPDB — Berman, Henrick & Nakamura 2003 + RCSB economic impact
- S-0013aWestbrook & Burley 2019 — PDB contributions to FDA drug approvals 2010-2016
- S-0014CLOCKSS Archive preservation statistics
- S-0015Worldwide LHC Computing Grid (CERN)
- S-0016DNS scale and reliability metrics
- S-0017SMTP — RFC 821 (1982) and RFC 5321 (2008)
- S-0018Email statistics 2026 (Radicati / Statista / EmailToolTester)
- S-0019BitTorrent 2 billion installations milestone (2020)
- S-0020Internet Archive uses BitTorrent for over 1M items
- S-0021GitHub Octoverse Report 2025 + Git release history
- S-0022Stack Overflow 2024 Developer Survey — 93.87% Git adoption
- S-0023kernel.org compromise 2011 (Corbet, Linux Foundation)
- S-0024GitHub Octoverse Report 2025 (scale)
- S-0025GitHub blocked sanctioned-region developers 2019
- S-0025aGitHub email refusing repository export to sanctioned developer
- S-0026Researcher career timelines (NCSES SED 2023, CGS 2008, Kahn & Ginther 2017, Woolston 2020)
- S-0027Avelino et al. 2016 — Truck factor analysis of 133 GitHub projects
- S-0028Schembera & Durán 2020 — Dark Data at HLRS Stuttgart
- S-0029Kyoto University IIMC 77TB loss notice (Dec 2021)
- S-0030Brazil National Museum fire — Escobar 2018, AHA Perspectives, Smithsonian Magazine
- S-0031NIH/NSF 2025 grant terminations
- S-0032NOAA Billion-Dollar Disasters retirement + Mauna Loa defunding proposal
- S-0033Data.gov dataset removals 2025 (EDGI, Data Rescue Project, Harvard LIL)
- S-0034Blumstein 2025 — The end of long-term ecological data?
- S-0035NASA Astronomical Data Center closure 2002
- S-0036AHRC AHDS closure decision 2007-2008
- S-0037BIIACS shutdown December 2023
- S-0038Twitter API termination impact analysis
- S-0039GISAID — governance and access controversies
- S-0040CKNI cross-border subscription cutoff April 2023
- S-0041CERN ended cooperation with Russian/Belarusian institutions Nov 2024
- S-0042UK Biobank cloud-only metered access transition
- S-0043Elsevier acquisition of Bepress (2017)
- S-0044Mendeley Desktop EOL September 2022 + 2018 update annotation loss
- S-0045Academia.edu pricing escalation timeline
- S-0046Elsevier and Springer Nature publisher-gated TDM APIs
- S-0047Nature reproducibility survey 2016 (Baker)
- S-0048Begley & Ellis 2012 — Amgen replication of cancer biology
- S-0049Miyakawa 2020 — Molecular Brain editor's raw data audit
- S-0050Stern, Casadevall, Steen & Fang 2014 — Financial costs of NIH retractions
- S-0051Trisovic et al. 2022 — Reproducibility of computational research code
- S-0052Pew Research Center 2024 — Web decay
- S-0053Klein et al. 2014 — Scholarly Context Not Found
- S-0054Zittrain, Albert & Lessig 2014 — Perma.cc / link rot in legal scholarship
- S-0055Piwowar & Vision 2013 — Open data and citation/reuse advantage
- S-0056Agh et al. 2009 — Iranian Artemia morphometric and genetic characterization
- S-0057CORDIS — EU ICA4-CT-2001-10020 award record (Artemia INCO-DEV)
- S-0058Urmia Lake collapse documentation
- S-0059Colavizza et al. 2020 — Open-data citation advantage in PLOS+BMC corpus
- S-0060Asem et al. 2019 — Post-collapse genetics of Artemia urmiana
- S-0061NCSES HERD FY2023 + 2025 Carnegie classification
- S-0062R1 faculty-and-publication scaling reference
- S-0063NIDDK extramural human-subjects share FY2014-FY2023
- S-0064Harvard / Brigham / Anversa $10M FCA settlement (April 2017)
- S-0065Duke / Potts-Kant $112.5M FCA settlement (March 2019)
- S-0066Dana-Farber $15M FCA settlement (December 2025)
- S-0067NIH Data Management and Sharing Policy (NOT-OD-21-013)
- S-0068Gates Foundation OA.Works programmatic compliance (Jan 2025)
- S-0069Horizon Europe FAIR data mandate, Grant Agreement Article 17
- S-0070Wellcome Trust compliance framework
- S-0071NSF DMP requirement — proposals returned without review
- S-0072OSTP Nelson Memo (August 2022)
- S-0073SEC Rule 17a-4 records retention requirements
- S-0074SEC/CFTC/FINRA $3.5B+ in records-related fines since 2021
- S-0075HIPAA Security Rule penalties
- S-007621 CFR Part 11 + Applied Therapeutics rejection
- S-0077Financial-sector cybersecurity IT-budget allocation
- S-0078Nature 2025 reader poll + Choose Europe for Science
- S-007944% of US faculty report adequate institutional IT support
- S-0080Standalone Tier 3 protocol-node hosting costs
- S-0081PDB and FDA cancer drug approvals 2019-2023
- S-0082NOAA NCEI 60+ PB across 4 sites; Hurricane Helene 2024
- S-0083STScI/CADC/ESAC astronomical archive partnership (IVOA)
- S-00848% declared / 2% actual data availability across 2.1M articles
- S-0085EDUCAUSE Core Data Service — median IT spend ~$10.6M; staffing 8% turnover
- S-0086Higher-ed physical-plant cost amortization + cabling 61% of campus networking
- S-0087LYRASIS DSpaceDirect / Digital Commons hosted-repository pricing
- S-0088Self-hosted repository cost analyses (MIT DSpace, Southampton ePrints)
- S-0089Cloud archival storage pricing — AWS S3 Glacier, Google Cloud Archive
- S-0090CLOCKSS membership pricing
- S-0091LOCKSS Global Network membership
- S-0092HathiTrust membership
- S-0093APTrust pricing
- S-0094Portico pricing
- S-0095Digital Preservation Network dissolution 2018
- S-0096MetaArchive Cooperative sunset 2025
- S-0097BitTorrent daemon footprint + Academic Torrents
- S-0098Tor relay requirements + 45+ universities running relays
- S-0099BitTorrent WebSeed specification
- S-0100TU Dortmund Fachschaft Informatik Matrix homeserver
- S-0101TU Dresden Matrix for 18,000 users
- S-0102MIT SIPB-operated Mastodon and Forgejo
- S-0103Labovitz & Chang 1992 — 1:10:100 quality cost rule
- S-0104EMBL-EBI economic-impact study (~£50M/yr → £1B-£1.3B/yr)
- S-0105UK Archaeology Data Service — £13M/yr efficiency gains
- S-0106Australia NCRIS — $7 returned per $1 invested
- S-0107XSEDE — $4.7B-$22.7B value on $257.5M investment
- S-0108Apon et al. — $14.3M HERD per $100K research-computing salary
- S-0109PDB ROI — $5.5B annual impact / $6.1M federal funding
- S-0110Carnegie 2025 R1 threshold — $50M R&D, 70 doctorates
- S-0111Human Genome Project + genomics — $14.5B → $965B economic impact
- S-0112Landsat — 53/day → 5,700/day; $25.6B/yr economic value
- S-0113COVID-19 vaccine value — $1.9T BioNTech / $5.2T total
- S-0114European Commission — €10.2B/yr cost of not having FAIR data
- S-0115Erasmus University 2012 / 2014 Smeesters report
- S-0116NASEM 2020 — Funding system not conducive to data life-cycle cost forecasting
- S-0117NSPM-33 implementation guidance — research integrity / COI disclosure
- S-0118EDUCAUSE 2025 Top 10 IT Issues — Data-Empowered Institution
- S-0119NSF NAIRR Pilot launch (January 24, 2024)
- S-0120NAIRR Pilot expansion + NAIRR-Open / NAIRR-Secure focus areas
- S-0121National AI Initiative Act of 2020 (Thornberry NDAA FY2021)
- S-0122CREATE AI Act (pending in Congress)
- S-012329 NSF National AI Research Institutes funded
- S-0124DOE Genesis Mission EO November 2025 + $320M+ DOE AI investments
- S-0125DOD Chief Digital and AI Office
- S-0126DARPA AI Next + AI Forward — $2B+ since 2018
- S-0127Aggregate federal nondefense AI R&D ~$3.3B/yr
- S-0128NSF National AI Research Institutes solicitation NSF 23-610
- S-0129AI faculty market — academic vs industry compensation differential
- S-0130European Open Science Cloud (EOSC)
- S-0131EuroHPC Federation Platform + AI Factories
- S-0132China Science and Technology Cloud — 11 of 20 NSDCs