space · manufacturing · digital · family: the solution exists but nobody can afford it
space computersfall short
Space computers fall short of what NASA's next missions need, and NASA ranks high-performance onboard computing its third-highest civil space shortfall
Problem statement
Spacecraft computers must function reliably in radiation environments — galactic cosmic rays, solar protons and heavy ions, and trapped protons and electrons — that produce single-event effects, total ionizing dose, and displacement damage, and whose tolerance NASA says commercially available solutions might not naturally provide. NASA's own solicitation says existing space-grade coprocessor and accelerator devices "are not yet powerful enough to meet the performance requirements of NASA's next-generation systems," and envisions a 100X increase in computational capability at the same power for the next generation, while the missions they must support — increased autonomy, real-time processing of sensor data, onboard AI — demand more computing capability. What NASA names as missing is foundational: the technologies to design, fabricate, and package the next generation of radiation-hardened microelectronics, and the libraries and tools to put space microelectronics onto emerging semiconductor fabrication processes. This write-up's working hypothesis is that the market for such parts is too small to fund that work; none of its sources state a market size, a lag in years, or a supplier count.
Why this matters
NASA's #3 and #6 ranked civil space shortfalls are both computing-related: high-performance onboard computing and extreme-environment avionics. Signals can take as much as 24 minutes each way to Mars, part of a communications scenario that NASA says drives a need for increased autonomy — autonomous decisions made on space-grade coprocessor and accelerator devices that NASA's solicitation calls not yet powerful enough, while the commercial accelerators built for self-driving cars, data centers, and smartphones are continuously being developed. For the Europa Clipper, with a mission life of 10 years, the radiation environment is estimated at 2.9 Mrad total ionizing dose behind 0.1-inch-thick aluminum. In Europe, ESA, the European Commission and the European Defence Agency agreed in 2008 to join forces to develop critical space technologies in Europe; ESA's announcement calls the development of space technologies for ensuring non-dependence "of strategic importance for Europe and for its role as a major space power and credible international partner."
What’s been tried and why it hasn’t worked
Radiation hardening by design (RHBD) — parts custom-built to military specification and assured for complete radiation immunity, in Goddard's description — is the traditional route, and NASA still lists RHBD libraries and tools among its foundational avionics shortfalls. Commercial-off-the-shelf (COTS) approaches — off-the-shelf parts designed, assessed, or tested for tolerance of a given radiation environment rather than immunity — are where Goddard says the industry is now pivoting. NASA's shortfall list, in its entry on high-rate communications across the lunar surface, names among the key challenges characterizing the radiation tolerance of COTS system-on-chip components, which are increasingly difficult to replace due to their sophistication and power efficiency, and architecting around them to yield system reliability sufficient for high-criticality applications; its solicitation adds that commercial accelerator devices can be leveraged, but that a redundant system design with health monitoring is needed.
What would unlock progress
NASA's shortfall descriptions and solicitation frame the needs as foundational technology: (1) libraries and tools that let the next generation of space microelectronics be implemented on emerging semiconductor fabrication processes; (2) testing technologies that make COTS device testing more efficient and effective, so that leading-edge COTS devices can be used more broadly in avionics; (3) for compute offload, either reliable use of COTS coprocessors in space systems or fault-tolerant IP cores implemented in a radiation-hardened FPGA, with a redundant, health-monitored system design; and (4) open, modular avionics architectures that interoperate across vendors. ESA's parallel strategic approach, announced in 2008, is a joint EC/ESA/EDA task force to continue developing the list of critical space technologies for a Europe-wide approach.
Entry points for student teams
The two radiation effects need two different machines, and only one of them is anywhere near a student team. A cobalt-60 source gives total ionizing dose only — dose-accumulated drift in a commercial microcontroller's current draw, clock stability, and memory retention — and that arm requires a university health-physics or nuclear-engineering department that already holds a licensed Co-60 irradiator and will supervise the exposures; single-event upset rates cannot be measured there at all, because they need heavy-ion or proton beam time at one of a handful of facilities (Texas A&M's Cyclotron Institute Radiation Effects Facility, https://cyclotron.tamu.edu/ref/, runs dedicated heavy-ion and proton lines for this and schedules beam time by request), which is a funded-program access, not a semester one. The door that needs no beam and no license is the software-mitigation half done against real cross-sections: NASA Goddard's radiation database publishes decades of its own test reports openly (https://nepp.nasa.gov/radhome/RadDatabase/RadDataBase.html), so a team can take measured SEU cross-sections for a specific commercial part family, fold them through a chosen orbit's particle environment, and then evaluate TMR voting, checkpoint/rollback, and watchdog schemes by fault injection in emulation — reporting the compute, power, and mass cost of each mitigation at a given upset rate. The architecture arm is equally tractable without a facility: design and simulate a hardened supervisory core with health monitoring managing a higher-performance commercial compute core, and map how much radiation hardening can be delegated to architecture rather than fabrication.
Genes
Structural cousins: stuck the same way, in other fields
Sources
"Civil Space Technology Shortfall Ranking," NASA STMD, July 2024. (accessed 2026-02-14). Shortfall #3 (score: 7.4345) and #6 (score: 7.2076). Supplemented with the NASA Fiscal Year 2025 SBIR Phase I Solicitation (80NSSCFY25SBIRPhaseI; ), subtopics Z-ENABLE.02 ("High-Performance Space Computing Technology"), S13.03 ("Extreme-Environments Technology") and H15.01 ("Autonomous Capabilities for Lunar Surface Mobility Systems"). (accessed 2026-02-14; solicitation PDF read 2026-09-30). Also "Critical space technologies for European strategic non-dependence," ESA corporate news on the EC/ESA/EDA joint effort, 19 September 2008. (accessed 2026-02-14). go to source 1 ↗ go to source 2 ↗ go to source 3 ↗ go to source 4 ↗
Working notes
Notes made while writing and checking this brief, shown as written. They use working shorthand.
- The `temporal:static` tag is justified because the fundamental physics of radiation damage to semiconductors does not change. What changes is the process node — smaller transistors are more radiation-sensitive, creating a race between commercial shrinkage and hardening techniques.
- The `constraint:supply-chain` tag is critical: in 2008 ESA, the EC and EDA called the development of space technologies for ensuring non-dependence of strategic importance for Europe. The sources read for this write-up do not give a count of qualified rad-hard processor suppliers.
- The `failure:tech-limitation-now-resolved` tag reflects the pivot Goddard describes toward commercial-first spaceflight architectures that use off-the-shelf parts assessed or tested for tolerance rather than custom RHBD parts.
- Cross-domain connection: this shares the supply-chain concentration structure with `critical-minerals-waste-extraction` and `energy-grid-transformer-supply-chain-crisis` — a small number of specialized producers controlling a critical input.
- The small size of the rad-hard market is this write-up's hypothesis for the structural constraint; none of the sources read give a market figure.
Reconciliation 2026-08-21: Entry-point realism (C37 triage, score 1) — flag CONFIRMED, and the physics error triage named is real: the section told teams to characterize "single-event upset rates and total ionizing dose thresholds" at "a university radiation facility (many physics departments have small accelerators or cobalt-60 sources)," conflating two incompatible test capabilities. A Co-60 gamma source produces total ionizing dose only; SEU cross-sections require heavy-ion or proton beam time, available at a small number of facilities. The claim is now split: the TID arm is kept with an explicit access line (a department holding a licensed Co-60 irradiator, with supervision), and the SEU arm is named as beam-facility access — Texas A&M's Cyclotron Institute Radiation Effects Facility (https://cyclotron.tamu.edu/ref/, fetched: "dedicated beam lines for both heavy ion and proton testing," serving commercial, governmental, and educational customers, beam time scheduled by request; pricing not stated on the page and not asserted here). Unflagged-door check: the chiplet-architecture door was already facility-free and was kept, tightened only in wording. The new facility-free carrier is the software-mitigation half rebuilt on published cross-sections rather than on beam time a team will not get — NASA Goddard's Radiation Effects and Analysis Group publishes its radiation test reports in an openly browsable database (https://nepp.nasa.gov/radhome/RadDatabase/RadDataBase.html, fetched; reached via https://etd.gsfc.nasa.gov/capabilities/capabilities-listing/radiation-effects-and-analysis/, which describes 30+ years of archived characterization data and an open-data commitment; the old radhome.gsfc.nasa.gov address 301-redirects there). Declined to cite: ESA's ESCIES and the CREME96 online tool, neither of which was fetched or confirmed open in this session, and no vendor or university irradiator is named because none was verified.
Reconciliation 2026-09-30: Whole write-up read against every source it names, all fetched 2026-09-30 by a first reader and again, independently, by a second reader the same day, who went over each repair, undid none, and reworded several where the new sentence still said more than the source. NASA's shortfall page (https://www.nasa.gov/spacetechpriorities/) was read in full, with the two PDFs it links: the July 2024 Civil Space Shortfall Ranking (civil-space-shortfall-ranking-july-2024.pdf) and the July 2024 shortfall descriptions, both in full. The SBIR landing page (https://www.nasa.gov/sbir_sttr/) carries no subtopic text; the Fiscal Year 2025 SBIR Phase I Solicitation PDF (80NSSCFY25SBIRPhaseI, at the address now on the Source line) was read in full, as were the 2022, 2023 and 2024 Phase I solicitations as a check. The ESA page, the Texas A&M Radiation Effects Facility page and the Goddard Radiation Effects and Analysis page were read in full; the NEPP Radiation Data Base page timed out for the second reader and was read from its Internet Archive copy. Confirmed in the sources and left as written: the July 2024 ranking places 1554 High Performance Onboard Computing to Enable Increasingly Complex Operations third at 7.4345 and 1552 Extreme Environment Avionics sixth at 7.2076; signals take as much as 24 minutes each way to Mars (subtopic H15.01); Europa Clipper's 10-year mission life and 2.9 Mrad TID behind 0.1-inch aluminum (subtopic S13.03); Texas A&M's dedicated heavy-ion and proton beam lines and beam-time request; Goddard's 30+ years of archived test data, its open-data commitment, the database of Goddard-only test reports, and the radhome.gsfc.nasa.gov redirect. Changes: (1) The long title said space computers are 20 years behind and only two companies can make them, and the display title said "Two Companies, Twenty Years Behind." No source gives a lag in years or a supplier count. The long title is now "Space Computers Fall Short of What NASA's Next Missions Need, and NASA Ranks High-Performance Onboard Computing Its Third-Highest Civil Space Shortfall" and the display title "Space Computers Fall Short," both resting on the ranking and on the solicitation's statement that space-grade coprocessor/accelerator devices "are not yet powerful enough"; the titles do not say space computers lag commercial ones, because no source makes that comparison outright. (2) The Source line cited 2025 SBIR subtopics S12.07 and S12.04 with radiation-hardened-electronics titles. In the 2025 solicitation S12.04 is X-Ray Mirror Systems Technology, there is no S12.07, and neither title appears in the 2022–2025 solicitations. The line now names the subtopics the write-up actually draws on, copied from the solicitation, and the solicitation PDF's address: Z-ENABLE.02, S13.03 and H15.01. (3) The ESA source was cited as an ESA/EC/EDA Joint Task Force document of 2024; the page is an ESA corporate news item dated 19 September 2008 about the joint effort. The Source line now says so, and the body dates the agreement and the task force to 2008. (4) The problem statement said the environment would destroy commercial electronics within hours and named latchup, the RAD750 and Honeywell RH32/RHPPC, 1990s–2000s desktop performance, a ~$1.8B market, billions in fabrication cost and a 15–20 year lag; none of these are in any source. The paragraph now names the environment components and effects Goddard lists (single-event effects, total ionizing dose, displacement damage), the solicitation's note that commercially available solutions might not naturally provide radiation tolerance, its statement that space-grade coprocessor and accelerator devices are not yet powerful enough, its 100X capability target, the mission demands it and shortfall 1554 name (increased autonomy, processing sensor data in real time, onboard AI), and the shortfall list's foundational design/fabrication/packaging and library needs, and states plainly that the market-size explanation is the write-up's hypothesis. (5) "Orders of magnitude slower than a modern smartphone" is not in the source; the sentence now says the 24-minute delay is part of a communications scenario the solicitation says drives a need for increased autonomy, and contrasts commercial accelerators built for self-driving cars, data centers and smartphones with space-grade devices that are not yet powerful enough. (6) "Will accumulate 2.9 Megarad ... behind 100 mil aluminum" is now "estimated at 2.9 Mrad ... behind 0.1-inch-thick aluminum," the source's wording. (7) The claim that ESA identified 41 critical dependencies, that European missions depend on ITAR-restricted U.S. parts, and that an export change could ground programmes is not in the ESA page; the sentence now reports what the page says, with its own words quoted. (8) The "What's Been Tried" section's guard rings, TMR and ECC costs, RHBP with SOI and hardened oxides, the two U.S. suppliers plus ST, COTS failing against latchup, and the RAD5500 on GlobalFoundries 12nm at 2x are in no source; the section now carries Goddard's description of RHBD and the COTS pivot, the key challenges on COTS system-on-chip components from the shortfall list's lunar-surface communications entry (1558), named as such, and the solicitation's line that commercial accelerator devices can be leveraged but need a redundant system design with health monitoring. (9) The unlock list's named foundries, BAE partnerships, chiplets, GaN/SiC, photonics and a European rad-hard FPGA are in no source; the list now follows shortfall descriptions 1553, 1554 and 1555 and subtopic Z-ENABLE.02, introduced as NASA's framing of the needs rather than as a contrast with economics. The entry-point "chiplet arm" was renamed the architecture arm. (10) In Source Notes, the 1960s discovery date, the two-facility claim, the BAE/GlobalFoundries and BAE/Intel partnerships and the refusal of fabs to take defence work, the $1.8B figure, and the European Component Initiative bullet (50% target, Horizon Europe FPGA) were removed or softened to what the sources say; the supply-chain bullet now dates the ESA/EC/EDA statement to 2008 rather than saying "since 2008." Could not check: nothing; every named source opened. Noticed, not fixed here: the family C3 (unviable economics) framing rests on no source in the write-up, and the Source Notes sentence that smaller transistors are more radiation-sensitive is unsourced and predates this round. Genome tags untouched.