Skip to content
Substrata
Join

Science

The pipeline: new papers, grants and who is doing them →

A map of constraints is half the picture. This is the other half: what could remove each one, how it would work, and how close it is to being usable at scale.

Technologies tracked
How this is computedTechnologies tracked = science entries; bottlenecks addressed = distinct bottlenecks named as relieved by any entry; reaching production = entries judged 8 or 9; sourced = entries whose readiness cites a source.Method →Code →
across 4 fronts
Bottlenecks addressed
How this is computedTechnologies tracked = science entries; bottlenecks addressed = distinct bottlenecks named as relieved by any entry; reaching production = entries judged 8 or 9; sourced = entries whose readiness cites a source.Method →Code →
at least one candidate relief each
Reaching production
How this is computedA judged level on a nine-step scale from "idea with a physical basis" (1) to "in production at scale" (9).Method →Code →
judged 8 or 9 of 9
Sourced
How this is computedTechnologies tracked = science entries; bottlenecks addressed = distinct bottlenecks named as relieved by any entry; reaching production = entries judged 8 or 9; sourced = entries whose readiness cites a source.Method →Code →
of 14, each citing the claim behind its score

Read this first. The readiness numbers are judgements written by hand, each now citing a source that bears on it — a company announcement, a filing, an agency report or a paper. Reading those sources moved three of the twelve scores, one of them because the entry turned out to describe the wrong half of the problem. The reasoning sits next to every number so you can disagree with it.

01AI & compute

Training and running models: the chips, memory and packaging they need.
  • High-numerical-aperture EUVReaching productionEstimate, not a measurementPast development: ASML says Intel Foundry entered high-volume manufacturing on High-NA-patterned layers in 2026, at yields matching the previous generation. This row said "not yet volume production" until the source was read.By Substrata, open to challenge · 2026-09-15Supporting source →

    The next generation of the machine that prints chip patterns, with larger optics that resolve finer features in one pass.

    Would relieveLeading-edge foundry capacity

  • Dry and metal-oxide photoresistsBeing provenEstimate, not a measurementDemonstrated on production-class tools and being qualified; qualification is per process and per factory, which is the slow part.By Substrata, open to challenge · 2026-09-15Supporting source →

    A different chemistry for the light-sensitive layer on a wafer, deposited as a film rather than spun on as a liquid.

    Would relievePhotoresist formulation

  • Hybrid bondingBeing provenEstimate, not a measurementIn production for some image sensors and logic stacks; ramping for memory stacks.By Substrata, open to challenge · 2026-09-15Supporting source →

    Joining two chips copper-to-copper with no bumps between them, so they behave more like one piece of silicon.

    Would relieveHigh-bandwidth memory stacking yieldAdvanced packaging capacity

  • Diamond heat spreaders in production packagingBeing provenEstimate, not a measurementUsed in specialist radio-frequency and laser parts; not standard in high-volume logic packaging.By Substrata, open to challenge · 2026-09-15Supporting source →

    Putting a layer of laboratory-grown diamond next to the hottest part of a chip, because nothing else moves heat as fast.

    Would relieveCVD synthetic diamond heat spreaderTwo-phase dielectric immersion coolant

02Robotics & autonomy

Machines that act in the world: motors, drives, sensors and the magnets inside them.
  • Rare-earth-lean and rare-earth-free motorsReaching productionEstimate, not a measurementGrain-boundary diffusion is standard practice and rare-earth-free traction motors are in production; the trade-off is mass and efficiency, not feasibility.By Substrata, open to challenge · 2026-09-15Supporting source →

    Motor designs that use less heavy rare earth, or none: ferrite magnets, wound rotors, and diffusion methods that place dysprosium only where it is needed.

    Would relieveDysprosium metalDidymium (Nd-Pr) metal, magnet feed

03Energy

Generating power and getting it to where the work happens.
  • Amorphous metal transformer coresReaching productionEstimate, not a measurementIn production at scale, but only for part of the problem. ABB has sold amorphous metal distribution transformers for two decades; the constraint is applicability rather than maturity, because the lower saturation induction means a larger core cross-section, larger coils and a larger footprint — acceptable on a pole, not on a grid-scale unit.By Substrata, open to challenge · 2026-09-18Supporting source →

    Transformer cores wound from rapidly cooled metal ribbon that has no crystal structure, instead of the grain-oriented silicon steel the industry is short of.

    Would relieveGrain-oriented electrical steel (GOES)

  • High-temperature superconducting magnetsBeing provenEstimate, not a measurementFull-scale magnets have passed acceptance testing and are in serial production, validated by the US Department of Energy. No power plant has run on them. The constraint has moved from physics to tape supply.By Substrata, open to challenge · 2026-09-15Supporting source →

    Magnets wound from ceramic tape that carries current without resistance, strong enough to hold a fusion plasma in a machine small enough to build quickly.

    Would relieveREBCO superconducting tape, 12 mm

  • Small modular reactorsBeing provenEstimate, not a measurementDesigns are licensed or in licensing and first units are under construction; none is yet delivering power at scale in a Western market.By Substrata, open to challenge · 2026-09-15Supporting source →

    Nuclear reactors small enough to build in a factory and truck to site, rather than constructing each one in place.

    Would relieveHeavy-duty gas turbine order booksGrid interconnection queues

  • Gallium nitride power conversionReaching productionEstimate, not a measurementIn volume production for consumer and datacentre power; higher-voltage grid use is earlier.By Substrata, open to challenge · 2026-09-15Supporting source →

    Power electronics built on gallium nitride rather than silicon, which waste less energy as heat when converting voltage.

    Would relieveLarge power transformer slots

  • 200 mm silicon carbide wafersReaching productionEstimate, not a measurementIn production at several makers and expanding; the market is currently oversupplied rather than short.By Substrata, open to challenge · 2026-09-15Supporting source →

    Making silicon carbide crystals in bigger discs, so each one yields far more usable chips.

    Would relieveSilicon carbide substrate, 200 mm semi-insulating

  • Superconducting transmission cableBeing provenEstimate, not a measurementCity-scale links have run for years as demonstrations; no routine procurement path exists.By Substrata, open to challenge · 2026-09-15Supporting source →

    Power cables cooled so they lose nothing in transit, carrying far more current through a much smaller duct.

    Would relieveHigh-voltage cable and switchgear

04Advanced manufacturing

Making physical things faster: fabs, additive manufacturing, precision assembly.
  • Nanoimprint lithographyBeing provenEstimate, not a measurementCanon commercialised the FPA-1200NZ2C in October 2023 and shipped one to the Texas Institute for Electronics in 2024, at 14 nm minimum linewidth — equivalent to a 5 nm node. That is a real machine doing real patterning, but not volume logic production: throughput, defectivity and the supply of 1:1 imprint masks are the open questions, and a stamp touching the wafer fails differently from light.By Substrata, open to challenge · 2026-09-18Supporting source →

    Printing chip patterns by pressing a patterned mask into resist like a stamp, instead of projecting an image through lenses and mirrors.

    Would relieveEUV lithography scannersEUV projection optics

  • On-site neon recycling and separationBeing provenEstimate, not a measurementScored 8 here until the source was read. Two years after the shortage, a leading memory maker was still calling its own system an industry first, recovering 72.7% with 77% as a target — which is a working pilot at one firm, not an industry that has adopted it.By Substrata, open to challenge · 2026-09-15Supporting source →

    Capturing and re-purifying the rare gas a chip factory uses, instead of buying it fresh from an air separation plant.

    Would relieveNeon, excimer laser grade

  • A fully synthetic crucibleIn the labEstimate, not a measurementThis entry originally claimed the synthetic inner layer was the open question. The source says the opposite: fusing synthetic powder onto the inside of a natural-quartz crucible is established practice. So the relief is narrower and earlier than scored — replacing the natural BODY has no demonstrated production route.By Substrata, open to challenge · 2026-09-15Supporting source →

    Making the whole crystal-growing crucible from manufactured quartz, rather than only its inner lining.

    Would relieveCrucible-grade high-purity quartz sand

05The readiness scale

  1. 1Idea with a physical basis
  2. 2Concept worked out on paper
  3. 3Demonstrated in a laboratory
  4. 4Works as a component in a lab
  5. 5Works in a realistic setting
  6. 6Prototype at meaningful scale
  7. 7Pilot running in the real environment
  8. 8Qualified and entering production
  9. 9In production at scale
What these columns mean
In the lab
Levels 1 to 4: demonstrated in a laboratory or earlier. Real physics, no product.
Being proven
Levels 5 to 7: works outside the laboratory, not yet something a buyer can order at scale.
Reaching production
Levels 8 and 9: qualified and shipping, or already standard practice.
AI & compute and the other fronts
The front is which direction of progress this would unblock, not who sells it.