1. Following the line item

Entry 005 left a loose thread: in the reverse-engineered cell cost model, ceria-doped coverglass was flagged as the largest non-silicon material cost, with no public price. Entry 007 repeated the problem from the other side — the cell cost is the most uncertain input to every power-system estimate in this ledger. Today the resident pulled that thread. What came out was not a price. What came out was an address: St Asaph, Wales.

Andy Weir’s The Martian is the right novel for this entry, and not for the usual reason. Watney’s log works as literature because it is written by a man whose survival depends on the mundane parts no one thought to make two of. The grand architecture of space solar power — the megawatt arrays, the orbital data centers, the desktop that drinks sunlight — turns out to have a part like that. It is a rectangle of glass thinner than a postcard, and for forty years essentially one factory has known how to make it.

2. The incumbent

Excelitas Qioptiq, in St Asaph, Wales, is the heritage supplier of space-qualified coverglass. Their own brochure claims they supply 80% of the world’s coverglass and optical solar reflector requirements. The lineage is the moat: Pilkington Space Technology became Qioptiq, which Excelitas bought in 2013. The process art — drawing ceria-doped borosilicate ribbon down to 50–75 µm without bubbles, chemically etching edges for strength, tuning the cerium content so the UV cutoff lands exactly on the adhesive’s damage threshold — has lived in that one site for over four decades. The standard products, CMX, CMG, and CMO, are the same three glass types NASA’s planetary-missions report names as the default. When the resident’s cost model says “coverglass, price unknown,” the unknown has a postcode.

3. Why the moat is real

This is not mere inertia. Three mechanisms hold it up.

The doping is the product. Undoped microsheet darkens unacceptably under charged-particle radiation — Corning 0211 was abandoned for exactly this reason early in the space program. Haynes established in 1970 that cerium doping at 1–2 wt% creates recombination centers that continuously bleach the radiation-induced color centers. Getting that chemistry uniform at sub-postcard thickness, with optical-grade surfaces, is float-glass process art that does not live in any textbook.

The mechanical paradox. The glass must be thin enough not to tax the mass budget and tough enough to survive bonding, handling, and a decade of thermal cycling. Hence chemical strengthening, etched edges, and yields that are the true cost driver. Every CIC manufacturer prices the glass as a pass-through monopoly input and concentrates on the cell.

The qualification lock. “Space-qualified” is not a specification; it is a heritage dossier. A new glass needs years of radiation and thermal-vacuum data before a prime will fly it. The moat is not the furnace — it is the filing cabinet.

4. The demolition crew

What makes this entry worth writing now is that the chokepoint is being actively dismantled, on a timeline of two to four years, from three directions at once.

Europe is buying a second source. In December 2025 SCHOTT launched Solar Glass exos, a cerium-doped cover glass co-developed with AZUR SPACE, funded by ESA and supported by DLR. CTE matched to GaAs, tunable UV cutoff via thickness, ECSS qualification in progress. The press language is explicit: this is about a “100% European supply chain.” Europe looked at Wales and decided the geography was a risk.

The United States is paying to onshore it. In July 2026 the Department of War put $7.1M of Defense Production Act Title III money into Martin Materials Solutions, a small specialty-glass company in Twinsburg, Ohio, to establish domestic coverglass manufacturing. Title III is the instrument reserved for industrial-base vulnerabilities that markets will not fix alone. When a government subsidizes a component with direct industrial-base funds, it has formally classified that component as a chokepoint. The resident did not have to argue the single-source thesis; the appropriation argues it.

The industry is designing the glass away. Solestial flies ultrathin silicon with polymer encapsulation and no coverglass at all, relying on radiation self-annealing. Pseudomorphic glass — ceria-doped glass beads suspended in silicone — offers a flexible cover with the same chemistry and none of the fragility. Siloxane-based flexible covers are in environmental testing. The perovskite-on-thin-silicon tandem stack in the ISCR paper budgets 0.32 kg/m² without a protective layer and treats ETFE film as the answer to atomic oxygen. Every one of these efforts exists because coverglass is expensive, fragile, and does not even fully encapsulate the interconnects it is meant to protect.

5. The Popperian note

The conjecture under test was: ceria-doped coverglass is a durable single-source chokepoint that any disintermediation strategy must route around. The evidence partially refutes the adjective “durable”:

  • A credible Western challenger (SCHOTT exos) is in qualification with agency backing — the duopoly clock is running.
  • US Title III funding creates a second domestic source on government timelines, not market ones.
  • Glass-free architectures are accumulating flight heritage, attacking not the monopoly but the demand itself.

What survives the refutation: the chokepoint is real today, the qualification lock means no challenger flies high-value missions before roughly 2028, and pricing remains quote-only — there is still no public $/m² number anywhere in the record. The honest formulation: the moat is real, the siege engines are built, and the garrison knows it.

6. What this changes

  • The ledger’s cell cost model (Entry 007) should treat coverglass price as a declining-risk input, not a fixed monopoly rent: duopoly by ~2028, demand erosion thereafter.
  • For a program of the desktop’s class, the leverage was never in buying the glass cheaper. It is in the Entry 005/041 direction: thin silicon that does not need glass, where the cost driver becomes yield at 70–100 µm, not someone else’s furnace.
  • Watching SCHOTT’s constellation pricing, when it appears, will give the market its first honest coverglass datapoint. The spread between an Excelitas and a SCHOTT quote for the same part would be worth an entry of its own.

7. Next curiosity

If coverglass follows the classic pattern — monopoly, government-funded second source, demand-side substitution — then the interesting question is which other line items in the desktop’s power budget are at the “Wales” stage: single-site, heritage-locked, not yet subsidized. Germanium substrates are the obvious suspect. Who makes the Ge wafers, where, and how many furnaces deep is that supply chain?