Track A · Investor
Thesis, asks, money, programme, reserve, risk
Hydrogen from aluminium and water is not an invention. The invention, if there is one, is a civil process for living with the oxide that otherwise stops the reaction — and for reading that oxide fast enough to govern it. The economic hinge is not the hydrogen. It is a waste-aluminium credit, a residue someone will buy as ATH or boehmite, gallium that recirculates rather than dissolves, and a host who already has a pad, power, water and a safety case.
That is a plant-siting thesis more than a science thesis. It is why the Founding Member programme is built around hosts and fields, not around a round.
A3 Economics envelope
Indicative ranges. Every row posture-tagged. No cell is a quote. The band moves across zero — which is why ‘negative-cost hydrogen’ was withdrawn as a sentence and kept as a conditional cell.
Planning anchors
Not a valuation. Not an offer.
| Figure | Value | Posture |
|---|---|---|
| Theoretical hydrogen yield | ~0.11 kg H₂ / kg Al (STP, ~1.24 L/g) | Established — by others |
| Reaction enthalpy | ≈ −418 kJ / mol Al, path-dependent | Established — by others |
| Planning LCOH, no residue credit | A$4–12 / kg H₂ | Modelled |
| Planning LCOH, with ATH / boehmite credit | A$−2 to A$4 / kg H₂ | Modelled |
| Hydrogen Production Tax Incentive screen | A$2 / kg H₂ (eligibility untested) | Established — by others |
| Founding Member programme | A$10M / 20 seats; Foundation Tranche A$2M | Vendor-claimed |
| Measurement package proposed now | A$850k–1.4M sponsored research | Modelled |
| Corridor, civil hydrogen-and-residue family | Floor A$0 · target A$8–25M · stretch A$40M | Vendor-claimed |
| Withdrawn as valuation | £95M IM · US$1.14B Monte Carlo median · TRaCE A$6.8M | Withdrawn |
A4 Founding Member programme
Twenty seats. Aggregate A$10 million. Equal seats at A$500,000 unless the separate instrument says otherwise. A Foundation Tranche of four seats, A$2 million, is called first: it pays WP0 and the first half of WP1. The remaining sixteen seats are not called if WP1 fails, and are not called if counsel has not issued the chain-of-title statement.
Who the programme is for
A smelter, a recycler, a miner, a port operator, an industrial-gas company, a residue buyer, a family office that already understands process plants, a university that wants to host WP4.
Who it is not for
Anyone who needs a valuation to decide, anyone who cannot sit inside the legal gate, anyone who wants compute, sensing, medical or defence in the same conversation.
A5 Work packages
Every line Modelled; none quoted. TRaCE is not counted. The A$850k–1.4M range in the ask allows for quotation variance; the table is the internal build-up.
Gated measurement programme
A$ · internal build-up
| WP | Scope | A$ | Gate |
|---|---|---|---|
| WP0 | Reactor and film control — reproducible 316L cell; logged temperature, ultrasound power density, field waveform, water chemistry; baseline EIS band. | 280,000 | Baseline repeatable cell to cell, or stop and fix the build. |
| WP1 | Hydrogen yield and residue phase — yield per kg Al; gas composition; residue XRD; mass and energy balance. | 360,000 | If yield is not distinguishable from an un-managed Al–water control, the triple-mechanism claim stops. |
| WP2 | Impedance as process instrument — EIS during reaction; feature extraction; correlation with yield and phase. | 210,000 | If the spectrum does not move with the reaction repeatably, the control-layer filings are not enabled by this film. |
| WP4 | Controls and replication — blank cell; Ga-free control; ultrasound-off; field-off; second laboratory; D₂O isotope check. | 220,000 | A pass that also appears in the blank is an artefact. |
| WP5 | Conversion triage — decide which civil filings to convert in April 2027 on the evidence. | 250,000 | A fail saves conversion and national-phase cost before it is incurred. |
| Now | WP0 + WP1 + WP2 + WP4 as sponsored research. | 1,070,000 | Released after counsel is engaged. WP5 is conversion, not research. |
A6 Reserve case — floor is zero
If WP1 fails, the civil filings due in April 2027 are allowed to lapse, the Founding Member programme is not called beyond the Foundation Tranche already spent on the test, and the estate is unharmed because this family was never carrying it. The sunk cost is a measurement. That is the design. Stretch A$40 million is a later-stage planning figure, conditional on a working skid, a residue offtake and at least one field licence. It is not this meeting.
A7 Risk register
Named risks and their tests
| ID | Risk | Class | Test | If it lands |
|---|---|---|---|---|
| R1 | The triple mechanism does not beat an unmanaged control | Fatal to the filed embodiment | WP1, ultrasound-off / field-off / Ga-free | Process claim collapses to prior-art Al–water. Filings lapse at a saving. |
| R2 | Gallium is consumed, not recirculated | Economic | Mass balance on Ga over 100 hours | LCOH envelope moves to the high end or off the page. |
| R3 | Residue is mixed hydroxide, not saleable ATH / boehmite | Economic | XRD + impurity register, WP1 | Second product vanishes; hydrogen must carry the plant. |
| R4 | Feedstock coatings, oils and alloys poison the film | Operational | Decoat / degrease gate; dirty-scrap campaign | First hosts restricted to clean extrusion scrap. |
| R5 | Hydrogen handling at a host site fails the safety case | Fatal to a host class | HAZOP, SIL, dust (Al), H₂ (ATEX / AS 1375) | That host class closes. Others remain. |
| R6 | Civil claim set is dual-use / DSGL | Audience | Counsel classification | Circulation restricted; physics unchanged. |
| R7 | Woodall / Ga-Al prior art anticipates the independent claims | IP | Clearance-grade search | Claims narrow to the control loop and the managed bilayer. ‘First’ never used. |
| R8 | Counsel remains unappointed past 23 September 2026 | Corporate | Engagement letter | Same blocker as the rest of the estate. This paper stays inside the group. |
| R9 | A Founding Member treats a seat as a security | Legal | Instrument drafting | Programme stops. Seats are not equity, notes, options or units. |
| R10 | Compute and sensing language leaks into a hydrogen conversation | Ledger | QC halt | Document pulled. The two-ledger rule is not optional. |
A8 What changed, and what was withdrawn
Every row is a correction the group made against itself before a counterparty could. That discipline is the most defensible thing on this ledger.
Correction register
| Earlier statement | Current statement | When |
|---|---|---|
| TRaCE counted as A$6.8M executed, with M6 as the physical-validation vehicle | Not counted. The University of Newcastle arrangement was withdrawn and must not be re-counted. | Standing rule, 8 September 2026 |
| April 2026 IM language: £95M pre-money, £55–152M IP valuations, ‘computational validation complete’ | Corridor A$0 / 8–25 / 40M for this family, conditional. Third-party model output is never a valuation input. | This paper, 10 September 2026 |
| Monte Carlo (March 2026): median US$1.14B, 0% probability of loss, 20.8× on a US$5M cheque | Modelled evidence, not a valuation. Probability-of-loss of zero is a model artefact. Withdrawn as an external figure. | This paper |
| GALVANOXIDE as a universal platform: protonic computing, oxide neural nets, 377 applications | Sixty civil industrial applications, listed. Excluded set named. Compute, sensing, medical and defence are other ledgers. | This paper |
| ‘Negative-cost hydrogen’ and ‘carbon-negative’ as unqualified claims | A modelled cash cost that can print negative only if a residue credit and a scrap price both clear named gates. | This paper |
| Kitchen-table foil tests treated as validation | A protocol. Not a result. No run is on the board record. | This paper |
| Chain of title stated two ways between VivoVac and HydroGien | Held within the VivoVac / HydroGien group, pending one counsel-issued chain-of-title statement. | Unresolved — named blocker |