The 2027 cliff, restated. More than 100 GW of announced electrolysis loses any realistic chance of operating by 2030 without an investment decision by end-2027. This week showed the mechanism by which the pipeline survives without committing: the state pays the option premium. Four of the five money items this week were public.

Repsol Cartagena's DIA carries a water number, and it checks out. The favourable-with-conditions DIA of 16 July specifies 332,000 m³/yr from the Mancomunidad de Canales del Taibilla, all wastewater to the refinery treatment plant with no direct environmental discharge, soil monitoring after petroleum hydrocarbons were detected above reference levels, and archaeological supervision near Roman Escombreras. 332,000 m³/yr against 15,000 t/yr is 22.1 L/kg — against the IRENA alkaline benchmark of 22.4 L/kg, for a pressurised alkaline plant. That internal consistency is the strongest argument the figure is real rather than a permitting placeholder.

RFNBO review still autumn. The five-member-state Directors-General ask: additionality transition 2028 → 2035 with grandfathering to 2040; monthly matching retained to 2035 for member states with clean power systems; and — new this week — the Commission to assess lowering the 90% renewable-electricity general exemption threshold. A competing analyst read expects a more modest 2032–2033 delay. Treat 2035 as the industry ask, not the expected outcome.

State aid. The Commission approved a €780 m Dutch renewable hydrogen scheme on 6 August (~400 MW): grants of up to 80% of investment cost plus a variable premium over 5–10 years, bidding concluding Q1 2027. A materially more generous structure than the EHB's fixed premium per kg, and the second Dutch approval in a week.

The 10 GW question, resolved. Following the autumn-2025 monitoring report the economy ministry said it would scrap the target and make it flexible. No formal proposal has been presented and no amending instrument exists; the 2023 NWS Fortschreibung remains operative. Verdict: 10 GW is still the legally standing target on 10 August 2026, with an announced intent to replace it. The apparent source conflict is a conflict between announced and enacted policy, not between facts. This edition stops treating it as unresolved.

Kernnetz. Third market information package, published 4 August, data to 23 July: just under 6 GW of cluster entry/exit capacity reserved on a paid basis, ~0.6 GW cross-cluster contracted and 0.5 GW under review — a doubling against 13 May. Network approved at 9,040 km / €18.9 bn, ~400 km operational, KfW €24 bn backstop. No updated operational-kilometre figure published.

Correction: HYDRIX is not mentioned anywhere on the official Klimaschutzverträge portal. W32's claim that it is the round-2 reference index is downgraded to unverified.

Direction of travel — this part is solid. 11 closures by 31 March 2025, 11 more by 30 June 2025 ("Netzkonsolidierung 700 bar", phase 2), 14 more on 31 December 2025 — Berlin-Rothenbachstraße, Biebelried, Braunschweig, Duisburg, Essen, Hasbergen, Ingolstadt, Kirchheim, Laatzen, Limburg, Lohfelden, Metzingen, Rheda-Wiedenbrück, Wendlingen. H2 Mobility's CEO: the first-generation stations "cannot be adapted to technically changed and growing requirements." The inflection marker: in March 2025, 350-bar sales exceeded 700-bar for the first time, at well over 50% of volume sold. Counter-flow is real but small — Düsseldorf (May 2025, 5 t/day, three pressures, €3.1 m BMDV) and Ludwigshafen (December 2025). Europe-wide, LBST's 2023 census already found 92% of new stations heavy-duty capable.

An arithmetic check worth publishing. Moeve Onuba is stated at 300 MW producing 45,000 t H₂/yr. At 50–55 kWh/kg that implies an 86–94% capacity factor — not achievable on an RFNBO-compliant renewable profile. Either the figure assumes grid power, or it is nameplate arithmetic. This is precisely the gap between announced and delivered output that the industry does not publish, and it applies to most headline t/yr figures in this brief.

No new large-scale supply contract was verifiable between 1 and 10 August. The one strategic move: thyssenkrupp nucera and BHEL announced an alkaline manufacturing collaboration in India on 7 August — not an order, but nucera hedging away from a dead European order book toward India's SIGHT-subsidised market while building non-Chinese capacity that would qualify under auction 4. Read it as a response to the content rule, not to demand.

1 — Four hydrogen applicants are competing for one water system. The Chanza-Piedras Anillo Hídrico in the Tinto-Odiel-Piedras basin carries four hydrogen applicants in a single competencia de proyectos totalling 7,171,099 m³/yr — Moeve 2.9 M, Iberdrola 2.1 M, Rolwind 1.2 M, Galena 0.97 M. The competing-demand story with hard numbers attached, in a water-stressed basin, from one document. It is also the clearest argument that project-by-project water screening is not an environmental nicety but a capacity-allocation question.

2 — La Joya has designed for dry or hybrid cooling; Lingen has not. Cooling is 18% of La Joya's draw against 75% of Lingen's. Cooling is the swing variable, and this is the first empirical evidence in the dataset that a Mediterranean project has designed around it. La Joya's all-in intensity of 16.4 L/kg sits below the PEM benchmark and well below the ~30 L/kg Mediterranean rule of thumb — because it is not evaporating its water. Inferred from the ratio, not stated by the promoter.

3 — Permit headroom at Lingen is roughly twentyfold. The NLWKN permit authorises up to 23,979,000 m³/yr of abstraction against the 1,200,000 m³/yr the electrolysis actually uses. The permit covers the hydrogen-ready gas turbine plant as well, so it is not apples to apples — but the water right is plainly not the constraint at Lingen, which raises the question of what headroom the undisclosed German projects hold.

RD 1085/2024 confirmed in live use for hydrogen feedstock. BOJA 102/2026 cites article 7.6 explicitly to authorise Moeve's production and supply of 590,000 m³/yr of regenerated water at Palos — the first observed instance of the new reuse regulation applied to a hydrogen project. The Andalusian application form is set by the Resolución of 10 April 2026 (BOJA 73/2026).

EU Water Resilience Strategy, Council document ST-10288-2026-REV-1 of 15 June: 10% EU water-consumption reduction by 2030; joint methodology for water-efficiency targets due Q4 2026; abstraction reduction potential 45–95% for electricity, 30–50% for manufacturing. No hydrogen-specific provision. The dry-cooling public-private partnership is not evidenced in this document — downgraded from last week and carried as unconfirmed.

Still true, and still the point: no hydrogen-specific water permitting rule at EU level; no water criterion in the RFNBO delegated acts; no EHB auction round scores water; no hydrogen-specific water-footprint certification beyond generic ISO 14046. No aggregated project-level dataset exists — the IEA Hydrogen Production and Infrastructure Projects Database, the IEA Hydrogen Tracker and the Hydrogen Council Compass all lack a water field. Nature Communications Sustainability 2026 finds 43% of planned electrolysis capacity sits where freshwater evaporative cooling would be unsustainable — and explicitly concedes it did not survey project-level water plans. This section is built bottom-up because there is nothing to build it from.