Ask a developer when the RFNBO rules get hard and you will usually be told 2030, when hourly matching arrives. That answer is wrong in a way that matters, and the reason is one paragraph of the delegated act that gets far less attention than it deserves.
Read Article 11 before you read Article 6
The additionality tests — the requirement that the renewable generation be new, and unsupported — are in Article 5. Article 11 then does something unusual:[1]
Article 5, points (a) and (b) shall not apply until 1 January 2038 to installations producing renewable liquid and gaseous transport fuel of non-biological origin that come into operation before 1 January 2028. This exemption shall not apply to capacity added after 1 January 2028.
Read the two dates together. A plant that reaches operation on 31 December 2027 buys itself ten years of relief from the additionality tests. A plant that reaches operation a day later gets none. The value of hitting that date is not marginal, and it is not a compliance detail: it is a decade of being allowed to source electricity that a later project cannot.
This is why "the deadline is 2030" is the wrong framing. 2030 is when a rule tightens for everybody. 2028 is when a door closes, and it is the only date on the whole timeline that a project currently in construction can still act on.
Figure 1
What the delegated act actually says, as it stands today
The rules in force on 30 August 2026. No amending act has been adopted, or formally proposed.
Article 11. This is the only date on this chart a project under construction can still act on.
Article 11: Article 5(a) and (b) do not apply until 1 January 2038 to installations in operation before 1 January 2028.
Article 6, first paragraph.
Article 6, second paragraph.
Article 6: following a notification to the Commission, Member States may apply the hourly rules from 1 July 2027 for fuel produced in their territory.
Article 11 exempts only points (a) and (b) of Article 5 — the 36-month rule and the prohibition on support — not the whole of Article 5. And the exemption does not extend to capacity added after 1 January 2028, so a plant that qualifies cannot expand under its own grandfathering.
Source: Commission Delegated Regulation (EU) 2023/1184, Articles 5, 6 and 11 · Data as of 2026-08-30
Show the data ▸Hide the data ▾
| Rule | In force | Article |
|---|---|---|
| Deadline to qualify for additionality relief | 1 January 2028 | Art. 11 |
| Additionality relief, if qualified | until 1 January 2038 | Art. 11 |
| Monthly temporal correlation | until 31 December 2029 | Art. 6 |
| Hourly temporal correlation, EU-wide | from 1 January 2030 | Art. 6 |
| Hourly correlation, national early adoption | from 1 July 2027, on notification | Art. 6 |
Two details in that chart are worth stating in words, because both are routinely missed.
The exemption covers only points (a) and (b) of Article 5, not the whole article.[1] And it does not extend to capacity added after 1 January 2028 — so a qualifying plant cannot expand under its own grandfathering. Phase two of a two-phase project is a different regulatory animal from phase one, even on the same site.
Hourly matching may not wait for 2030
The second thing the headline framing gets wrong is that hourly correlation is a fixed 2030 event. Article 6 sets 2030 as the EU-wide default, and then adds:[1]
Following a notification to the Commission, Member States may apply the rules set out in this paragraph from 1 July 2027 for renewable liquid and gaseous transport fuel of non-biological origin produced in their territory.
So 2030 is a floor, not a ceiling. A project sited in a Member State that chooses to move early faces hourly matching from mid-2027 — inside the same window as the additionality door. Whether any Member State does is a political question we are not going to forecast. That it is available, on notification, without any amendment to the act, is a fact worth having in the model as a sensitivity.
The escape valve nobody models
Now the clause that cuts the other way, and that we have not seen quantified anywhere. The third paragraph of Article 6:[1]
The temporal correlation condition shall always be considered complied with if the [fuel] is produced during a one-hour period where the clearing price of electricity resulting from single day-ahead market coupling in the bidding zone is lower or equal to EUR 20 per MWh or lower than 0,36 times the price of an allowance to emit 1 tonne of carbon dioxide equivalent.
Note the word always. This is not part of the transitional regime — it does not expire in 2030, and it survives whatever the revision does to the correlation rules unless the revision touches this paragraph specifically.
The practical consequence is that the burden of hourly matching is not uniform across Europe. It falls hardest where prices rarely go low, and it is materially softer in a bidding zone with deep renewable penetration and frequent depressed hours — which is to say, in exactly the zones where an electrolyser wants to be anyway. Any hourly-matching cost estimate that ignores this clause is measuring a stricter rule than the one that exists.
The revision is a commitment, not a rule
The reason the 2028 door gets ignored is that most of the market expects the rules to be loosened before it closes. That expectation is not fantasy — it is a reading of the Commission's own words. In AccelerateEU on 22 April 2026 the Commission committed to a targeted review of the production criteria, ahead of the review that was scheduled for 2028.[10]
But the distance between a commitment and a rule is the whole subject of this piece. The review was promised for the second quarter of 2026. As of the beginning of July it was neither adopted nor formally proposed, with a first draft expected in the third quarter.[11] Eight days into July the industry was still writing joint letters asking for it — asking, among other things, that additionality be extended to at least 2035 and monthly correlation kept until at least 2035.[12] You do not write that letter about a rule that has already changed.
The most-quoted forecast of what the revision will contain expects additionality delayed to 2032–2033, grandfathering extended to at least 2040, hourly correlation pushed to 2032–2033, and a price impact of roughly €2/kg.[13] That is a policy advisory firm's political forecast, and it is worth reading. It is also worth noting that the same note expected publication by June 2026.
The evidence that complicates all of this
If RFNBO compliance were the ruinous burden it is often described as, the European Hydrogen Bank's third auction should have struggled. It did the opposite.
Figure 2
The awkward evidence: the third auction cleared at under €1.50/kg in the general baskets
Bids awarded in the European Hydrogen Bank's third auction, by topic. A fixed premium per kilogram, not a production cost.
Lowest: Cloudberry, Finland, €0.44. Highest: Lotse, Germany, €1.10.
Lowest: AN-1-B, Greece, €0.57. Highest: two Spanish projects at €1.48.
Both Norwegian: Gen2-LH2 and RogalandH2. Eight times the cheapest general bid, for the same molecule under a different end use.
These are subsidy bids, not LCOH. A developer bids the premium it needs on top of everything else it earns; reading them as production costs is the single most common misuse of this dataset. The ceiling was €4.00/kg and applied across all three topics, not only to the maritime and aviation basket.
Source: European Commission / CINEA — IF25 hydrogen auction results · Data as of 2026-05-07
Show the data ▸Hide the data ▾
| Topic | Awarded range (€/kg) | Budget |
|---|---|---|
| Topic 2 — RFNBO and low-carbon electrolytic | 0.44–1.10 | €400m |
| Topic 1 — RFNBO general | 0.57–1.48 | €600m |
| Topic 3 — maritime and aviation | 3.48–3.49 | €300m |
| Ceiling (all topics) | 4.00 | — |
Fifty-eight bids from eleven countries, requesting €8.4 billion against a €1.3 billion budget — oversubscribed more than six times.[5] Nine projects awarded €1.09 billion for around 1.1 GW, and the general baskets cleared between €0.44 and €1.48 per kilogram.[4]
Those numbers deserve to be read carefully rather than used. They are fixed-premium subsidy bids, paid for up to ten years against certified and verified production, ranked lowest to highest until the budget is exhausted.[9] They are not production costs, and they are not LCOH. What a bid represents is the premium a developer says it needs on top of everything else it earns — and how a developer arrives at that number is its own commercial judgement, not something the Commission defines.
Two things the auction does tell us. First, at least nine projects believe they can produce compliant hydrogen with a subsidy of well under €1.50/kg, under the rules as they stand. Second, the same molecule under a different end use needs eight times the support: the maritime and aviation basket cleared at €3.48–3.49/kg against a €4.00 ceiling.[4] The gap between those baskets is a demand-side fact, not a production-cost one.
And the evidence that complicates the auction
Figure 3
And the awkward evidence cuts both ways: an award is not a project
Second auction: what was awarded in December 2024, against what had signed a grant agreement by January 2026.
2.3 GW of electrolysis.
381.25 MWe. Twenty-seven per cent of the money, seventeen per cent of the capacity.
This was not a straight fall from 15 projects to 6. Seven of the fifteen withdrew, taking 1.9 GW with them; the Commission then invited ten reserve-list projects totalling 774 MW; six signed. Capacity fell from 2.3 GW awarded to 381.25 MWe signed. The Commission's own explanation is that sponsors were forced to reassess project maturity or financial viability between bidding and having to post the completion guarantee. Every withdrawal came from the general basket; the three maritime projects proceeded.
Source: European Commission, DG CLIMA — grant agreement signature announcement, and S&P Global on the withdrawals · Data as of 2026-01-20
Show the data ▸Hide the data ▾
| Stage | Amount (€m) | Capacity |
|---|---|---|
| Awarded, December 2024 | 992 | 2.3 GW, 15 projects |
| Withdrawn | — | 1.9 GW, 7 projects |
| Reserve list invited | — | 774 MW, 10 projects |
| Grant agreements signed, January 2026 | 270.6 | 381.25 MWe, 6 projects |
The second auction awarded €992 million to fifteen projects covering 2.3 GW. By January 2026, six projects had signed grant agreements for €270.6 million and 381.25 MWe.[6]
The route between those two numbers matters, and quoting "six of fifteen" without it would be unfair. Seven of the fifteen withdrew, taking 1.9 GW with them, and the Commission then invited ten reserve-list projects totalling 774 MW.[7] The six signatures come from that enlarged pool.
The Commission's explanation is the useful part: sponsors were forced to reassess project maturity or financial viability between bidding and having to provide the completion guarantee.[6] A winning bid costs nothing. A completion guarantee costs money, and the draft terms for the fourth auction put it at 8% of the maximum grant, backed by a minimum rating of BBB– or equivalent, with financial close required within two and a half years of signature and operation within five.[8]
That is the filter. It is not the RFNBO criteria that removed 1.9 GW from the second auction — it is the moment the sponsor had to post real money against a real date.
What to do with the dates
- Find out whether your project can be in operation before 1 January 2028. If it can, the additionality relief to 2038 is worth quantifying, and it may be worth accelerating for. If it cannot, stop modelling as though it might.
- Do not assume grandfathering covers phase two. Capacity added after 1 January 2028 is outside the exemption, on the same site, under the same permit.[1]
- Check your Member State's position on early hourly adoption. The option exists from 1 July 2027 and needs no amendment to be exercised.[1]
- Model the low-price exemption explicitly. It is permanent, it is bidding-zone specific, and leaving it out overstates the cost of hourly matching by an amount that depends entirely on where you are.
- Put the revision in a sensitivity with a name and a date, not in the base case. It has been committed to, promised for one quarter, and not delivered in that quarter. Both of those facts belong in the note.
- Price the completion guarantee as a real pre-close cash requirement, not a formality. It is what removed most of the second auction's capacity between award and signature.
None of this is a prediction about what the Commission will do. It is the observation that a project reaching FID this autumn will be built under the rules that exist, and that the most valuable of those rules expires as an opportunity in sixteen months.