Petroleum products
from captured carbon.

Incumbent cost stack versus FFA production cost

What a location spends now, and what FFA would cost instead. The amber block is distribution and marketing \u2014 the layer FFA removes by producing on site.

National fuel spend and the distribution component

Total national spend on road and aviation fuel, and the share of it that is distribution rather than product.

Logistics cost avoided by siting at the demand point

FFA's units are shipping containers. Siting them where the fuel is used does not reduce the haul \u2014 it designs it out.

Capex and units to supply ten percent of national demand

The smallest achievable programmes. Iceland's aviation sector is the only ten-percent target under $100M in the database.

Real PPA prices versus the database retail proxy

The database carries retail tariffs. A plant operator signs a power-purchase agreement \u2014 far cheaper than retail.

The honest basis for comparison

A pump price contains crude, refining, distribution, tax and retail margin. FFA's figure is a factory-gate production cost. Presenting the difference as a national saving would be a false comparison, and this analysis does not make that claim.

Two comparisons are defensible. First, whether FFA can produce below the price at which fuel currently sells — that determines whether there is a business. Second, the distribution and marketing layer, a substantial part of the pump price. FFA removes that layer structurally, because the fuel is made where it is used rather than shipped to it.

What the analysis supports, and what it does not

Iceland's aviation sector is the clearest entry point in the dataset: around fourteen base units — or one plant scaled to the same output — for $55–65M of capital and 26 MW of power, in a country whose electricity is cheap, whose fuel is not subsidised, and where all three products are already viable. It is the only ten-percent target anywhere in the database costing under $100M.

The counterweight is scale. Even FFA's largest standard unit supplies under a hundredth of a percent of a mid-sized country's road fuel. India at ten percent would need nearly thirty thousand base units and $135bn. FFA cannot address large markets at a ten-percent share, and the plan does not claim it can. The strategy is small markets, or well-chosen niches inside large ones.

A note on the electricity price

The 394-location database carries retail tariffs — the business column is labelled a proxy in FFA's own workbook. A plant operator signs a power-purchase agreement instead. Real corporate PPAs are far below retail, and FFA's own workbook already records the Cambodia National Solar Park. The figures throughout this section use a representative PPA; the basis is available on request. On the database's retail proxies the same analysis would understate FFA's position several-fold.

Labour — two operators at the local rate

Two operators per location regardless of output, at the local average wage plus 25% employer on-cost. Sources: OECD Taxing Wages 2026 (Europe), ILO and national statistics elsewhere. A 64× spread — $4,500/yr in DR Congo to $290,215 in Switzerland. This replaces the local rate: 2 operators x local wage x 1.25 placeholder in FFA's V42.1 model, which was 2.4× too high for Switzerland and 50× too high for Myanmar.

Capital cost — the scaling curve

FFA's three quoted plant sizes, fitted to a smooth scaling curve that follows the pattern seen in real process plants. 2015 pricing, stable to 2026: the technology gain has come through efficiency, not cost.

The pilot proved the process; the design is scalable. This analysis assumes output rises linearly with plant size, while production cost falls with scale to a cap. The cap is reached at a few times the base unit and holds thereafter — which agrees with FFA's own note that there are minimal economies of scale beyond 3 t/day, once the electrolyser is at its limit.

For reference, FFA's measured data shows a far steeper curve at small sizes: $4.50/gal at 33 gal/day, $2.40 at 333, $1.90 at 1,000 — reductions of 47% then 21%. The 8% rule applies beyond the 3 t/day base, where the step changes are smaller.

FFA's base units are 8′×40′ shipping containers — the company's own specification. At these volumes there is no single-plant option in the conventional sense: supplying a tenth of a country's aviation fuel takes between fourteen and a hundred and nineteen base units. Because each goes to the demand point, the road haul and the remote-delivery premium are not reduced but designed out.

The tension worth naming: the scaling rule rewards building one large plant, while the logistics argument rewards dispersing many small ones. Both are true, and they pull in opposite directions. The resolution depends on how concentrated the demand is: a single airport justifies one scaled plant on site; a national road network does not.

What the ten locations show together

Power is not the constraint; margin is. Every airport programme in this list draws between 0.4% and 2.3% of its national grid — large, but absorbable, and Iceland's 25 MW is under one percent of a grid with surplus renewable capacity.

Only two of the ten support aviation on price. Iceland (32% kerosene margin) and Cambodia (41%) clear the local jet price. Finland, Norway and Myanmar are thin. Sri Lanka, Spain, Sweden, the Netherlands and Portugal lose money against untaxed jet fuel — for those five the airport programme only works under a RefuelEU-style mandate, where the buyer must comply and the alternative is a penalty rather than cheaper fossil fuel.

Iceland is the entry point. Fourteen base units, or one plant scaled to the same output for $52M, at Keflavík — the only aviation programme in the database under $100M, in a country with cheap unsubsidised power, no fuel subsidy, all three products viable, and 0.9% of the grid required.

Per litre of fuel produced

Source: fuelfromair.com FUELS array — FFA's own per-fuel constants.

Per 3 t/day unit, per year

One unit produces 759,921 litres (200,750 US gallons) of finished fuel a year. On the diesel basis that draws 1,996 tonnes of CO₂ from the atmosphere and returns 2,146 tonnes of pure O₂ to it. On the kerosene basis — the aviation case — it is 1,919 tonnes and 2,064 tonnes. A single unit handles the carbon of roughly 430 cars.

What displacement means for a single car

A car covering 12,000 miles a year at 25.5 mpg burns 470.6 gallons. On fossil gasoline that adds 4.09 tonnes of new carbon to the atmosphere — carbon that had been underground for millions of years. On FFA gasoline the same 4.09 tonnes is drawn from the air to make the fuel and returned when it burns. Net addition: zero.

A correction worth making internally

Arizona Fuels Corporation's CO₂ Extraction deck states that each car running on AFC fuel “will remove 13,176 lbs of CO₂ from the atmosphere each year”, and that the US fleet would remove 2,423 million tonnes annually. Neither figure survives scrutiny, and both should be corrected before the deck is shown to an investor or a regulator.

The arithmetic does not match the slide. 470.6 gallons × 28 lbs is 13,177 lbs of CO₂ consumed as feedstock. The same slide states that burning the fuel emits 8,188 lbs. A removal figure would be the difference — 4,988 lbs, not 13,176. The larger number is gross consumption presented as net removal.

And even 4,988 lbs is not removal. A circular fuel cannot be a net sink: the carbon that goes in comes back out when the fuel is burned. Cross-checked against EPA figures, a gallon of gasoline contains about 2.42 kg of carbon — 19.6 lbs of CO₂. The deck's 28 lbs is 1.43× that, and the excess is process loss, not sequestration. It does not stay out of the atmosphere.

The honest claim is stronger than the overstated one

“Removes 2,423 million tonnes” is false, and a regulator or a serious investor will find that out in an afternoon. “Eliminates 1,700 million tonnes of new fossil carbon a year” is true — and it is the entire decarbonisation case for US ground transport in a single line.

FFA does not need to claim removal. Displacing every tonne of new carbon from a sector that cannot electrify is a complete argument on its own, and it is one that survives audit.

Fleet scale

Market figures from Arizona Fuels Corporation, CO₂ Extraction (2015 basis). Displacement computed on fuelfromair.com's own constants: 2.296 kg CO₂ per litre of gasoline (8.691 kg/gal, 19.2 lbs/gal). Note this differs from the AFC deck's 28 lbs/gal; the site figure is used throughout this analysis.

The honest framing — circular, not sequestration

This is the point on which the environmental claim stands or falls, and it should be stated before anyone else states it.

FFA's carbon is circular, not sequestered. The CO₂ drawn from the air to make a litre is released again when that litre is burned. Fuel From Air is not a carbon sink and does not claim to be one.

The benefit is displacement. A litre of fossil diesel adds 2.626 kg of new carbon to the atmosphere — carbon that sat underground for millions of years. A litre of FFA diesel recycles 2.626 kg that was already in the air. Net addition: zero. That is a real and substantial benefit, and it is a different claim from removal.

The oxygen is a genuine by-product. Splitting water to make hydrogen releases O₂, and the process vents it. It is not a carbon claim and should not be presented as one — but it is a real output, and industrial and medical oxygen is a real market worth investigating as a by-product line.

Circular carbon against fossil — the comparison

What a carbon credit actually is

A carbon credit is one tonne of CO₂ that either was not emitted, or was taken out of the air and kept out. Those two things sound similar and price completely differently.

Avoidance credits represent emissions that did not happen — a forest not felled, a coal plant not built. They are abundant, contested, and cheap: generic avoidance trades below $5 a tonne, and the weighted average spot price across the whole voluntary market is about $5.60.

Removal credits (CDR) represent a tonne physically extracted from the atmosphere and stored. They are scarce and expensive: biochar $189–200, enhanced rock weathering $400, direct air capture with storage $450 to over $1,000 a tonne in 2026 forward offers.

Compliance markets are a separate system entirely. EU ETS allowances are projected to average $91–93 a tonne in 2026 and are set by auction, not project economics. They cannot be used against a voluntary net-zero claim, and voluntary credits cannot be surrendered against an ETS obligation.

The eligibility question — and it is the whole question

The EU's Carbon Removals Certification Framework (Regulation 2024/3012) sets three durability tiers for a removal credit: permanent removals requiring centuries of storage, carbon storage in products requiring a minimum of 35 years, and carbon farming requiring a minimum of five.

FFA's carbon comes out of the air and goes back into it the moment the fuel is burned — a matter of weeks. It meets none of those tiers.

Fuel From Air cannot sell removal credits on its fuel carbon. Any model showing Pillar 4 earning $200 a tonne of removal credits is modelling revenue that does not exist.

The price ladder, and where FFA can actually play

The gap between the placeholder and the reality is a factor of 25 to 36.

What FFA can legitimately sell

1. Avoided emissions, not removals. Every litre of FFA fuel displaces a litre of fossil fuel that would have added new carbon. That avoided tonne is real and saleable — but it prices as avoidance, at single digits per tonne, not as removal.

2. The fuel itself, into a mandate — and this is where the value sits. Under RefuelEU Aviation the buyer must blend synthetic fuel regardless of price; the alternative is a penalty, not cheaper fossil. FFA is not selling a credit at all here — it is selling a compliance product at a premium. That is Anchor 2, and it is worth far more than any credit line.

3. Aviation compliance channels. CORSIA Phase I (2024–2026) has created the first concrete compliance demand in aviation. Decisions on including CDR in the UK SAF mandate and in national ETS programmes could open further channels — but for the fuel, not for FFA's circular carbon.

4. The oxygen, as an industrial commodity. Each unit vents roughly 2,064 tonnes of pure O₂ a year. Medical and industrial oxygen is a genuine market and worth investigating as a by-product line. It is not an environmental credit and must not be dressed as one.

Marketing to large users — the realistic route

The question of whether FFA can offset and market to big users has a clear answer, but not the expected one. A corporate buyer with a net-zero commitment wants removal, and increasingly wants durable removal — the Oxford Principles push buyers up the durability ladder, not down it. FFA cannot supply that.

What FFA can supply is a fuel with a materially lower carbon intensity than the fossil alternative. An airline, a haulier or a mining operation does not want a certificate — it wants to reduce the emissions it reports, and buying FFA fuel does that directly and verifiably. The transaction is a fuel supply contract, not a credit trade.

That is a better business in any case. Credit markets are volatile, contested and thin: the voluntary market is worth around €3bn in 2026 against a global fuel market measured in trillions, retirements fell 7% last year, and integrity standards are tightening against exactly the kind of claim a circular fuel might be tempted to make. Selling fuel into a mandate is a larger, firmer and more defensible position than selling credits into a market that would rightly question them.

Atmospheric exchange — per unit and per programme

One 3 t/day unit draws 1,745–1,996 tonnes of CO₂ a year depending on product, and returns 1,983–2,146 tonnes of pure oxygen.

Circular carbon, not sequestration

A litre of fossil diesel adds 2.626 kg of new carbon. A litre of FFA diesel recycles 2.626 kg already in the air. Net addition zero — displacement, not removal.

The carbon price ladder and Pillar 4 revisited

Removal credits price at $189–1,000 a tonne but require durable storage FFA cannot provide. Avoidance credits are open to FFA at $5.60–8. The gap against the plan's $200 placeholder is 25–36×.

Electricity tariff — the basis that decides everything

Electricity is almost the whole cost of the fuel. The original 394-location database holds retail tariffs — its business column is labelled a proxy in FFA's own workbook, median $0.185/kWh. A plant operator signs a power-purchase agreement: real corporate PPAs are far below retail, and FFA's own workbook records the Cambodia National Solar Park. This analysis uses a representative PPA as the base case throughout, with retail shown as a sensitivity (basis available on request). On the proxy, the case is understated several-fold — gasoline reads $2.59/L instead of $0.73/L, and viability 8 of 68 instead of 63.

Is a PPA location-dependent? Yes — sharply

A PPA prices the local generation cost, resource quality and cost of capital — none of which are uniform. Real 2026 PPAs run from $0.0163/kWh in Saudi Arabia to $0.1523 in Sweden, a 9.3× spread. In five countries — Iran, Cambodia, Finland, Norway and Sweden — the PPA is dearer than the retail tariff, so an operator there signs a retail contract instead. The model uses min(PPA, retail) at every site.

Electricity price by location

The 15 cheapest and 15 most expensive sites, consumer against contractor tariff. The green line is the tariff implied by FFA's own $1.90/gal figure — very few sites reach it.

Production cost by location

The same sites, converted to production cost. Two reference lines: FFA's stated $1.90/gal ($0.50/L) and the $5.50/gal ($1.45/L) retail price. A site only works if its bar sits left of the retail line.

Where the economics work — all 394 original locations

Every site plotted against its tariff. 93 of 394 produce below the retail price; 301 do not. Knowing which is which is what Pillar 2 licenses.

Top 20 countries by best site

The cheapest-power location in each country, labelled with its tariff. These are P1's deployment candidates.

Production cost by fuel — 20 cheapest-power countries

All three products at each site, with the local pump marked. Diesel viable in 11 of 68 countries, gasoline 8, kerosene 5.

Carbon balance by fuel

CO₂ consumed and O₂ vented per unit produced. Diesel removes the most carbon per litre (2.626 kg/L, 9.94 kg/gal) — which is Pillar 4's basis.

Monetisation targeting map — four groups

All 394 original locations by distance from breakeven, per fuel. Diesel gives 73 profitable today plus 44 within 10% — 117 live or near-live targets.

Every location ranked by margin

The full distribution for gasoline. 55 sites profitable, 48 within 10%, 45 within 30%, 246 beyond.

Group 1 deep dive — 73 sites, 11 countries

Ranked by payback rather than margin, the order changes completely. Cambodia has the best margin (56%) but Iceland pays back twice as fast, because Iceland clears on a $2.35/L pump rather than a cheap tariff.

Can FFA undercut the local pump?

Every location with a known pump price. Every point sits above the parity line.

FFA synthetic petrol versus local pump, by country

Best site per country. The ratio label shows how far short each falls.

Why local-electricity petrol is hard

The mechanism: pump price and electricity price correlate at r = 0.62. Where fuel is worth replacing, power is dear; where power is cheap, fuel is cheaper still.

What would close the gap?

At 90% yield and $0.10/L conversion, gasoline falls to $0.65/L — below the US pump. The route is unproven rather than hopeless.

The tariff threshold for pump parity

How cheap power must be for FFA to match the pump at each price level.

Is the local pump price subsidised?

IEA/IMF price-gap method. Only Iran, Brunei, Malaysia, Egypt and Saudi Arabia price below cost recovery. The other 63 tax fuel above it.

The subsidy trap — why this is structural

The single chart that explains the problem. Cheap-power countries subsidise their fuel; fuel-taxing countries have dear power. FFA is squeezed at both ends — but 14 of its 19 viable countries are not subsidised.

Set A — with subsidy (the market as it is)

Actual 2026 pump prices, including government subsidy where it exists.

Set B — subsidy removed

Subsidised countries repriced to cost recovery. The result barely moves: 65 of 68 with subsidy, 67 without — removing subsidy changes almost nothing.

Brent crude, 2016–2026

Mean $69.45, standard deviation $17.02, high 2.3x the low. Mean absolute annual move 26%.

What oil volatility means for FFA

FFA's cost is flat against oil, so volatility is pure upside optionality. But petrol needs Brent at $168 — 70% above the highest annual average ever recorded.

The 2026 stress test — when oil supply fails

The Strait of Hormuz closure: 14 mb/d shut in, supply down to 94.5 mb/d, OECD stocks at their lowest since 1990.

Supply security — FFA's cost does not move when oil does

Through a doubling and halving of Brent, FFA held at $1.25/L. At the $144 April peak fossil reached $1.11/L — still short. The worst crisis on record got fossil to 86% of breakeven and no further.

Three route-to-market anchors

Annual EBITDA per $4.5M unit. Same plant, same cost base; only the customer and the pricing basis change. Petrol on the open market is the only losing case.

Anchors × pillars — where each pillar finds its buyer

P3 Aviation on Anchor 1 is the single red cell — and precisely the case the pump analysis dissects. P3 survives because Anchor 2 is its route.

Demand cases — what one unit serves

One unit serves 427 cars or 13 trucks; a regional jet needs 10 units, a transatlantic jet 32.

Location versus containers and capex — 10% of national demand

Units and capital to supply a tenth of each country's demand, road against aviation.

The entry point — smallest achievable programmes

Iceland's aviation sector: 14 units, $64.9M. The only ten-percent target in the database under $100M.

One plant as a share of national road-fuel demand

Even the largest FFA unit supplies under a hundredth of a percent. Scale is the constraint, not cost.

Scaling rule — cost per gallon and capex

Linear output with a capped cost reduction; consistent with FFA's own note on minimal economies beyond 3 t/day.

Scaled plant versus many base units

The scaling rule cuts Iceland's aviation capex from $64.9M to $55.2M.

Incumbent cost stack versus FFA

What a location spends now. The amber block is distribution and marketing — the layer FFA removes by producing on site.

National fuel spend and the distribution component

Total spend on road and aviation fuel, and how much of it is distribution rather than product.

Logistics cost avoided by siting at the demand point

Honest caveat: $0.2–3.6M a year against $3–30M of capex. It supports the siting decision; it does not justify the investment.

Real PPA prices versus the database retail proxy

The database carries retail tariffs. On a real PPA, viability goes from 11 to 63 of 68 countries for diesel.