Mazda's Onboard CO₂ Capture: Why Carbon-Negative Racing on HVO Does Not Add Up Yet
Mazda captured 804 g of CO₂ in a 24-hour race. Carbon-negative operation needs about 20 percent capture and durable storage. Neither is in place.
Mazda Motor Corporation (TYO:7261) has put a number on a claim that most of the industry has left vague. Its onboard CO₂ capture system recovered 804 grams of carbon dioxide over the Fuji 24 Hours in June 2026, running a race-prepared Mazda3 on hydrotreated vegetable oil, and the company says the result confirms the potential for carbon-negative operation in production vehicles, albeit for a limited period [1]. The claim is testable, though it fails two tests. The captured mass is at most about 0.3 percent of the car's exhaust CO₂, against a break-even threshold near 20 percent for the best commercial HVO. None of the uses Mazda has named for the captured carbon stores it durably. A physically genuine carbon-negative pathway exists, but it is narrower than the announcement implies and it is not reachable by the November 2026 target.
What Mazda demonstrated
The Mazda Mobile Carbon Capture system adsorbs CO₂ from the exhaust onto porous zeolite, releases it using exhaust heat during driving, and compresses it into an onboard tank with an electric compressor [1]. It ran on the No. 55 MAZDA SPIRIT RACING 3 Future concept in Round 3 of the 2026 Super Taikyu Series at Fuji from 5 to 7 June, capturing 804 grams across the 24 hours, about 9.6 times the 84 grams recovered during a four-hour, 120-lap race in November 2025 [1][2][3]. Mazda describes the fuel as a carbon-neutral fuel already in practical use in Europe and states an ambition to realize mobility that reduces CO₂ the more you drive by 2035, with carbon negativity in racing targeted for Round 7 in November 2026 [1][4].
The arithmetic behind the ambition was set out at the 2025 Japan Mobility Show alongside the Vision X-Coupe concept. Carbon-neutral fuel cuts CO₂ by about 90 percent, onboard capture recovers 20 percent of the CO₂ in the exhaust, and the two together give 110 percent, which Mazda presents as 10 percent carbon negativity [4]. Engineers have illustrated the result for a B-segment car emitting about 100 grams per kilometre, which would become approximately negative 10 grams per kilometre [3][5]. Australian motoring press reporting from the same event puts the prototype device at about 50 kilograms, raising fuel consumption by 2 to 3 percent, with a tank that fills within 50 to 200 kilometres [5]. Those figures come from company briefings rather than published test data.
The foundation: what HVO actually emits
Two figures govern everything that follows, and both come from the regulatory lifecycle framework rather than from Mazda.
The first is the tailpipe. HVO is roughly 85 percent carbon by mass with a lower heating value near 44 megajoules per kilogram, so combustion releases about 71 grams of CO₂ per megajoule of fuel energy. That carbon is biogenic, absorbed by the feedstock while growing, and every major framework treats it as neutral. Uptake cancels exactly that quantity and nothing more.
The second is what remains on the books. Against the European Renewable Energy Directive comparator of 94 grams of CO₂-equivalent per megajoule [6], attributional central estimates place used cooking oil HVO near 14 g CO₂e/MJ, an 85 percent saving [6][7][8]. Intermediate winter canola reaches about 32 g [9], U.S. soybean about 58 g once central indirect land-use change is counted [8][10], main-season rapeseed about 70 g [6][8], and palm oil exceeds the fossil baseline at about 105 g [6][8]. Tallow sits near 19 g [6][8]. These are fertilizer, soil nitrous oxide, hydrogen, process energy, freight and land-use emissions that no amount of tailpipe capture addresses, because they were emitted before the fuel reached the tank.
No commercial HVO pathway is carbon-neutral. The credible claim for the best of them is a reduction of roughly 75 to 88 percent per megajoule, which is a different statement from neutrality and a very different one from negativity.
Why the 110 percent does not hold
Mazda's sum adds two percentages measured against different bases. The 90 percent is a lifecycle saving against a fossil comparator. The 20 percent is a share of the carbon leaving the tailpipe. Restating both per megajoule resolves the mismatch.
A fuel achieving a 90 percent saving against 94 g leaves about 9.4 g CO₂e/MJ of upstream burden. Capturing 20 percent of the 71 g of biogenic CO₂ at the tailpipe removes about 14.2 g. The net is about negative 4.8 g CO₂e/MJ, roughly half the negativity the 110 percent framing implies. The direction is right and the magnitude is not.
That 90 percent is itself only available at the favorable end of the evidence. It corresponds to waste-oil HVO on attributional accounting, where Germany's Federal Office for Agriculture and Food reported average savings of 87.16 percent for waste-based HVO in 2023 [7]. Mazda has not disclosed the feedstock or supplier of its race fuel, and that omission decides the whole calculation. Asian-sourced used cooking oil, once displacement of its former uses is counted, carries a consequential intensity that the International Council on Clean Transportation puts at an additional 12.2 g CO₂e/MJ for the U.S. case [11], and that a 2025 study for Deutsche Umwelthilfe finds can erase the benefit entirely when palm oil replaces the diverted material in origin markets [7].
The break-even rule
A simpler test follows from the same two numbers, and it generalizes to any onboard capture claim from any manufacturer.
A vehicle is carbon-negative only when its sustained capture rate exceeds the fuel's lifecycle carbon intensity divided by the tailpipe CO₂ per megajoule, which is about 71 g. Everything upstream of the tank must be paid for out of carbon caught at the exhaust.
Applying that rule to commercial pathways gives break-even capture rates of about 20 percent for used cooking oil HVO on attributional accounting, about 27 percent for tallow, about 37 percent for U.S. used cooking oil once displacement is counted, about 45 percent for intermediate canola, about 82 percent for U.S. soybean, and essentially 100 percent for main-season rapeseed [6][7][8][9][10][11]. Palm oil with indirect land-use change cannot break even at any capture rate, because its lifecycle intensity exceeds the carbon available at the tailpipe.
Mazda's 20 percent target therefore lands exactly on the boundary for the cleanest commercial fuel and nowhere near it for anything else. At that rate, attributional used cooking oil HVO reaches approximately zero, not negative, and only if the captured CO₂ never returns to the atmosphere.
The reported 2 to 3 percent fuel consumption penalty from carrying the device [5] does not move the threshold, since it raises fuel burned and CO₂ available for capture in the same proportion, but it enlarges the net burden at every capture rate below break-even.
What the race actually showed
Mazda has published neither the car's fuel consumption nor its distance, so the capture share must be bounded rather than calculated. Assuming only 2,000 kilometres over the 24 hours, well below the pace implied by the November race, and a road-car consumption of 6 litres per 100 kilometres, the car burned roughly 120 litres of HVO. At about 34 megajoules per litre [6], that is around 4,100 megajoules, releasing approximately 290 kilograms of biogenic CO₂. The 804 grams captured is therefore at most about 0.3 percent of exhaust carbon, and less at realistic race consumption and distance.
Expressed another way, 804 grams equals about 40 kilometres of capture at the 20 grams per kilometre that Mazda's own target implies for a 100 g/km car [3]. Reaching the 10 percent capture rate that the company's arithmetic requires would take a gain of at least about 35 times by November. Reaching the 20 percent needed for break-even on the best HVO would take at least about 70 times. The 9.6-fold improvement between November 2025 and June 2026 is substantial engineering progress and remains two orders of magnitude short.
Mazda's own account is consistent with these bounds. The phrase "temporarily exceeded the target recovery level" indicates capture at target rates over short windows rather than sustained [1], and the company lists water cooling, removal of moisture that interferes with capture, and a means for customers to empty the tank among the problems still to be solved before production [2].
The permanence problem
Even at target capture rates, the claim would fail on storage. The uses Mazda has described for the recovered CO₂ are growing microalgae for future fuel, enriching greenhouse crops such as tomatoes and strawberries, producing carbon materials for vehicle parts, and supplying raw material for fertiliser and for plastics including disposable cutlery [3][4][5].
Greenhouse and fertiliser CO₂ returns to the atmosphere within weeks to months. Disposable plastics release their carbon at incineration or through decay. Carbon materials in vehicle parts last a vehicle's service life at most. None of this is durable removal, which conventionally means storage measured in centuries, in geological formations, mineral form or genuinely stable materials.
The microalgae route carries an additional accounting hazard. If exhaust CO₂ feeds the algae that become the next tank of fuel, that carbon cannot be counted once as a tailpipe removal and again as the biogenic uptake that makes the fuel neutral. Utilization of this kind can at most displace CO₂ that would otherwise have been sourced elsewhere, which earns a substitution credit. That is an accounting route to a lower reported figure, not a physical removal from the atmosphere.
What would make it viable
The mechanism Mazda is pursuing is sound in principle. Under the biogenic convention, the roughly 71 g of CO₂ per megajoule leaving a biofuel tailpipe has already been netted against feedstock uptake. Capturing part of it and storing it durably is true removal; it's bioenergy with carbon capture and storage relocated from the plant to the vehicle. It is also the only stream in the HVO chain large enough to matter, since the process CO₂ available for capture inside an HVO refinery amounts to at most about 4 g per megajoule of fuel on a stoichiometric basis.
Four conditions would have to hold at once.
The fuel must sit near the near-zero threshold, meaning waste-oil HVO made with renewable electrolytic hydrogen and renewable process energy from verifiably surplus feedstock, at roughly 3 to 5 g CO₂e/MJ [7]. That alone would cut the break-even capture rate to about 4 to 7 percent, within reach of the announced hardware.
Capture must run above that rate across a full duty cycle, net of the device's own energy use and embodied emissions, not in temporary windows.
The captured CO₂ must reach geological storage or durable mineralization, with monitoring. No such route has been named.
The feedstock chain must be disclosed and assessed on consequential as well as attributional accounting, because a fuel that is clean on paper and displaces palm oil in its origin market is not clean.
The first condition also caps the scale of anything built on it. Waste and residue supply does not grow with demand, and Transport & Environment has estimated that Europe used close to seven million tonnes of used cooking oil for biofuels in 2023, roughly eight times estimated collected volumes [12]. Additional HVO demand is met at the margin by crop oils, whose break-even capture rates of 45 to 100 percent place carbon-negative operation out of reach.
The claims environment is tightening
As of late 2026, Directive (EU) 2024/825 applies across member states. It prohibits claiming, on the basis of greenhouse-gas offsetting, that a product has a neutral, reduced or positive environmental impact, and it restricts generic environmental claims that cannot be substantiated [13][14]. Mazda's announcements are made in Japan, and the Directive governs consumer-facing claims in the European Union, but describing HVO to European buyers as a carbon-neutral fuel now faces a substantiation test that no commercial pathway currently passes.
The credible claim
Mazda has no viable route to carbon-negative racing by November 2026 in any sense a lifecycle auditor would accept. It could satisfy its own definition if the metric were an instantaneous capture rate, a limited window, or a tank-to-wheel boundary that ignores everything upstream of the fuel filler. Those are not the terms under which buyers, regulators or investors should read the claim.
What Mazda can credibly say is what every HVO user should say: a stated percentage reduction against a named comparator, with captured CO₂ reported as a separate, gross quantity until durable storage exists to convert it into a removal. On the evidence available, 804 grams over 24 hours is a promising engineering result and not a carbon-negative vehicle.
John D
John D
John D
References
[1] Mazda Motor Corporation. 2026. "Mazda Advances Toward Commercialization of Onboard CO₂ Capture System." News release, 8 June. https://newsroom.mazda.com/en/publicity/release/2026/202606/260608a.html
[2] Mazda Motor Corporation. 2025. "Creating a Positive Future for Combustion Engine Cars: Mazda's Public Proof-of-Concept Testing for Carbon Neutral Combustion Engine Car." MAZDA MIRAI BASE, 10 December. https://www.mazda.com/en/mazda-mirai-base/articles/20251210-CO2-capture/.
[3] Ward, James. 2026. "Mazda's New Technology That Could Keep Petrol Engines Alive in an Electric Age Is Now Race-Proven." Drive, 14 June. https://www.drive.com.au/news/mazdas-carbon-capture-system-just-proved-itself-in-a-24-hour-race-and-captured-nearly-10-times-more-co2-than-before/.
[4] Mazda Motor Corporation. 2025. "Japan Mobility Show 2025: The More You Drive, the Cleaner the Planet? The MAZDA VISION X-COUPE and the Future of Joy of Driving." MAZDA MIRAI BASE, 29 October. https://www.mazda.com/en/mazda-mirai-base/articles/20251029-jms2025-mazda-vision-xcoupe/.
[5] Law, John. 2025. "Emissions Breakthrough Promises to Beat Electric Cars, but Is Mazda's Algae Fuel and Carbon Capture Really a Silver Bullet?" Chasing Cars, 2 November. https://www.chasingcars.com.au/news/car-industry/emissions-breakthrough-promises-to-beat-electric-cars-but-is-mazdas-algae-fuel-and-carbon-capture-really-a-silver-bullet/.
[6] European Parliament and Council of the European Union. 2018. "Directive (EU) 2018/2001 of the European Parliament and of the Council of 11 December 2018 on the Promotion of the Use of Energy from Renewable Sources (Recast), Annex V." Official Journal of the European Union L 328. Consolidated text, 7 June 2022. https://www.seai.ie/sites/default/files/publications/RED-II-directive-ANNEX-V-RULES-FOR-CALCULATING-THE-GREENHOUSE-GAS-IMPACT-OF-BIOFUELS-BIOLIQUIDS-AND-THEIR-FOSSIL-FUEL-COMPARATORS.pdf.
[7] ifeu (Institut für Energie- und Umweltforschung Heidelberg). 2025. Carbon Footprint of Biodiesel and HVO Based on Used Cooking Oil (UCO). Study for Deutsche Umwelthilfe. https://www.duh.de/fileadmin/user_upload/download/Projektinformation/Naturschutz/Agrokraftstoffe/2025-08-07_DUH_ifeu_study_GHG_balance_UCO_english.pdf.
[8] International Civil Aviation Organization. 2022. CORSIA Default Life Cycle Emissions Values for CORSIA Eligible Fuels. June. https://www.icao.int/sites/default/files/environmental-protection/Documents/ICAO-document-06-Default-Life-Cycle-Emissions-June-2022.pdf.
[9] Kwon, Hoyoung, Xinyu Liu, Hao Cai, and Michael Wang. 2026. "Life Cycle Analysis of Growing Canola for Biofuel Production in the United States." ACS Sustainable Chemistry & Engineering 14 (24): 11209–19. https://doi.org/10.1021/acssuschemeng.6c03620.
[10] Li, Yuan, Hoyoung Kwon, Michael Wang, et al. 2023. "Life Cycle Analysis of Renewable Diesel from Soybean Oil." Argonne National Laboratory. https://pubs.acs.org/esthag/article-pdf/56/12/7512/12960473/es2c00289.pdf
[11] O'Malley, Jane, Stephanie Searle, and Nikita Pavlenko. 2021. Indirect Emissions from Waste and Residue Feedstocks. International Council on Clean Transportation. https://theicct.org/publication/indirect-emissions-from-waste-and-residue-feedstocks/.
[12] Transport & Environment. n.d. "UCO (Unknown Cooking Oil): High Hopes on Limited and Suspicious Materials." https://www.transportenvironment.org/articles/uco-unknown-cooking-oil-high-hopes-on-limited-and-suspicious-materials.
[13] European Parliament and Council of the European Union. 2024. "Directive (EU) 2024/825 of the European Parliament and of the Council of 28 February 2024 Amending Directives 2005/29/EC and 2011/83/EU as Regards Empowering Consumers for the Green Transition." Official Journal of the European Union L, 6 March. https://eur-lex.europa.eu/eli/dir/2024/825/oj.
[14] Government of Ireland, Department of Enterprise. n.d. "Directive (EU) 2024/825: Empowering Consumers for the Green Transition." Accessed 26 September 2026. https://enterprise.gov.ie/en/what-we-do/the-business-environment/empowering-consumers-for-the-green-transition/.
