The Infrastructure Carbon Curve

Ten infrastructure build-outs, ten abatement-cost curves. The pattern is consistent: a slug of carbon can be cut at negative cost using recycled materials that are cheaper, not dearer. Do this before any premium tonne is bought.

A plain-language MACC analysis across roads, bridges, water, wastewater, grids, power, rail, ports, data centers & buildings. This means built & operational carbon, $/tCO₂e, cost per functional unit, and seven regions (US, EU, Asia, Australia, India, Japan, Brazil). Built on Infrastructure Australia, Carbon Leadership Forum/RMI, GCCA, IEA & Ember data (2022–2026)1,2,3 — with CarbonSig research-tool notes.

Bottom Line Up Front

The cheapest carbon is the carbon you don’t buy

The argument in one breath. Across all ten infrastructure types, the abatement-cost curves share a shape: a band of cost-negative levers (recycled aggregate, reclaimed asphalt, lower-clinker concrete, structural lightweighting) sits below the zero line cutting carbon while saving money.

This followed by a rising premium for the deeper cuts (electrified fleets, green steel, bio-fuels, CCS). Infrastructure Australia found four of eleven material strategies are net cost-saving and one is cost-neutral;5RMI/CLF reach the same verdict for buildings.4The strategic implication: most of the early abatement is a procurement decision, not a cost. Two numbers set the scale: embodied carbon is already ~10% of national emissions in a developed economy, and it is locked in the day the asset is built. Unlike operational carbon, it cannot be cleaned up later by a greener grid.

This brief gives each of ten build-out types its own marginal abatement cost curve (MACC): levers ranked cheapest-first, bar width proportional to how much carbon each can cut, with the zero line separating money-saving moves from premium ones. Each curve is paired with the functional units stakeholders actually budget in — $/km, $/MW, $/m³, $/m² and both embodied (built) and operational carbon. A closing regional map shows how the same asset’s footprint shifts across the US, EU, Asia, Australia, India, Japan and Brazil, and how CarbonSig is used to model any of it at the project level.

The framing treats carbon as money. Where embodied carbon is priced using Buy Clean procurement (request CO2e declarations on offer), CBAM on materials, low-carbon-concrete mandates the MACC stops being an environmental chart and becomes a cost curve a project engineer optimizes against, exactly like a bill of materials.

Method · How to read this report

How to read a MACC and what’s in each cost

A marginal abatement cost curve ranks every carbon-cutting lever by its cost per tonne avoided ($/tCO₂e), cheapest first, with each bar’s width showing how much carbon that lever can cut. Levers below the zero line are net cost-saving; above it, they carry a premium that rises steeply for the last tonnes. McKinsey originated the format in 2007; it is now standard in net-zero roadmaps.

Read: Work left-to-right and stop where the curve crosses your carbon price. Everything left of that point pays for itself or beats your price; everything right of it is a deliberate premium for deeper cuts.

Cost basis: each curve covers both embodied (materials + construction) and operational (energy + maintenance over the asset life) carbon; $/tCO₂e is the lever’s net lifecycle cost per tonne avoided. Values are indicative syntheses of the cited sources, normalized for comparability across sectors.

The two cost bases, kept separate Embodied (built) carbon is fixed the day the asset completes — it can only be designed out, never retrofitted away. Operational carbon (pumping, lighting, traction, cooling, 40-year maintenance) depends on the grid and decays as electricity decarbonizes. The split matters: embodied-dominated assets (roads, bridges) reward material choices; operational-dominated assets (water pumping, data centers) reward energy choices. Each section flags which kind it is.

Where the carbon and the free savings sit

Before the ten curves, one orienting picture: infrastructure types differ sharply in whether their carbon is embodied (locked in materials) or operational (burned over decades of use). That split decides which levers matter and how much free abatement is on the table.

Read: Embodied-dominated assets (bridges 88%, ports 82%, roads 80%, rail 75%) are won or lost at the material-procurement stage: recycled steel, low-clinker concrete, lightweighting.

Operational-dominated assets (data centers ~82% operational, wastewater ~70%, water ~65%) are won at the energy stage using clean power and pump/process efficiency. Buildings and power-plant builds sit in the middle.

Source: Infrastructure Australia (embodied locked at build); IEEE/Schneider (data-center ~60% operational, ~40% embodied including devices); CLF benchmarks. Splits indicative.

Why this ordering matters for spend For the embodied-heavy four, the cost-negative levers (recycled materials) capture real abatement on day one with no grid dependency are the fastest, cheapest wins in the whole report. For the operational-heavy assets, the lever is a power contract and an efficiency design, and the payoff compounds as the grid cleans up. CarbonSig models both in one canvas (see closing section).

Roads & highways

Roads are embodied-dominated: most whole-life carbon is in pavement materials and earthworks, with operational carbon from lighting and 40-year maintenance cycles. Whole-life carbon runs ~800–2,700 tCO₂e/km depending on scale (single-2-lane to dual-3). The cheapest cuts are reclaimed asphalt and recycled aggregate these can be cost-negative.

Read: Roads are the clearest cost-negative case: reclaimed asphalt and recycled crushed concrete reduce material spend while cutting carbon (Infrastructure Australia). The premium tonnes (fleet electrification, bio-fuels) wait until grids and HVO supply mature.

Bridges & elevated structures

Bridges are the most embodied-dominated infrastructure. Nearly all carbon is in structural concrete and steel, with negligible operational energy. Span choice and material strength dominate: lower concrete grades and steel right-sizing cut carbon at little or no cost; green steel and low-carbon concrete carry a modest premium.

Read: Because operational carbon is near zero, a bridge’s whole-life answer is set entirely at design. Site/span selection alone can swing embodied carbon materially (BSCES);82green steel is the big-ticket residual lever.

Water supply & treatment

*Read: Efficiency is genuinely cost-negative* as every kWh saved cuts both the power bill and the operational carbon. The grid factor is decisive: the same plant’s operational carbon varies ~6× across regions (see regional exhibit).Infrastructure 04**

Wastewater & sewerage

Wastewater is strongly operational-dominated, with a twist: aeration electricity and direct process emissions (CH₂ and N₂O from treatment) that no grid greening touches. The curve mixes energy levers with process-control and biogas-capture levers.

Read: Biogas capture can make a plant energy-positive — a cost-negative lever that also displaces grid power. Process N₂O is the hard residual: high GWP, no cheap fix, a prime target for measurement & control.

Electricity transmission & distribution

Read: Aluminium conductor is the embodied hotspot. Recycled or renewable-powered aluminium is the key materials lever (ties to the aluminium brief). Line losses convert directly to operational carbon at the local grid factor, so low-loss design pays twice.

Power generation build-out

For the physical build of generation, carbon is embodied in foundations, towers and equipment where operational carbon belongs to the fuel, not the structure. Embodied intensity per kWh delivered: wind ~10–15 g, solar ~40–70 g (life-cycle), so the lever set is concrete, steel and panel/turbine supply chains.

Read: The build-out’s own footprint is small per kWh but scales with the gigawatts deployed for the transition. Foundation concrete is the single biggest embodied item. Low-carbon mixes are the highest-leverage materials lever.

Rail & transit

Read: Tunnels and viaducts dominate embodied carbon with alignment choices (avoiding tunneling) are the largest single lever, set at planning. Electrified traction shifts the asset onto the grid’s decarbonization curve.

Ports & marine

Read: Marine concrete’s durability requirement keeps cement content high while low-carbon marine mixes are technically harder but high-impact. Shore power moves berth emissions onto the grid, compounding with grid greening.

Telecom & data centers

Read: Because operational carbon dwarfs embodied, where you site it (the grid factor) is the dominant lever. The same MW emits ~6× more on a coal grid than on Brazil’s or France’s. Efficiency (PUE, server life) is cost-negative; clean-power procurement is the strategic spend.

Buildings / social infrastructure

Read: The RMI finding is the headline: low-embodied-carbon buildings frequently cost the same or less when material efficiency is pursued early. Heat-pump electrification plus clean power closes the operational half.

The same asset, seven carbon footprints

Identical infrastructure carries very different carbon depending on where it is built and operated. Two regional factors drive it: the grid carbon intensity (which sets operational carbon and the carbon of any electrified equipment) and the material carbon intensity (local cement and steel, which set embodied carbon). A data center in Brazil and one in coal-heavy Asia can differ 5× in operational carbon for the same design.

Read: Operational carbon — and the carbon benefit of electrifying any fleet or process — scales directly with these numbers. The same electrified paving fleet or pumping station abates far more in Brazil (~103) or the EU than in coal-heavy Asia (~573), where electrification can even raise emissions until the grid cleans up.

Source: Ember Global Electricity Review 2025 (Brazil 103, Japan 482, China 560, Asia 573, global 473); IEA Electricity 2025 (global 445→400 by 2027); US/EU/India/Australia from Ember & Enerdata.

The two-factor regional rule of thumb Operational carbon follows the grid map above — build power-hungry assets (data centers, water/wastewater pumping, electrified rail) where the grid is clean (Brazil, EU, hydro-rich regions). Embodied carbon follows local cement and steel intensity with CBAM-style border pricing and Buy Clean rules are now making low-clinker concrete and recycled/green steel a procurement requirement in the EU and parts of the US, while India and parts of Asia carry higher default material intensities. The cost-negative recycled-material levers work everywhere, regardless of grid.

Read: In clean-grid regions, electrifying fleets and processes is a top lever; in coal-heavy grids, the same move waits behind recycled materials and efficiency, which save carbon regardless of the grid. CBAM (EU) and Buy Clean (US) are turning low-carbon materials from option into requirement.

*Source:* grid values per Ember; lever logic per Infrastructure Australia & GCCA.5, CarbonSig Research Tool**

CarbonSig as an infrastructure research tool

Every curve in this report is a generic answer. A real project: this highway, in this region, with this concrete supplier needs its own curve. CarbonSig is the tool that can help build it: a digital twin of the infrastructure asset where each material, fuel and process is a node carrying its own carbon factor, so a planner can test a lever and watch carbon, cost and $/tCO₂e re-price instantly with the MACC, computed live from the actual bill of materials. This is currently 2 seperate tools being integrated into one.

How it maps to this report: every lever in the ten curves becomes an editable node; embodied vs operational (cross-sector exhibit) and the regional grid factor (regional map) are input; the output is the project’s own MACC plus an auditable, Buy-Clean/output.

CarbonSig platform notes: any value chain digitally twinned with PCF/CI, CAP certificates, scenario modeling & 3rd-party verification.

What a planner, contractor or owner does with it

• Build the primary asset model. Assemble the materials→transport→construct→operate chain as connected nodes; enter the real concrete mix, steel source, fleet fuel and the regional grid factor; get whole-life baseline carbon per functional unit ($/km, $/MW, $/m³, $/m²).
• Generate the a project’s own MACC. Toggle each lever — recycled aggregate, low-clinker concrete, green steel, electrified fleet, clean-power PPA — and watch carbon, re-rank. The generic curves in this report become your curve for your bill of materials.
• Rank a portfolio. Compare projects by abatement-per-dollar; find the cost-negative tonnes first; stress-test against a carbon price or a Buy Clean threshold before procurement closes.
• Issue the proof. Turn the verified footprint into a Carbon Attested Product (CAP) — the EPD/Buy-Clean/CBAM evidence that could win low-carbon-procurement bids and survives third-party assurance.

The synthesis. Every infrastructure type has a cost-negative band of carbon to cut before any premium is paid — but the exact curve depends on the asset, the materials and the region. CarbonSig is tool that computes that specific curve, turns the generic findings of this report into a project decision, and issues the certificate that makes a verified low-carbon asset worth more than its high-carbon twin.

Carbon-Neutral Infrastructure

Removals close the gap: ISO 14068, and the cost of neutral

The ten cost curves answer “how far can we cut?” They never reach zero — every asset has a residual of embodied carbon (calcination CO₂ in cement, process emissions in steel) that no material substitution removes. The correct accounting standard for closing that last gap is ISO 14068–1:2023, which defines carbon neutrality through a strict hierarchy: reduce first, enhance removals second, and offset only the residual — increasingly with durable removal credits, not avoidance offsets. Pair the MACC (reduce) with a removal instrument (neutralize the residual) and an infrastructure asset can be made verifiably carbon-neutral — at a calculable, and falling, premium.

The argument in one breath. A road, bridge or data center that has exhausted its cost-negative and low-premium levers still emits a residual. Under ISO 14068–1:2023 that residual may be neutralized with high-quality carbon credits — and the standard pushes buyers toward permanent removals.

The cost depends entirely on the instrument: nature-based removals run $6–50/t, biochar ~$125–180/t, enhanced weathering ~$200/t, BECCS ~$390/t, and direct air capture (DAC) ~$300–600/t today.

Because the residual after a good MACC is small, even premium DAC adds a bounded, modelable cost while CarbonSig tools can compute the least-cost blend of reduction + removal that reaches neutral. The worked examples below cover all ten infrastructure types — each with its achievable reduction (and uncertainty) and the resulting neutrality premium.

Read: The MACC does the heavy lifting (here ~75% cut); durable removals neutralize only the ~25% residual that material and energy levers cannot reach. A smaller residual means a smaller removal bill. Spending on reduction first is what makes neutrality affordable.

Source: ISO 14068–1:2023 hierarchy (iso.org);13cement calcination ~40–50% of concrete CO₂ is process-inherent. Shares indicative.

Read: Cheap nature-based offsets ($6–50) carry permanence risk; ISO 14068 steers neutrality claims toward durable removals. DAC at ~$300–600/t is the most expensive but offers permanence, scalability and verifiability — the gold-standard tonne for a credible carbon-neutral asset. Sources: Sylvera 2026 (ARR $22, biochar $177, ERW $200+, BECCS $389, DAC $500+); IEA & ETH Zürich DAC $230–630 scaled (ETH 2024); Puro.earth CORC biochar index ~$125–145 (2025).

Read: ISO 14068 is what separates a defensible “carbon-neutral road” from a greenwash. Offsetting is permitted only for the residual that survives reduction and in-boundary removals, and the standard pushes the credit mix toward permanent removals over time.

Source: ISO 14068–1:2023 (iso.org/standard/43279); hierarchy & residual-only offsetting per ISO preview & Seedling/ECA summaries (Seedling 2026).

Read: Neutrality is an optimization, not a checkbox. CarbonSig sets the project’s marginal abatement cost against the marginal removal cost and solves for the cheapest mix that reaches zero under ISO 14068 and then issues the verified CAP. As DAC and biochar costs fall, the model will re-balance toward removals automatically.

Source: method follows the MACC-vs-backstop logic (McKinsey); CarbonSig CaRMa platform capabilities (digital twin, scenario modeling, CAP, ISO 14067/14068 alignment).

Read: The bridge is a deliberately hard case — embodied-dominated, with process CO₂ in cement/steel capping reduction near 45%. Even so, neutrality lands at ~1.7% (biochar) to ~6.9% (DAC high).

Sources: bridge carbon/cost from CLF/IStructE; DAC $300–600/t (IEA, ETH Zürich); biochar/BECCS (Sylvera 2026). Illustrative.

Read: Two patterns. (1) Capital-heavy, operational assets (water, wastewater, power, data centers) neutralize for well under 1% — their build cost dwarfs the residual removal bill. (2) Material-light, embodied assets (roads, bridges) carry the highest premium (up to ~7–8% with DAC) because their residual is large relative to a modest build cost and process CO₂ caps reduction. The lever everywhere: cut deeper to shrink the residual before buying removals.

Sources: functional-unit carbon & cost per the ten sections above (CLF/IStructE, IEA, Infrastructure Australia); DAC $300–600/t (IEA/ETH); biochar $150 (Sylvera). Reduction %, residuals & premiums are illustrative syntheses.

Why this matters for the carbon-as-money thesis Once an asset can be made verifiably carbon-neutral under ISO 14068 at a known, single-digit-percent premium, neutrality becomes a procurement option with a price tag — not an aspiration. The reduced-plus-removed asset earns the low-carbon premium, qualifies for Buy Clean / CBAM-aligned tenders, and carries a CAP certificate proving it. CarbonSig is where the reduction MACC and the removal ladder meet, so an owner can price neutral before breaking ground.

References

1. Infrastructure Australia — Embodied Carbon Projections for Australian Infrastructure and Buildings: embodied carbon ~10% of national emissions (2023, upfront 7%); four of eleven material strategies net cost-saving (recycled crushed concrete, reclaimed asphalt, structural steel lightweighting, hydrated lime), one cost-neutral; pipeline upfront 37–64 MtCO₂e/yr. link [ref 5]
2. Carbon Leadership Forum / RMI / Univ. of Washington — Embodied Carbon Pathways to 2050 (US); WBLCA Benchmark Study V2 (292 projects); SE 2050 database (1,000+ LCAs); low embodied carbon often achievable without cost premium. link [refs 4, 81]
3. Global Cement & Concrete Association (GCCA) / Sustainability Atlas (2025–26) — construction materials ~15% of global CO₂ (cement ~8%, steel ~7%); low-carbon cement green premium 10–25%; commercial low-carbon cements achieve 30–70% reductions vs OPC (OPC ~600–900 kg CO₂/t). [refs 3, 86]
4. Ember — Global Electricity Review 2025: 2024 grid intensity gCO₂/kWh — Brazil 103, EU low/declining, global avg 473, Japan 482, China 560, Asia avg 573. link [refs 1, 92]
5. IEA — Electricity 2025 (global CO₂ intensity 445→400 g/kWh by 2027, −3.6%/yr); Emissions Factors 2025; data centres & transmission embodied + operational guidance. link [refs 2, 95, 102]
6. Univ. of Strathclyde / Univ. of Leeds / Transport for the North — whole-life carbon of roads ~800–2,700 tCO₂eq/km (single-2-lane to dual-3), embodied + 40-yr operational (lighting, maintenance). link [ref 76]
7. Arup / IStructE — Embodied Carbon Priority Actions: C30/37 concrete mix breakdown; concrete ~7.5–8% of anthropogenic CO₂; design efficiency as the most effective lever; reinforcement carbon depends on recycled steel content. [ref 80, 83]
8. Low Carbon Concrete (UNSDSN / Springer / One Click LCA) — SCMs (GGBS, fly ash, calcined clay/LC3), RCA, carbonation curing; 80% of concrete emissions from cement, 40–50% from calcination (not abatable by renewables); clinker reduction often saves capital cost; lifecycle costs down up to 15%. [ref 86]
9. IEEE Spectrum / Schneider Electric / IEA — data-center carbon: operations ~60% / embodied ~40% (incl. devices); core & shell ~6.6% of pre-power Scope 3; whole-life (embodied + operational) accounting; renewable PPAs as primary operational lever. [refs 102, 105]
10. Rio Tinto / ESG Today / Sustainability Directory — construction-fleet decarbonization proof points: renewable diesel (drop-in), battery-electric and green-hydrogen haul trucks; fleet electrification as a Scope 1 lever. [ref 78]
11. McKinsey — origin of the marginal abatement cost curve (MACC), 2007; ~25% of 2030 reductions from negative/near-zero-cost levers. link [ref 72]
12. Enerdata — world CO₂-intensity trends by region (US, EU, China, India, Brazil, Japan, Australia), 2024. link [ref 97]
13. CarbonSig — CaRMa platform: digital twin of any value chain (materials→transport→construct→operate→asset); PCF/CI, CAP certificates, Scope 1–3, scenario modeling, EPD / Buy Clean / CBAM alignment, 3rd-party verification. (Project files.) [ref 12]
14. ISO 14068–1:2023 — Climate change management — Transition to net zero — Part 1: Carbon neutrality: defines carbon neutrality via the hierarchy reduce → enhance removals → offset residual only; life-cycle boundary, all Kyoto GHGs, independent verification; successor to PAS 2060. iso.org/standard/43279 [ref 13]
15. ISO 14068 explainers — Seedling, ECA Business Energy, Sphera, Tunley: hierarchy, residual-only offsetting, durable-removal preference, anti-greenwash framing. Seedling (2026); ECA. [ref 14]
16. IEA — Direct Air Capture: scaled operational DAC cost ~$230–630/tCO₂, depending on energy cost; CDR need ~85 Mt (2030) → ~980 Mt (2050). iea.org/reports/direct-air-capture-2022 [ref 15]
17. ETH Zürich (Sievert, Schmidt & Steffen, 2024) — projected DAC costs ~$230–540/tCO₂ at scale (vs current ~$600–1,000); solid-sorbent ~$374/t, liquid-solvent ~$341/t at 1 Gt/yr. ScienceDaily summary; Carbon Herald. [ref 16]
18. Sylvera (2026) & Puro.earth / CDR.fyi (2025) — removal-instrument price benchmarks: ARR ~$22, REDD+ ~$6, biochar ~$125–180, ERW ~$200+, BECCS ~$389, DAC >$500; durable-CDR order prices ~$320/t (2024). sylvera.com; OTCflow market outlook. [ref 17]

All ten abatement-cost curves and the cross-sector split are indicative syntheses normalized for cross-sector comparability: lever orderings and the cost-negative-band finding are grounded in the cited sources (notably Infrastructure Australia and CLF/RMI), but specific $/tCO₂e values and abatement widths are illustrative and will vary by project, supplier, region and year. Headline figures (embodied ~10% of national emissions; ~15–25% abatement at zero/negative cost; 30–70% cement reductions; the regional grid-intensity values) are sourced as cited. This is a strategic-planning aid, not a substitute for project-level LCA which is precisely the gap the CarbonSig research tool fills. Reference numbering preserves the source IDs used across the CarbonSig brief series.

Originally published at https://nickgogerty.github.io.

Data and methods: the $/tCO₂e levers in this brief are now backed by the CarbonSig infrastructure carbon-cost database with 880+ sourced datapoints from ~120 independent organizations (state DOT bid tabulations, SINAPI, CPWD, MLIT, USGS, SteelBenchmarker, World Bank, EC3, Sylvera), each value URL-cited and confidence-tiered, and validated against an independent held-out benchmark to 7% median cost error, inside the 15% best-in-class target. The CarbonSig MACC engine computes a live curve for all ten infrastructure types above and solves the least-cost path to ISO 14068 carbon-neutral. Explore the live data summary dashboard: https://nickgogerty.github.io/car-netzero-sim/macc-database/