The Infrastructure Carbon Curve — Net-Zero Embodied Carbon is Affordable Today

01 · The answer

Net-zero embodied carbon costs a few percent of the build — and almost nothing per year

Feasible today. Every lever needed — recycled materials, low-carbon cement, right-sized steel, durable removals — is commercially available now. Nothing here waits on a breakthrough.

Feasible
now

The technology and the market both exist. Low-carbon cement is already cheaper than ordinary Portland; durable removals are transacting at scale. This is a procurement decision, not an R&D problem.

0.3–7% of capex

The one-time premium to neutralize residual embodied carbon at $350/t, after taking the cheap material cuts first. Most sectors land at 2–5%.

0.02–0.4% / yr

Amortized over the asset's life at a 4% real rate, the premium is a fraction of one percent of the build, each year — below the noise of any capital budget.

The logic is a pyramid. Because the cheapest 20–45% of embodied carbon is cut at negative or trivial cost (recycled materials that are cheaper, not dearer), and because the residual is neutralized with a durable-removal portfolio for a few percent of capex, and because infrastructure is financed over 25–100 years, the annual cost of a zero-embodied asset rounds to nothing. The rest of this brief proves each step and then asks: what happens to global markets if we do this at scale?

02 · Why it is affordable

A one-time few-percent premium, spread over a lifetime of service

The sticker number — 0.3–7% of capex — sounds like a cost. Amortized, it is not. A road, a bridge, a water plant delivers service for 40 to 100 years; the removal premium, financed like the rest of the build, becomes a fraction of a percent of the budget per year, and fractions of a cent per unit of service delivered.

What it costs per unit of service

0.06¢ per vehicle-km

Road — residual removal cost spread over 40 years of traffic (40k AADT/km)

0.36¢ per m³ of water

Water plant — ~0.2% of a typical water tariff, over a 30-year plant life

<0.1¢ per bridge crossing

Bridge — over a 100-year design life at 20k crossings/day

The amortized premium never exceeds ~0.4% of capex per year, and for long-lived assets (bridges, rail, buildings) it is under 0.2%. Data centers sit highest only because their 15-year life gives the least time to spread the cost — and even there it is ~0.15%/yr.

Method: one-time premium = residual embodied × $350/t ÷ capex (see sector cards). Amortized with a standard capital-recovery factor at a 4% real discount rate over each asset's service life (roads 40 yr, bridges 100, rail 60, ports 50, T&D 40, power 25, water/wastewater 40, data centers 15, buildings 60). Per-unit figures divide the one-time removal cost by lifetime service volume. amortization: moderate — rate & life assumptions stated

03 · How it works

Reduce first, then remove — never buy your way past the cheap cuts

1 · Cut what the materials allow 15–45%

Cost-negative and low-cost levers — reclaimed asphalt, recycled aggregate, SCM/low-carbon cement, EAF steel, structural right-sizing, mass timber — cut 15–45% of embodied carbon before a single tonne of removal is bought. Many are cheaper than the high-carbon default.

2 · Neutralize the residual durable portfolio

What material choices cannot reach is neutralized with a blended long-duration removal portfolio at ~$350/t. This is the ISO 14068 hierarchy: reduce, then remove; offset only the residual. Buying against gross carbon would inflate every figure 1.5–3× and is not neutrality.

Boundary & corrections. The zero-carbon number covers construction/materials embodied carbon (A1–A5), not operational carbon (which decarbonizes with the grid). A "mid-size" bridge is ~600–1,000 tCO₂e (not 2,500 — that is a large multi-span); rail tunnels reach ~25 ktCO₂e/km (not 1–5); data-center embodied is grid-dependent (10% of lifetime on a dirty grid → ~50% on a clean one). All reflected below.

04 · The removal

The $350/t is a blended portfolio of long-duration removals — DAC included

No single removal pathway carries the residual. A defensible net-zero claim buys a portfolio weighted toward permanence: cheap, durable biochar and enhanced rock weathering at the base; BECCS in the middle; direct air capture (DAC) at the top, priced up to $500–600/t. Blended, the portfolio averages ~$350/t today — and the $200–$500 band in every chart is really a portfolio-mix band, not a guess.

Reading the band

$200/t near-term blend

Biochar/ERW-weighted, less DAC. What a cost-minimizing durable portfolio clears at today.

$350/t balanced central

Diversified >1,000-yr portfolio with a DAC slice — above today's transacted average, deliberately conservative.

$500/t DAC-heavy blend

Maximum-permanence, DAC-forward. The prudent ceiling for a >10,000-yr claim.

Why a portfolio, not a single credit? Permanence, delivery risk, and price all differ by pathway. A blend hedges reversal risk (forestry burns; DAC does not), smooths delivery, and averages the price down while keeping a durable, verifiable claim. It is how Frontier and Microsoft actually buy.

Sources: CDR.fyi biochar snapshot 2025 · Oxford Energy CDR review 2025 · WEF cost of carbon removal 2025 · CDR.fyi 2024 Year in Review

05 · All ten sectors

Roads cost most to reach zero; rail almost nothing

Two master views. The first ranks sectors by the cost to neutralize residual embodied carbon at $350/t, as a share of capex. The second shows why the embodied-only boundary flatters some sectors: where the bar is mostly gold, most lifetime carbon is operational and sits outside this number.

06 · Sector by sector

The ten sectors, each with three MECE views

Each carries the same three visualizations — what to cut (marginal abatement cost curve), how far cutting gets you (reduction waterfall), and what neutrality costs (cost-to-zero band) — plus a matched archetype, worked examples, and sources. Every chart is copy-paste extractible (⧉ PNG, ↓ SVG) under CC BY 4.0. Jump to a sector:

07 · At scale

If just 5% of new infrastructure went net-zero, it would remake two markets

The world builds roughly $3.3 trillion of new infrastructure a year, carrying an estimated 2–3 GtCO₂e of embodied carbon. Take 5% of that to net-zero embodied — a deliberately modest wedge — and the downstream demand dwarfs both the green-materials and the carbon-removal markets as they exist today.

The removals market: demand outruns delivery

30–75 Mt/yr

New durable-removal demand from residual embodied carbon (after material cuts)

4–9×

today's entire contracted durable-CDR market (~8 Mt/yr, 2024)

100–230×

today's delivered durable removals (0.32 Mt in 2024) — delivery, not price, is the binding constraint

$6–26 B/yr

residual-removal bill at a $200–350/t blended durable price

The green-materials market: a demand pull, mostly cost-neutral

~75 Mt/yr

Low-carbon cement (LC3-type) demand pulled in to cut the material stage — ~2% of the 4.0 Gt/yr global cement market

LC3 is ~25% cheaper than ordinary Portland and ~40% lower CO₂ — the premium is negative

~10–15 Mt/yr

Green steel (EAF / H₂-DRI) demand for right-sized, low-carbon structures — under 1% of the 1.84 Gt/yr global steel market

Green steel does carry a real premium (~$186/t at $5/kg H₂); the one lever that is not free

Order-of-magnitude

Green-materials tonnages are indicative — carbon cut ÷ savings-per-tonne — and scale with the material mix

materials sizing: low confidence

The takeaway. A 5% wedge is enough to be transformational for removals (a market that delivered 0.32 Mt last year) while barely denting global cement and steel supply — and the cement half of the shift is cost-negative. The scarce resource is not money or materials; it is durable-removal delivery capacity. That is the market signal this brief is built to send.

Method & sources: global infra spend $2.5–4.4 T/yr (McKinsey, GI Hub); new-infra embodied 2–3 Gt/yr (derived from cement+steel CO₂, WEF, Global Carbon Budget 2024; infra/buildings split is the soft assumption); durable CDR 8 Mt contracted / 0.32 Mt delivered (CDR.fyi 2024); cement 4.0 Gt, steel 1.84 Gt (worldsteel, GCCA); LC3 economics (RMI).

08 · Reuse

Every graphic is free to reuse — credit travels with the image

Licensed CC BY 4.0. Each chart's ⧉ button copies a ready-to-paste PNG with CarbonSig attribution baked in; the ↓ SVG button downloads an editable vector. Both carry the licence and source, so credit and provenance travel with the file.

Suggested citation. CarbonSig (2026). The Infrastructure Carbon Curve: net-zero embodied carbon is affordable today. Carbon Finance Lab. carbonsig.com. Licensed CC BY 4.0.

09 · The tool

From chart to procurement line-item

Once Buy Clean and CBAM price embodied carbon, the MACC stops being an environmental chart and becomes a cost curve a project engineer optimizes against — like a bill of materials. CarbonSig computes it per project at four resolutions:

1 · Material / component. Swap a concrete mix or steel grade; get carbon per m³ or tonne, and the $/t abatement it buys.

2 · Asset / project. Build the whole road, bridge or data center; get a live MACC and whole-life carbon per functional unit.

3 · Portfolio / program. Roll assets into a capital program; rank every lever by abatement-per-dollar across the book.

4 · Market / compliance. Issue a verified Carbon Attested Product (CAP) certificate, residual neutralized at a disclosed removal-portfolio price.

10 · Method & sources

Methodology notes & references

Embodied boundaries are cradle-to-gate (A1–A3) unless a card states A1–A5; where only A1–A3 exists, a stated A4–A5 uplift (~5–15%) is applied and flagged.
MACC levers de-duplicate to one lever per material — competing alternatives never sum. The removal block is always the last, most expensive step.
Cost-negative levers carry a documented net saving; low-cost levers sit under ~$50/t and typically under a 1% build-cost premium.
Amortization uses a capital-recovery factor at 4% real over each asset's service life. Figures are indicative syntheses, not project quotes. Confidence: high · moderate · low.

Sources

Extends the live Infrastructure Carbon Curve (431 datapoints, 71 sources) with archetype-matched capex, a reduce-then-remove model, an amortization layer, and a scaled-market analysis. Removal pricing verified against CDR.fyi, Frontier, Oxford Energy, WEF; capex against FHWA, IRENA, NREL, Lazard, Eno, MISO, Turner & Townsend, World Bank; embodied intensities against Carbon Leadership Forum, IStructE, IPCC AR5; scaled-market inputs against McKinsey, GI Hub, worldsteel, GCCA and RMI.