💡 Inspiration
In January 2026, the EU's Carbon Border Adjustment Mechanism (CBAM) entered its definitive phase — a carbon tariff on imported steel, aluminium, and cement. The first real bill arrives 30 September 2027.
We kept coming back to one overlooked group: mid-market importers sitting just above the 50-tonne exemption line. The EU's Omnibus simplification removed roughly 90% of the smallest importers, but the companies just over that threshold are stuck — large enough to owe, too small to staff a compliance team.
These importers face a punishing default: if they can't prove a supplier's emissions intensity, the EU applies a punitive default value. For Chinese steel slab, that default is ≈ 3.17 tCO₂e/t against a real benchmark of ~1.37 — roughly 2.3× the cost, purely for being unable to verify. We wanted to build the honest tool that tells these companies what CBAM will really cost, how much of that is avoidable, and which suppliers to verify first.
🌉 What it does
CarbonBridge turns a compliance burden into an incentive to decarbonise. It answers three questions:
- What will CBAM really cost? We model the true cost trajectory — 2026 only accrues at a 2.5% factor, ramping to 100% by 2034 — instead of a misleading big-today number.
- How much is avoidable? We quantify the avoidable overpayment: the gap between paying punitive defaults and paying verified actuals.
$$\text{cost} = \text{embedded_emissions} \times \text{CBAM_factor} \times \text{certificate_price}$$ $$\text{embedded_emissions} = \text{tonnes_imported} \times \text{emissions_intensity}$$
- Which suppliers to verify first? A private triage signal ranks suppliers by how far their self-reported figure diverges from an independent estimate — always shown as a range with confidence, never as an accusation.
The product spans four screens: an Overview with a live cost curve (punitive defaults vs. verified actuals), a Suppliers triage queue, a Simulator showing the ROI of decarbonising each supplier, and an Evidence tab with auditable provenance and two genuinely live data feeds.
🔁 The moat: the Verified Pool
The defensible idea isn't the calculator — calculators are table stakes. It's a cross-company data network. A supplier verifies their emissions once, by consent, and that proof is reused by every unaffiliated importer who sources from them. No re-auditing, no re-collection.
This creates a flywheel: importers chase verified data → suppliers verify to win business → that proof compounds in the pool → the next importer's decision gets easier. We don't penalise suppliers — we reward them. The pull is well-documented: CDP found suppliers are 52% more likely to cut emissions when buyers ask, with buyer engagement driving 43 Mt of reductions in a single year — more than Sweden emits annually.
🛠️ How we built it
CarbonBridge is a frontend-only, serverless proof-of-concept — it runs entirely at the edge, keyless, hitting real public data straight from the browser. Nothing to run, nothing to leak.
- Stack: TypeScript / JavaScript, HTML, CSS, React + Vite (confirm against your
package.json) - Mapping: Leaflet (toggleable standard/satellite views with facility fly-to)
- Real facility data: a static extract from Climate TRACE (manufacturing v5.7.0, CC BY 4.0) — 8 core facilities with real owners, LEIs, locations, and multi-year emissions intensity
- Two genuinely live feeds, fetched client-side:
- GLEIF LEI API — type a company name and real legal-entity records resolve live
- UK Carbon Intensity API — live GB grid intensity and generation mix
- CBAM policy layer (default values, certificate price, CapEx presets) — illustrative, calibrated to published orders of magnitude, and labelled as such on screen
🎓 What we learned
- CBAM is a system, not a number. The cost ramps over years, defaults escalate with mark-ups (10/20/30%), and the regulation's real teeth are in the punitive defaults — not the headline rate.
- Grid is destiny. Our sharpest finding: Alcoa vs. Albras — same metal, same CBAM-priced intensity (~2.5), but cradle-to-gate footprints of ~12 vs. ~4 tCO₂e/t. The difference is entirely the electricity grid (coal vs. hydro) — emissions CBAM doesn't even price. The UK Carbon Intensity feed became living proof that a cleaner grid is achievable and trackable.
- Honesty is a feature. In a space full of products that overclaim "live" data and satellite "precision" they can't deliver, the discipline of labelling every number real-or-illustrative became our credibility — and our pitch.
🧗 Challenges we faced
- Resolving entities honestly. Linking a trader to the actual producing facility is genuinely hard. Rather than fake automation, we kept it semi-manual with a confidence score — because that's the honest answer.
- Respecting the limits of estimates. Climate TRACE is satellite + ML at regional resolution — an estimate, not a per-facility fingerprint. This shaped the whole product: everything is a range with confidence, and we only ever recommend verification — never accuse.
- Keeping live feeds reliable in the browser. Calling GLEIF and the UK grid API client-side, keyless, meant handling latency and failure gracefully so the demo stays trustworthy.
- Modelling the real cost curve — accruals, the 2027 first-declaration deadline, and the phase-out of free allocation to 100% by 2034 — rather than a scary, inaccurate big-today figure.
🚀 What's next
- A thin serverless backend (managed Postgres + edge functions) for the shared Verified Pool and supplier consent records
- Expanding beyond steel into aluminium and cement at scale
- Onboarding pilot importers above the 50-tonne line and design-partner suppliers to seed the pool and prove the network effect
Built as a frontend-only proof-of-concept. All emissions facility data, owners, and LEIs are real (Climate TRACE / GLEIF); the CBAM policy overlay is illustrative and labelled as such.
Log in or sign up for Devpost to join the conversation.