The carbon credit market is broken. Here's the fix.

The voluntary carbon market grew from a niche concern of corporate sustainability officers to a roughly $2 billion-a-year industry, and then it cratered. The collapse wasn't a price collapse — though prices did fall — it was an integrity collapse. The institutional buyers stopped trusting the asset they were buying. And once that happened, you couldn't price the asset at all.

I want to walk through what actually happened, why the legacy fix is structurally insufficient, and what an on-chain retirement layer fixes at the protocol level. Then I'll be honest about what it doesn't fix.

What actually happened, 2022–2024

For about fifteen years, the voluntary carbon market ran on a small set of registries — Verra (VCS), Gold Standard, Climate Action Reserve, American Carbon Registry — that issued, tracked, and retired credits. A credit was a database row. The registry kept the row updated. If the row said "retired," the credit was retired.

The 2022 Guardian/Source Material/Die Zeit investigation into Verra's REDD+ rainforest credits was the moment this started to break in public. The headline finding: a substantial portion of the credits issued — by some accounts up to 90% in certain protocols — were probably not generating real, additional emissions reductions. They were paying for forest that wasn't going to be cut anyway.

That's a bombshell on its own. But the structural problem went deeper:

• Additionality theatre. A "carbon credit" is supposed to represent an emissions reduction that wouldn't have happened without the credit revenue. Proving additionality is hard. Most protocols quietly accepted weak proof.
• Permanence assumptions. A forest credit assumes the forest stays standing for at least a hundred years. Many didn't.
• Double counting between markets. Some credits got counted toward a country's NDC under the Paris Agreement and sold to a corporate buyer. Both parties claimed the same tonne.
• Re-sale opacity. A credit could be retired on a registry page, then quietly re-listed under a different identifier. Auditing this required a level of database forensics that most buyers couldn't do.

By 2024, the major institutional buyers had pulled back. Microsoft, Salesforce, Stripe were doing their own bilateral procurement. The mainstream voluntary market was discounted to a fraction of its 2022 prices.

Why the legacy fix doesn't work

The registries' response was mostly procedural. New methodologies. Stricter additionality tests. Buffer pools to handle reversal risk. ICVCM's Core Carbon Principles. The Integrity Council ratings.

All of this is necessary. None of it is sufficient.

The reason it isn't sufficient is that you cannot fix a trust problem with more procedure run by the same trusted intermediary. The buyer has no way to independently verify that the registry's database hasn't been edited. The buyer has no way to verify that a credit hasn't been re-listed elsewhere. The buyer is back to where they started — trusting the registry.

This is the moment a public ledger becomes a non-optional design choice. Not because blockchains are magical. Because they are the only data structure where retirement is a verifiable, tamper-proof public event.

What an on-chain retirement actually fixes

I want to be precise about what blockchain solves here, because there's a lot of vapourware around this space.

It does not solve additionality. If the project never reduced emissions in the real world, putting a token on a blockchain doesn't change that. The off-chain verification still has to be rigorous. Project quality is still project quality.

It does not solve permanence. A forest credit retired on Ethereum can still burn down. The on-chain layer can record the reversal but it can't prevent it.

What it does solve, completely:

1. Double-spending. Once a credit is retired on-chain, the retirement transaction is publicly visible. Anyone can verify the same serial number isn't retired twice.
2. Re-listing fraud. A retired credit cannot be silently re-issued under a different identifier. The chain tracks the lineage.
3. Audit cost. Verifying the integrity of a retirement used to require buyer-side database forensics. Now it requires looking up a transaction hash. Drops the cost from £thousands to seconds.
4. Public proof. The retirement event has a public link. The end consumer can show their employer, their auditor, their auditor's auditor.

That last one is the breakthrough for consumer-grade carbon. Until on-chain retirement, a "your booking offsets 1 tonne of CO₂" claim was unverifiable to the buyer. Today, it ends with a transaction hash they can paste into Etherscan. That's the gap that closed.

Project quality still matters more than anything

This is the part most blockchain-carbon pitches get wrong. They build elaborate retirement infrastructure for credits that shouldn't have been issued in the first place. On-chain garbage is still garbage.

The credits worth retiring are the ones that pass three independent tests:

• Additionality. Would this reduction have happened anyway? If yes, the credit doesn't deserve to exist.
• Verification. Was the reduction measured by a competent third party with skin in the game?
• Permanence. Will the reduction stick? For nature-based credits, this is hard. For renewable energy displacement, this is easier. For direct air capture, this is the easiest.

At IMPT, we lean toward project categories where these three tests are tractable. Peatland restoration in Ireland — measurable, additional, durable on a soil-carbon timescale. Wind and solar displacement in coal-grid regions — direct, measurable, instant. Reforestation with multi-decade verification protocols. Coastal reef and pollinator habitat with co-benefit metrics that protect against monoculture failure modes.

We avoid the problem categories: avoided-deforestation REDD+ credits with weak baselines, vague "improved cookstove" methodologies, biochar credits issued before the science settled. The market will sort these out eventually. We're not waiting.

What changes for buyers

For a corporate sustainability officer in 2026, the buy decision now has three layers:

1. Pick a project type with credible additionality and permanence.
2. Buy from a project with rigorous third-party verification.
3. Take retirement on a public ledger so the audit trail is independent of the seller.

Most legacy market participants nail layer 1 and 2 and fail on layer 3. Most blockchain-native participants nail layer 3 and are weak on layer 1 and 2. The buyer who wins is the one operating across all three.

What this means for consumer carbon

Consumer-grade carbon retirement was always going to be a hard sell while integrity was unverifiable. People won't pay for something they can't audit. Once the audit cost falls to zero — paste-the-hash zero — the question changes from "should I trust this?" to "do I want to do this?"

That's the question we want our customers asking. And the answer, given that we've made the cost zero to them, is reliably yes.

← Back to analysis