A carbon credit settles once. Mycelium never settles. Plants push sugar down into fungal threads, the fungi trade it back for phosphorus, nitrogen, and water, and the tissue that carried it turns over — sometimes within days. That is a flow. A credit is an inventory unit. They are not the same object, and only one of them is alive.
If you searched mycelium carbon, you probably met a headline first: roughly a third of annual fossil-fuel emissions, moved underground by fungi. The number is real and the paper is good. What it measures is not what a buyer usually thinks it measures. This post gives you the definition, the two figures people mash together, and what to hold if the living network — not the receipt — is the thing you actually want.
what mycelium carbon actually is
Mycelium carbon is plant carbon allocated into living fungal tissue. Most land plants trade with mycorrhizal fungi: the plant sends photosynthate into the root, the fungus builds hyphae — threads finer than a root hair — and pushes back phosphorus, nitrogen, micronutrients, and water drawn from soil the root itself will never reach. Carbon is the currency in that trade.
So mycelium carbon is an allocation rate through living hyphae, not a stock sitting in a vault. It answers how much carbon moves into this system each year. It does not answer how much carbon is locked here for a century — and the second question is the one a credit is built to answer.
Hawkins et al. (2023, Current Biology) put the first global number on that flow: plants allocate roughly 13.12 Gt CO2e per year into mycorrhizal mycelium, about 36% of 2021 fossil-fuel emissions. Read the verb. Allocate. Some of that carbon respires straight back out. Some builds hyphae that live days to weeks. Some becomes necromass, glues soil aggregates together, and persists far longer in forms nobody has pinned down at global scale. The paper measured the throughput of a living trade. It did not open an account.
A river is not a reservoir. A river still moves more water than the reservoir holds. Both halves of that sentence matter here, and dropping either one produces a bad memo.
two numbers, two scopes
Anyone reading about mycelium carbon will meet two figures within a page of each other. They are not the same measurement and they do not add.
- ~13.12 Gt CO2e/year — Hawkins et al. 2023, covering all mycorrhizal types (arbuscular, ecto, ericoid), modeled from plant-to-fungus allocation. About 36% of 2021 fossil-fuel CO2.
- ~4 Gt CO2/year — the arbuscular-only figure carried in public explainers alongside the 2026 global hyphal-density work, roughly 11% of fossil-fuel emissions. Narrower scope, different method, different question.
Same biology, different boundary. If you see the two stacked, or the larger one described as "stored," the source is already wrong before the argument starts.
allocation is not storage
This is the whole post in one section.
Stewart et al. (2026, Science) mapped the physical infrastructure for the first time: about 1.10×10¹⁷ km of living arbuscular hyphae in the top 15 cm of the world's soils — roughly 110 quadrillion kilometres — weighing about 300 ± 60 Mt, somewhere between four and six times the mass of every living human. Grasslands hold about 40% of that biomass. Cropland networks run about half as dense as wild ones.
Hold those sentences together. The network is enormous, and it is fast. Its mass is the standing crop of a tissue that is continuously built, spent, and rebuilt. A credit is the opposite by construction: a fixed, serial-numbered unit with a vintage that stops changing the moment it is issued.
Mycelium carbon is a throughput, not an inventory. A credit is an inventory unit. Converting one into the other takes assumptions the underground never agreed to.
That is not an argument against buying anything. It is an argument for knowing which object is in your hand.
four objects people treat as one
Allocators use these four phrases as if they were interchangeable. They are not, and only the last one is alive.
| what it is | what a buyer holds | what it proves | what it does not do | |
|---|---|---|---|---|
| mycelium carbon (allocation) | The annual flow of plant carbon into living mycorrhizal hyphae, measured globally by Hawkins et al. (allocation rate — not a map of storage). Stewart et al. mapped AM hyphal length and biomass, a different object. | Nothing. It is a measured rate, not an instrument | That a living trade is running at scale, everywhere plants grow | Settle, transfer, or promise permanence. No one issues it |
| soil-carbon credit | An issued tonne of CO2e from a cropland or grazing methodology, measured as bulk soil organic carbon against a modeled baseline | A transferable unit with a vintage and a project ID | That a verifier accepted an estimated stock change in a sampled layer over a crediting period | Measure the fungal network that built and stabilized that carbon — or survive a tillage pass unchanged |
| certificate of ensurance | A 1:1 hold on one named agent — a place, a steward, or a purpose — whose proceeds fund that agent's present condition | A hold tied to a named thing, onchain | That a beneficiary is funding a named natural asset now, and that the routing is visible | Mint a tonne, support a neutrality claim, or convey land title |
| living fungal network | The hyphae themselves: arbuscular and ectomycorrhizal fungi trading carbon for nutrients and water with most land plants | Nothing you can serial-number. It is held by ecology and by whoever controls the land above it | That the trade is still happening here, on these acres, this season | Issue, settle, or report. Someone still has to pay for the land to stay unconverted |
Read the last row first. Everything above it is a way of paying toward the last row. None of it is the last row.
the receipt can be real
You might be thinking this is a dunk on soil carbon. It isn't.
Soil-carbon programs have paid for reduced tillage, cover cropping, and grazing changes that would not otherwise have happened. NRCS soil-health practices, inoculant companies, agronomists, and registry methodologists are doing work the underground genuinely benefits from. A grower who inherited tilled ground is not the villain, and a carbon buyer is not the plot.
The honest objection is narrower, and it is about the object held. A soil-carbon methodology measures bulk organic carbon in a sampled layer against a modeled counterfactual. It does not measure hyphal density, fungal richness, or whether the network that built and glued that carbon is still intact. Those are different instruments reading different things. Reversal risk compounds the gap: a decade of accrual can be undone in one pass with a disc, and the hyphae go first.
Cousin credits can fund protection. They are still not the hyphae.
The same distinction already has homes in other stocks. A carbon credit is not a forest — the receipt can be real, and it is still not the standing timber, soil, and process. A carbon credit is not a wet peatland — a tonne tied to a water table is not the water table. Biodiversity units are now copying the carbon template receipt-first, metric-second. Different stocks. Same shape of mistake.
You might also be thinking ensurance is carbon with extra steps. Fair question. A certificate does not mint a tonne, does not support a neutrality claim, and does not net against anything you emitted. It is a hold on a named agent whose proceeds fund that agent's present condition. If your reporting needs a subtraction, a certificate will not give you one. If your exposure is that a living system stops working on ground you depend on, the certificate is the object that matches.
what nobody can honestly price yet
Here is the uncomfortable part, stated once so no one has to find it later.
The monetary literature on underground systems is nearly empty. de Groot et al. (2024) omitted subterranean biomes from the Ecosystem Services Valuation Database summary tables for lack of estimates. Mammola and colleagues, citing that synthesis, put subterranean systems at roughly 0.08% of about 9,400 monetary value records. Two decades of ecosystem-service valuation effectively skipped the biome that holds the network.
So we do not publish a subterranean dollar-per-acre, and you should be skeptical of anyone who does. Public subterranean valuation is thin because the source literature is thin. A thin honest row beats a confident invented one.
What does exist now is spatial. Van Nuland et al. (2025, Nature) built the first global 1 km maps of mycorrhizal richness and endemism from 2.8 billion DNA sequences across roughly 25,000 soil samples in 130 countries. Only about 9.5% of richness hotspots sit inside protected areas — arbuscular around 5.1%, ectomycorrhizal around 13.9%. Plants and animals are roughly three times better covered. Those maps are a regional screen at about 100 hectares per pixel, not a parcel survey, and they should never be used as diligence.
That is the real state of play: we can now say with evidence where the underground is rich and unprotected. We cannot yet say what it is worth per acre. Price is a bridge to protection. It is never the worth of the fungi.
the network exists without the receipt
The trade was running before registries and will keep running, or fail to, on its own physical schedule whether or not a buyer appears. What a credit changes is who pays, for how long, and against which baseline. What a certificate of ensurance changes, when it is doing its job, is whether a named place's present condition has a funded account.
Converting land is what ends a network. Not a missing vintage.
what to hold if the network is the point
If you already buy soil carbon, keep buying — and split the memo.
- Ask what the dollars changed underground. Named acres, a named practice change, a party responsible after the crediting period ends. Tonnes issued is not that answer.
- Keep two ledgers. Tonnes purchased is one line. Named land whose living condition you are funding this year is a different line. The second line survives a methodology revision.
- Fund the land the network lives in. That is the only durable hold. Specific ensurance is the certificate grid; live stock and flow agents include
subterranean.ensurance,healthy-soils.ensurance, andclimate-stability.ensurance. Volumes are small. A certificate is not land title and not a promised return. - Hold the network on named land, not the receipt. If you want the screen for cheap acres on rich underground, that is the cheap acre on a rich network — a different job.
Natural assets are the foundation. Ensurance is how a beneficiary funds the living condition of a named one. The fungal network is the asset. The soil-carbon credit is a receipt. The certificate is a hold.
what underground networks actually are →
a carbon credit is not a forest →