Why Soil Carbon Markets Hinge On A Measurement Few Can Agree On

England's Domesday Book of 1086 tried to put a value on every plow and pasture in the kingdom. The accountants of soil have a stranger job today. On June 24, Groundwork BioAg, an Israeli bio-agriculture company, announced the first verified carbon credits issued under Verra's VM0042 standard on American farmland: 19,568 units, third-party verified by SCS Global Services, with buyers already signed.

The issuance is small. The implication, less so. For roughly a decade, soil carbon has been pitched as the next great commodity of the regenerative economy, the second income stream that turns ordinary cropland into a stacked asset. Pension allocators have been waiting for delivered inventory rather than forward promises. They now have a sample.

What they do not have is agreement on how that inventory was measured.

Two camps split the soil carbon market, and the split is widening. One side, with Groundwork BioAg as its current exemplar, sinks augers into the dirt, ships samples to a lab, and burns them for carbon content. The other side, led by California-based Boomitra, reads soil carbon from orbit using satellite imagery and machine-learning models trained on more than a million ground samples. Both are approved by Verra. Both produce credits. Neither produces the same number for the same acre. Until the market can agree on whose tonne is a tonne, soil carbon will sell at a credibility discount that institutional money cannot price around.

The Promise Of A Stacked Acre

The agronomy is straightforward enough. Mycorrhizal fungi, microscopic partners that colonize plant roots and ferry nutrients in exchange for sugars, have done this work since plants moved onto land roughly 450 million years ago. Industrial agriculture, with its tillage and fertilizers, broke much of the symbiosis. Groundwork BioAg’s Rootella product line restores it through commercial inoculants. The result, the company says, is a root system effectively extended ten to a hundredfold underground, and the fungi function as carbon's main route into the soil.

What that means for an acre is this: better fertilizer efficiency, modestly higher yields, and according to Groundwork BioAg’s own data, between 1.5 and 3.5 tons of CO2 equivalent sequestered per acre each year. Enrollment has grown from roughly 9,000 acres in 2023 to over 700,000 today across the American Midwest, Mid-South, and Canadian Prairies. The company puts the addressable market at 450 million acres of reduced-tillage farmland across the Americas.

CEO Alon Werber, in a statement accompanying the issuance, called the milestone the “holy grail of CDR: scalable, durable, verifiable.” Verra's chief executive Mandy Rambharos, quoted in the same release, said the VM0042 methodology was designed for “rigorous, science-based accounting” of agricultural land management. Whether the methodology delivers that consistently is where the story gets interesting.

Lab Versus Algorithm

VM0042 v2.0, the methodology that governs Groundwork BioAg’s program, allows two quantification approaches. The first leans on biogeochemical models. The second uses what Verra calls “measure and remeasure”: physical sampling at the start of a project and at fixed intervals thereafter, with treated plots compared against untreated controls. Groundwork BioAg chose the second. Dirt to lab, dry combustion, third-party check.

It is the expensive option. It is also, according to its proponents, the only one currently legible to a risk-averse buyer.

“You cannot algorithm your way out of a physical baseline,” Dan Grotsky, Groundwork BioAg’s chief growth officer, said in an email. “The voluntary carbon market suffers from a trust deficit, not a technology deficit. Institutional buyers are rejecting phantom credits and model-derived estimates that fail to materialize in the ground.”

Boomitra would argue the opposite. The company was the first remote-sensing project to gain Verra approval for soil carbon and now monitors about 5 million acres across India, Africa, and Latin America. Its model fuses Sentinel-2 satellite imagery with what it describes as an archive of more than a million georeferenced soil samples. Boomitra's pitch is that remote monitoring can cut the cost of measuring soil carbon by more than 90%, which matters because the cost of measurement is what has kept smallholders and lower-acreage farms out of the market entirely.

The split between the two camps is not academic. A 2024 study in the Journal of Environmental Management by Indigo Ag's MRV team reported that an ML-based pipeline applied across 553,743 hectares of U.S. cropland delivered roughly 398,408 tCO2e of net reductions, but with material spatial and temporal variability in the impact of any single farming practice. The algorithms work. They work better in some places, in some years, than others. A separate 2025 paper using Sentinel-1 and Sentinel-2 data with XGBoost models reported an R-squared of 0.91 for soil carbon prediction across sites in Niigata, Japan, and Agbelouve, Togo. Impressive in the abstract. Less so when a buyer asks whether the missing 0.09 lives on their hectare or someone else's.

Standards bodies have so far accommodated both methods rather than choosing between them. The market fragments as a result. One project produces credits priced as physically verified durable removals. Another produces credits priced as modeled estimates with insurance pools attached for reversal risk. A third blends the two. Buyers comparing tonnages across the field are, in practical terms, comparing apples and Sentinel pixels.

The Permanence Question

There is a deeper problem beneath the measurement one. Even if everyone agrees on how much carbon went into the soil this year, no one fully agrees on how long it stays there.

Groundwork BioAg’s case for permanence rests on Mineral-Associated Organic Matter, or MAOM, which the company describes as durable for centuries to millennia. Grotsky says about 70% of carbon sequestered through Rootella programs is stored as MAOM, based on the company's own field testing. The mechanism, in plain terms: fungi exude compounds that bind chemically to soil minerals, producing a pool that microbes struggle to digest.

The peer-reviewed picture is more mixed. A 2012 paper in Science by Lei Cheng and colleagues found that under elevated atmospheric CO2, arbuscular mycorrhizal fungi could produce net soil carbon losses, challenging the assumption that they uniformly protect organic carbon from decomposition. Subsequent work has found that glomalin, once celebrated as the molecule that locked carbon into mycorrhizal soils, may have been overestimated; proteomic analyses indicate that much of what was extracted as glomalin is not actually of mycorrhizal origin. None of this means Groundwork's claims are wrong. It does mean the science is live, not closed.

For an institutional buyer asking whether mycorrhizal carbon credits should be priced like direct air capture, this is the load-bearing question. Engineered CDR locks carbon away in mineral or geological forms whose permanence is mostly a geology question, not a biology one. Mycorrhizal carbon, even the durable MAOM fraction, sits in fields that someone might one day plow.

What Has To Happen Before The Pension Fund Buys In

Ask anyone in the soil carbon trade what would turn it from a boutique product into an institutional line item, and you get a version of the same answer in three parts. Grotsky offered them speaking to me: consistent delivery histories rather than forward-crediting promises, market-wide convergence on the difference between durable and impermanent carbon, and a single agreed standard of measurement.

He is right on each count. Soil carbon, as it stands, has the shape of a credit market without the discipline of one. Forward sales have outpaced delivered inventory. Competing methodologies sit beside each other inside the same registry. Reversal risk is handled inconsistently. None of these are flaws of the underlying science. They are the unfinished plumbing of a market that grew faster than its accounting rules.

A pension allocator looking at the asset class today, faced with that picture, does what allocators do: discounts heavily and waits. The discount is the price of the disagreement. Groundwork BioAg’s June issuance narrows it because it converts pipeline into proven delivery. Boomitra's expanding footprint narrows it differently, by lowering the cost of measurement enough to bring smaller acres into the market at all. Each is a wager on what credibility will look like once the dust settles.

There is one revealing detail in Grotsky's response, however. Groundwork BioAg, the Mycorrhizal Carbon company that employs third-party lab-sampling, plans to use its physical data as a training set for its own mechanistic models in the future. Which is to say: even the camp betting against algorithmic measurement assumes algorithmic measurement is where the market eventually arrives. The question is who owns the baseline that trains it.

Domesday's commissioners had it easier in 1086. Counting a plow is a single act of measurement, and the answer holds until someone buys another plow. Counting carbon is continuous, and the answer depends on the instrument doing the counting. Until soil carbon's instrument question is settled, the asset will keep selling at a discount that has nothing to do with the dirt itself, and everything to do with the disagreement over how to read it.