Mitti Labs Signs Salesforce to Rice Methane Credits, With a Measurement Claim at Its Core
The release gives the number of credits and not the basis of the tonne. In a market where the unit is the product, that is the one detail a buyer needs.
Mitti Labs has signed a five-year agreement to supply Salesforce with 500,000 rice methane credits by 2030, its second offtake from a major technology company in a matter of weeks after Google committed to 1 million. The Bengaluru deep-tech firm works with smallholder rice farmers to switch from continuously flooded paddies to alternate wetting and drying, a practice that cuts methane emissions by up to 50 per cent and water use by up to 40 per cent without reducing yield. The announcement is about a farming practice. The thing that makes it sellable is a satellite system.
The practice and why it works
Flooded paddies are an anaerobic environment, and the microbes that produce methane thrive in them. Draining a field for a few days at safe points in the growing cycle interrupts that, and the practice costs a farmer almost nothing to adopt beyond a change in habit. Rice accounts for roughly 12 per cent of global methane emissions by Mitti Labs’ own reckoning, more than aviation, and agriculture overall contributes about 40 per cent of the human-caused total. Methane’s atmospheric lifetime is around a decade against carbon dioxide’s centuries, so abatement delivers its benefit quickly. That is the whole commercial argument for the category.
The claim that yields hold is the one that decides adoption. A farmer weighing a change to decades of practice will not accept a risk to the harvest for an atmospheric benefit they cannot see, and Mitti Labs’ argument is that water savings shorten and cheapen the irrigation cycle while the crop comes in unchanged. Whether that holds across soil types and rainfall years is an agronomic question the release does not address. AWD itself is not new. Agricultural research bodies across Asia have promoted it for two decades, and the reason it has not scaled is not agronomic scepticism but the absence of anyone paying for adoption at the far end. A farmer absorbs the effort of changing practice and receives a water saving that may never appear on a bill. A carbon credit is an attempt to put a payment where the effort is.
The measurement problem
Carbon markets have a credibility problem, and agricultural water and soil projects sit near the centre of it. A credit is a claim that a tonne of emissions did not happen, and the claim is only as good as the evidence that the practice took hold across thousands of small plots rather than appearing on a form.
Mitti Labs’ answer is a GeoAI platform built with support from NASA that fuses high-resolution satellite radar with ground truth. The detail is more specific than the release conveys. Remote sensing and geo-tagged photographs monitor every enrolled field, physical tubes in the paddy track water levels, and the programme measures emissions directly. Satellite radar matters because it sees through cloud and works at night, and because it detects surface water at a resolution that can tell one smallholder plot from its neighbour. The output is a digital twin per field, which is what makes a per-tonne claim auditable rather than estimated.
The two ratings, and the one the release left out
The strongest evidence in this story is third-party, and the release understates it. Mitti Labs received an A rating from Sylvera in July 2026. Sylvera’s A is its highest grade, and the company is one of only two agriculture projects to hold it. Sylvera assessed integrity, delivery and reputation and found the highest possible result on all three at the pre-issuance stage.
What the release omits is that Mitti Labs also received an A rating from BeZero in October 2026. Two independent agencies reaching the same conclusion is materially stronger evidence than one, and the company’s own published account names both. A reader of the press release alone would think a single rating existed. That is an odd thing to leave out of an announcement whose central claim is verifiability.
Three things follow from measuring at plot level rather than estimating at programme level. A dry-down that did not happen shows up as surface water where there should be none. A field enrolled on paper but not in practice fails the radar check. And a claim ties to a specific farmer and season rather than to an average, which is what an auditor needs when someone challenges a rice methane credit.
Why two buyers move the category
Rice methane projects are rare relative to forestry and cookstoves, and Mitti Labs’ own account notes that it is one of only two agriculture projects holding Sylvera’s top grade. That scarcity cuts two ways for a buyer. A thin market means less price discovery and fewer comparables, which is one reason neither agreement discloses a price. It also means an early commitment buys influence over a methodology that will be copied, and Google’s agreement already reads that way: it commits to more than 100,000 hectares and 1.5 trillion litres of water savings by 2030, with the company’s chief sustainability officer framing the appeal in terms of near-term atmospheric impact.
The GWP question
One number needs care. The release states that methane traps 85 times more heat than carbon dioxide. That is the 20-year global warming potential, GWP20, not the 100-year figure most corporate inventories use, which is roughly 28 to 30. Both are legitimate, and the 20-year basis is arguably the right one for a gas with a decade-long lifetime, but the release never says which basis it uses.
The omission matters because it determines what a credit means. Mitti Labs’ own announcement of the Google agreement is explicit about the distinction, describing 3 million tons of GWP20 impact as equivalent to 1 million tons of GWP100. On that convention, Google’s 1 million credits represent 1 million tons on a 100-year basis and 3 million on a 20-year basis, and Salesforce’s 500,000 credits would follow the same arithmetic. The release gives the number and not the unit. In a market where the unit is the product, that is the detail a buyer needs and a reporter should ask for.
The capital gap offtakes solve
These agreements get signed before any credit is issued because of timing. Sylvera’s published case study on Mitti Labs notes a 12-month gap between project launch and credit issuance. A developer carrying field costs, satellite processing and staff through that gap needs capital, and pre-issuance offtakes are how it arrives. That is also why the ratings matter commercially rather than decoratively: a pre-issuance assessment is what moves a buyer from interest into diligence.
The scale ambition follows from the contracted volume. Mitti Labs plans 2 million credits a year by 2028 from projects in India and Southeast Asia, against 500,000 for Salesforce across five years and 1 million for Google across four. The pipeline has to run several times the signed book, which means farmer recruitment rather than technology is now the constraint. That timing also explains the order of the announcements. Google in September, Salesforce in October, both before meaningful issuance, both framed as validation of a category rather than of a product. A developer that has closed two of the larger nature-based offtakes in the segment has solved its financing problem for the next several years and moved its constraint to farmland.
The water argument
The water figure deserves separate attention because it is the part a farmer feels. The partnership is projected to save 700 billion litres by 2030, which Mitti Labs describes as roughly eighteen months of Bengaluru’s supply. Across its programmes to date the company reports 500 billion litres, about a year of the city’s supply. Bengaluru’s groundwater carries well-documented strain and rice ranks among the thirstiest crops grown around it, so the local politics of the practice shift are more favourable than the climate argument alone would suggest. For the Google agreement, Mitti Labs projected more than 100,000 hectares covered and 1.5 trillion litres saved by 2030.
What Salesforce gets
For Salesforce the agreement is a supply-side move in a category it has already staked out. The company is a founding member of the Beyond Alliance’s Superpollutant Action Initiative, and its director of climate and energy frames superpollutant reduction as one of the fastest ways to slow near-term warming. Buying verified rice methane credits converts that framing into a measurable tonne rather than a pledge. It also places Salesforce in the same buyer cohort as Google, which is where the reputational risk sits: if the underlying methodology is later challenged, both companies carry it. The reputational arithmetic is not trivial: a company buying rice methane credits is making a claim about near-term warming that rests on a project type with a short track record, and the second rating matters precisely because it spreads that risk across two assessors rather than one.
What to watch
Four markers. The first is price, which neither agreement discloses; without it, the revenue these deals represent cannot be assessed. The second is enrolled area and farmer count, because 500,000 credits imply a land base the release does not state. The third is whether the credits rest on a GWP20 or GWP100 basis and whether buyers disclose which. The fourth is permanence, since a farmer who reverts to flooding next season reverses the abatement, and the release does not describe how long adoption is contractually maintained.