Mitti Labs - building the Ground Truth
When we wrote our first check into Mitti Labs in 2024, this was a carbon company, and the entire thesis came down to one word: trust. A rice carbon credit is worth nothing unless you can prove that a specific farmer, on a specific one-hectare plot, actually stopped flooding the field this season. Everyone else in the category was busy selling the credit but we backed Mitti Labs because they were the only team who focused a lot on ground-truth data - satellite imagery across microwave, SAR and high resolution bands, IoT sensors for soil parameters, time-series data on water usage, plant health and more. It was a lot of hard slog that had to be done outside of the headlines over two years. This team persevered.
So this one is personal. Today Mitti Labs announces a $9.5M Series A led by Aramco Ventures, their first ever investment in India, joined by Godrej Industries Group, Cisco Foundation, Francis Family Fund, and Volta Circle, with Lightspeed doubling down from the seed.
The problem Mitti Labs set out to attack
Nothing about why rice matters has changed since 2024, except we have entered an era of water bankruptcy. India carries 18% of the world's people on 4% of its freshwater, per-capita water availability has halved since 1970, and on current trends demand runs 50% past supply of freshwater by 2030. Roughly 80-90% of India's water goes to farming, and rice alone drinks more than 30% of all irrigation water on the planet. No other single decision on Earth moves as much freshwater as how a paddy is irrigated. The same flooded paddy is also a climate machine. Standing water also starves the soil of oxygen, the anaerobic microbes feast, and rice ends up responsible for 10 to 12% of global methane emissions. So the entire world’s agricultural mission boils down to (1) growing more rice (2) on less water (3) in worse weather (4) without torching the climate. Not for the faint of heart.
Why we backed Mitti Labs in 2024 and what two years unlocked
Two years ago, our thesis called out three things as core: (1) own the dMRV, and therefore the data / ground truth in a way that’s remotely measured & monitored (e.g. satellites, AI); without that there is no long-term moat in climate and it becomes a project-management business (2) accelerate farmer-behaviour shift via a top-down demand-first approach where buyers of rice in US/EU are demanding a change and (3) bring all stakeholders - the FPOs, the financiers, the buyers - on one platform to create transparency and trust.
On (1), what began as the world's best dMRV for rice is now a full GeoAI instrument: time series physical retrievals of soil moisture, field inundation, and residue, generated from a multi-sensor radar stack (C, L, X) that sees structure through monsoon clouds, all fused with deep learning architectures trained on hundreds of thousands of farmers to feed the bio-geochemical engine built by Mitti’s in-house scientists. As a result, Mitti Labs helps farmers optimize Alternate Wetting and Drying (AWD) management practices to save water and decrease GHG emissions. Mitti’s dMRV measures it all remotely how many dry-downs happened, how often, how long, etc. This is revolutionary in the category anywhere in the world. NASA has also given Mitti Labs an $850K grant as a result which is a strong proof point of our technical superiority.
On (2), the demand-first approach has paid off well so far. Mitti Labs now runs the largest AWD rice program in the world: more than 100,000 farmers across six Indian states, growing 3x every year from 8000 farmers in 2024 to 35,000 last year. The top-down buyers such as Ebro Foods (the world's largest rice company) and Syngenta, alongside global technology and financial-services names have all led to a brand and data flywheel for us that we frankly weren’t expecting to kick in so early.
On (3), the platform play, this is where we got surprised. The data record Mitti Labs built to satisfy a carbon auditor turned out to be something almost no one else on Earth has: a verified, season-by-season account of how 100K+ single-hectare/smallholder farms actually operate. We learned its value when buyers who did not want a single carbon credit began asking for the data itself. Food and agriculture companies wanted the field-level truth for their own supply chains. Others asked to license the platform outright. Mitti Labs now runs a second business selling that data and its insights, entirely separate from carbon.
Surprisingly enough, there is an interesting intersection of Geospatial data and AI that is emerging now. Models that can answer questions about the planet with a chat-like interface, or even predict the state of the earth for scenario modeling. In 2024, NASA and IBM launched Prithvi-2, one of the world’s first geospatial LLMs. Soon after, in 2025, Google DeepMind's launched AlphaEarth. Yet, the utility of these models is tethered to the quality of their training sets. In the physical world, ground-truth data remains a scarce commodity—a scarcity that deepens across India and Southeast Asia, reaching its peak in the isolated rice belts where almost nothing has been digitized. Mitti Labs bridges this gap by owning the data generation layer, verifying every season from the dirt up, and fusing it with GeoAI and bio-geochemical stacks that transform raw records into actionable truth.
Why we are doubling down
Once you can measure a hectare, you can predict its yield at sub-field resolution, time its irrigation, flag an input switch, and see a bad season coming. And the moment a hectare is measured, you can underwrite it, insure it, lend against it, and trade it. The interesting future customers are as much the food and agriculture companies, governments routing enormous energy and fertilizer subsidies toward resilience, banks, insurers, and commodity desks as they are the carbon buyers.
Moreover, the same field-level water ledger that underpins a credit could one day underwrite the replenishment requirements of AI hyperscalers as they build data centers in these markets. This kind of optionality only appears if you own the measurement layer. Rice alone is a $300B industry serving 150 million farmers that technology has almost entirely skipped. Within five years, we believe no serious model of Asia's food system, its water, its carbon, or its crop credit, will be trained without Mitti's ground truth somewhere in its loop.
None of this would be possible without the team and the founders. Devdut Dalal spent years as an operator in food and agriculture. Xavi Laguarta came from the demand side of climate, the corporates and carbon markets we needed cracked. They met in the same section at Harvard and spent a summer in India pressure-testing the idea. Nathan Torbick, who spent a career mapping rice from orbit for NASA and JAXA, drove down to Cambridge and says he was skeptical at first but after a few whiteboard sessions, he was convinced. We’ve been so lucky to see them grow from a 3-member team to now over 150 people across the world!
Congrats to our co-investors in this round, especially our friends at Aramco Ventures who are leading the $9.5M series A. With this capital, we plan to go deeper across India's rice belt, and expand into the Philippines and Indonesia, further up the GeoAI stack, and toward 2 million credits issued a year by 2028.
Welcome to our co-investors in this round, especially our friends at Aramco Ventures who are leading the $9.5M series A. With this capital, we plan to go deeper across India's rice belt, and expand into the Philippines and Indonesia, further up the GeoAI stack, and toward 2 million credits issued a year by 2028.
No comments yet.