On August 8, 2026, ClimateAi founder Himanshu Gupta announced on LinkedIn that the company was winding down and returning capital to investors.
The company had raised $38 million across seed, Series A and a $22 million Series B in April 2023 led by Four Rivers Group, with Radical Ventures, Neotribe, Yaletown Partners and Academy Investor Network on the cap table. Customers included Dole, Driscoll’s, Suntory and Nuveen Natural Capital, across 40-plus crops in more than 60 countries. Annual recurring revenue was in the region of $9 million.
I have written before about adaptation-linked risk analytics becoming an investable category. So, a shutdown inside that category is worth examining. While the company has not given concrete explanations beyond “geopolitical and climate headwinds”, I am going to look at plausible factors to make an inference on what might have caused troubles. A study of this wind down provides diligence questions in this fast evolving VC investment category.
The product logic was sound
ClimateLens, their marquee product, forecasted climate risk from one week to twenty years out, at 1km resolution. The company focused on an ICP comprising of players in food and agriculture sector, as they were directly impacted by extreme events and changing cultivation bands for different crops. The pitch was that the long-range forecasts could help identify shifts in growing locations, inform changes in crop management, and even retime harvests to manage inventory and avoid crop losses.
This was not a solution looking for a problem. So, what might have caused trouble?
Factor #1: Was the ICP ready?
In a December 2024 interview with AgFunderNews, COO Will Kletter named the adoption barriers. One, there was no existing budget line for what ClimateAi sold. Two, the typical ICP comprising of procurement and farmer relations teams usually planned on a two-week weather horizon and had to be persuaded to act on a six-month one with a wide, acknowledged error range.
From my experiences in the food and agriculture sector, I have seen that this ICP is one entrenched in traditional practices. For example, sourcing decisions in food and agriculture still run on supplier relationships and accumulated judgement. They are not used to probabilistic forecasts as a decision-making tool, and ClimateAi might have had to work through a behaviour change as well as selling a software product. So, while exposure to climate risk created a need for the product, it did not automatically create a readiness to buy.
A peer in this sector works slightly differently. Jupiter Intelligence also sells climate risk analytics based on forecasts, but has chosen a different ICP: banks, insurers, reinsurers and infrastructure owners. Here, the buyer is a team already trained to price risk through models.
Factor #2: Who signed off on the contract?
A product on climate risk analytics was also saddled by the label of a sustainability offering. In contrast, Jupiter Intelligence and the likes sell the decision makers on the financial side; transforming the offering as a core capital allocation decision factor. What would have truly moved the needle for ClimateAi was if their product was seen within a subset of the procurement and inventory management team, and not purely a sustainability desk’s line item. In January 2026, thirteen months after Kletter first described this perception problem, he said procurement and sustainability functions were only just beginning to merge inside customer organizations.
But the company did not have time on its side. Through 2025 and into 2026, corporate sustainability budgets and headcount contracted under political pressure in the United States, ClimateAi’s core market. A product read as a sustainability tool sits in the budget line that would have been among the first to get cut, especially when this product forecasts to many months out and the prevailing method of working would have been enough to manage immediate needs.
Factor #3: What is the real moat?
ClimateAi applied AI and patented models to climate and weather data points from multiple sources. The differentiation lived in the model layer. The inputs were public datasets, licensed datasets, and in some cases client-owned weather stations.
Compare that to Tomorrow.io, another peer. A much larger, capitalized firm, it also has the benefit of 13 satellites in orbit with 60-minute global revisit, NOAA validation of its observations through the Commercial Weather Data Pilot, and it raised a $210 million Series F this year to deploy a second constellation. While its primary ICP is aviation, defence and logistics, it is now starting to offer climate resilience-focused data analytics because it just needs to package in-house data differently to a new customer profile.
This is the same structural question I raised about Mitti Labs, where the moat rests on a farmer network producing multi-season baseline data a competitor cannot quickly replicate. If your edge sits entirely in the model layer, over inputs your competitors can also buy, then every capability gain is replicable by anyone with comparable talent and the same data. When COO Kletter was asked about the company’s profitability in a January 2026 interview, he said that the focus of the company is on staying competitive in AI and not immediate profitability. This, in my opinion, is a challenge one cannot overcome without a few conditions.
If you cannot own the data, then you need the product to be embedded deeply enough into a customer’s workflow to make switching costs prohibitive, keeping net revenue retention (NRR) high. Jupiter Intelligence has done this, so has Watershed in enterprise carbon accounting.
Another route is capitalization. If you are funded well enough, you buy the capability and the customer base instead of building them, which is what Asuene has been doing through a run of acquisitions in the climate data stack.
ClimateAi had neither route fully open. Its workflow embedding push came in January 2026 (as stated), which looks late, and at $38 million raised with roughly $9 million in ARR it was not capitalized to pursue M&As.
What the record actually showed
In December 2024, Gupta said he expected profitability by Q3 or Q4 of 2025, and a Series C by the end of that year, after proving value to around 200 partners. Neither milestone was met. In January 2026 the profitability target had been reframed as a strategic choice to invest in AI capability, and no Series C was mentioned. And now, seven months later, the company has wound down.
The diligence questions for investors in this category
Private capital has decidedly embraced adaptation risk analytics as a category; ClimateAi had raised through Series B on its strength. Going forward, two questions follow for anyone underwriting in this space.
First, has the target ICP, for whom a logical product-market fit exists, shown adoption readiness? Extending this question, an investor must test whether the buyer treats the product as a core decision input or a useful extra, and which budget line it actually sits in. ClimateAi’s own team identified its positioning problem in December 2024 and was still describing it as unresolved thirteen months later. A gap left open that long does not need a large shock to become terminal.
Second, does the company own a moat which controls the data input layer? If not, how is it working around this to avoid being commoditized in the era of AI?
Technology has made this category investable by creating a product and finding a payer, which was the prevailing argument against investing in adaptation. But technology is also asking the category to evolve extremely fast. The ones that don’t will find it hard to survive.
Leave a Reply