Deep in the Amazon, trees are being cleared to make way for crops and cattle, vast swathes of virgin jungle burned away so the locals can scratch out a meagre living, and feed global supply chains.

And somewhere else, in an office on the other side of the planet, an unwitting UK firm is buying carbon credits tied to those same patches of forest from an unscrupulous, or perhaps simply unknowing, middleman. A box can be ticked, ESG goals met, and the world, on paper at least, is a little closer to being saved. 

Carbon, the sticky, icky currency of climate compliance, is now traded in vast quantities, with over 25 million tonnes of greenhouse gas removal (GGR) credits sold in the past year alone.

Microsoft, for one, admits it can’t hit its ambitious net-zero goals without them. The company has bought tens of millions of tonnes in recent years, and at an average cost of $449 (£353) per tonne in 2024, the price of climate virtue is anything but cheap.  

However, while carbon credits and removals span everything from emissions accounting to ethical land use, deforestation is where the pressure is about to spike, an issue set to drive real-world business disruption over the next twelve months.

And yes, it’s all Brussels’ fault.

Compliant or Complicit

If your company relies on rubber for cables, wood for packaging, or palm oil derivatives in lubricants or adhesives, you’ll soon be required to prove, down to the GPS coordinates, that none of it comes from deforested land.

The EU’s Deforestation Regulation (EUDR), one of the biggest environmental rule changes to date, forces companies trading with Europe to prove, with precise geolocation data, that their products aren’t tied to deforestation.

Starting January 2026, even if your company isn’t personally hauling bags of cacao through Edinburgh airport, it can still be held liable for deforestation buried deep in the supply chain where EUDR-regulated materials like soy, palm oil, or timber are involved.

And if evidence of deforestation is found, the consequences can be severe. Companies could face penalties of up to 4% of annual turnover, restricted access to the EU market, the confiscation of affected goods, and even the revenues they generated, making the new rules a sweeping redefinition of environmental accountability for enterprise firms.

No wonder so many are scrambling.

“A company like Nestle can be fined 4% of its turnover for one coffee bean coming into its supply chain with deforestation attached,” says Professor Ed Mitchard, co-founder and chief scientist for Space Intelligence.

“EUDR has become a very large part of our business, probably half our business since last year.”

The Edinburgh University spinout has become one of the go-to specialists in geospatial data analysis, working with satellite images combined with software, machine learning and a lifetime of scientific expertise to map, monitor, and verify forest landscapes.

The firm, founded in 2017 by Mitchard alongside fellow Edinburgh Uni alumnus Dr Murray Collins, has a mission to help end deforestation within the next five years. Along the way, they’ve also added considerably to Scotland’s growing reputation in geospatial data.

In the past four years, Space Intelligence has grown its headcount from less than ten to more than fifty, raised over $7 million (£5.3m) in fresh funding, and gained contracts with the likes of Apple and the Intercontinental Exchange, which operates major financial marketplaces, including the New York Stock Exchange.     

With potentially millions on the line for multinationals, it’s not hard to see why the Scottish firm has grown so quickly, and is now taking a big bite from the geospatial solutions market – forecast to reach more than $990 billion (£752.5bn) by 2030.

Even as enterprise businesses are pouring more into ESG programmes, carbon reduction plans, and reforestation schemes, most lack the expertise and experience to track progress and deliver the kind of data needed to satisfy regulators.

Deforestation as a Data Problem

For a few firms, keeping track of deforestation requirements can be as easy as firing up Google Earth or delving into NASA Earthdata. As Mitchard points out, satellite and nature data are fairly open source, with the US first making its Landsat data publicly available back in 2009 (thanks, Obama), with the European Space Agency and EU following suit shortly after.

But just because you can, doesn’t mean you should.

“People with some funding from tech companies have made global maps of forest cover or land cover you can use to check an area for deforestation. And that can be quite useful, but I call that kind of global free and open data relatively low accuracy,” says Mitchard.

“There are lots of complications in using satellite data, like artefacts from clouds and from calibration and so on. Inevitably, you make errors.”

These errors range from the obvious, like confusing seasonal leaf fall in Africa with patches of deforestation, to the more complex, such as misreading routine changes in Scottish forest plantations as permanent loss, even though those trees are regularly replanted. 

Fortunately, this is the sweet spot for specialists in remote sensing, like Space Intelligence.

“It’s very tempting if you’re a satellite data company to use a single kind of type of data, maybe optical data from a camera in the sky, to find a single cloud-free scene over your area, then do some machine learning classification on it and make a map,” says Mitchard, “but that will be lower accuracy than if you used every optical data scene throughout the year.”

“We combine human intelligence with high-quality satellite data cubes and AI to build a really good map. Then we test it against independent ecological data and repeat the cycle, continuously improving it until it meets accuracy thresholds, often 90% accuracy for the classes, or 95% accuracy for forest versus non-forest.”

Processing this kind of data is no small feat. NASA’s Landsat 8 alone captures around 700 images a day, or more than 250 terabytes of raw data a year. That’s more than enough to show you the forest, but not what’s happening within it.

A single Landsat pixel, for instance, spans 900 square meters (roughly two tennis courts), making it all too easy to miss selective logging or smallholder clearing.

“We look at a 10 by 10 meter pixel in the study site about 400 times over a year at different wavelengths and with different types of satellites,” explains Mitchard, “then we bring that together into a data cube that allows us to make much more accurate maps, but it means that the data volumes we consider are far larger.”

Recommended reading

• Report: Glasgow Space Cluster Attracts £208M Enterprise Value
• Comment | Scotland’s Role in Greening The Space Sector
• £1.5M For UK Space Projects in Climate and Transport

It’s painstaking work, and too much for a human mind alone, but while LLMs are essential when working with petabytes of layered satellite data, Mitchard is mindful of AI hype creeping into compliance work.

“There’s some great work around large language models that can interrogate geospatial data, but it’s very easy to burn a huge amount of cash and carbon, doing very sophisticated things with those data sets,” he says. 

“I’m quite wary of that. This stuff really matters with carbon credits and compliance, so you can’t have an LLM hallucinating or confidently giving the wrong answer. It needs to be right.”

Saving Forests, Protecting Margins

The problem is, it’s also expensive. Mitchard freely admits that the clients who come through Space Intelligence’s door are already committed to high-end solutions, rather than those looking to greenwash their reputation or to tick a compliance box. Unfortunately, those committed clients are the minority.

“A lot of corporates in the UK have signed up for deforestation commitments,” says Mitchard, “and a lot of them don’t, as far as I can see, do much checking of that.”

He’s not wrong. Only 27% of firms have deforestation commitments covering all key commodities, according to the 2025 Forest 500 report, a drop from previous years, with many still missing pledges for the highest-risk materials. 

Yet under the EU’s new rules, that’s exactly what’s required. Instead, the burden is passed down the chain, onto suppliers, brokers, and producers, most of whom don’t have the tools to verify anything.

Given the costs of getting it wrong, investing in systems that can prove what’s happening on the forest floor is essential for modern enterprise. When one tiny coffee bean can cost you millions in hard-earned cash, saving the trees is now business critical.