London-based climate intelligence startup Sylvera has closed another chunky funding round — bagging $57 million in Series B funding led by Balderton Capital, with participation from existing investors Index Ventures, Insight Partners, Salesforce Ventures, Speedinvest, Seedcamp and LocalGlobe. New investors Fidelity Strategic Ventures, Bain & Company and 9Yards Capital are also joining the round.
The new funding follows a $32M Series A back in January 2022 and a $5.8M seed in May 2021 — with Sylvera banking close to $100M ($96M+) in total external investment since being founded back in 2020 to plough into plugging the data gap in carbon credit accounting.
The Series B is being put towards expansion into the US market — where it’s opening a New York office and building up a local team to target its services at US financial services companies and the asset management industry. Funding is also being earmarked for further recruitment to feed its engineering and product teams so it can keep driving transparency in carbon trading markets.
Commenting on joining Sylvera’s B round, Erik Mostenicky, principal at Fidelity International Strategic Ventures highlighted its planned focus on asset management use-cases as a particular area of interest. “We are excited to invest in Sylvera and contribute to the next stage of their growth by exploring asset management use cases,” he said in a statement.
“The company’s trusted and unbiased data solves a critical need for asset managers by helping them to better evaluate Net Zero plans of investee companies across the globe. Sylvera‘s unique data platform also enables the creation of new sustainable investment products and educates investors about the quality of carbon offsets. The institutionalization of carbon markets is necessary to help corporates and investors achieve their Net Zero targets and we believe Sylvera will be a key driver in facilitating this.”
Since Sylvera last raised it said it’s grown its customer base 13x but it’s still not disclosing customer numbers (though it names the likes of JP Morgan and UBS as recently joining its client roster).
The UK startup’s mission is to provide trusted data and ratings on carbon offset projects to support customers to meet Net Zero goals by investing in projects with the best credibility for carbon sequestration. (The basic idea for offsetting being companies pay to offset unavoidable emissions by funding stuff that sucks up an equivalent amount of carbon — meaning projects need to properly account for the carbon take up to ensure credits sold live up to the promise of emissions offset.)
One element of this trust flows from Sylvera not being a marketplace that makes money on carbon credit sales itself. It’s not in the business of selling carbon credits directly. Nor does it take payments from carbon project developers to rate their projects. Rather it’s positioning itself as an independent data layer — pitching rigorous analysis and data science to companies that want to meet their Net Zero targets and are looking for robust information on which offsetting projects they should invest in to help them get there, alongside (ideally) their best decarbonization efforts.
While it’s not the only business that offers verification of carbon offset project quality as a service, Sylvera’s pitch is it’s taken the lead in being methodical and rigorous in how it’s gone about gathering data and building models to benchmark projects that are selling carbon credits. Firstly because it’s devised a set of tests geared towards different types of carbon offsetting projects. So it’s not taking a one-sized fits all approach to assessing projects that can be as diverse as so-called “nature-based solutions” (e.g. forest conservation) to clean cookstoves, renewable energy or more emergent stuff like direct-air-capture (DAC) and other carbon removal technologies. (It makes its assessment frameworks available for download on its website.)
It’s also done this test building work in consultation with the market to further avoid any perception of bias — via a series of review committees that include buyers, project developers, investors and so on — reaching a consensus on a standard way to test that’s then “locked” and can’t be changed without undergoing another market review. Which means it can’t be accused of silently moving the assessment goalposts.
When it comes to capturing carbon sequestration data on the ground it’s using a variety of sensing technologies. This can include scanning forest biomass with lidar or using radar to probe soils, for example, depending on project type. But, again, it emphasizes that its approach aims to be methodical in how it samples data to ensure consistent training data underpins the machine learning models it builds — which can then be applied remotely to satellite data to quantify how much carbon is being locked away per project on an ongoing basis so buyers of associated carbon credits can be confident they’re not paying for, er, so much hot air.
Scanning the woods for the trees
“This is the magic of machine learning,” co-founder Dr. Allister Furey tells TechCrunch. “Once you have this calibration data — it’s literally like we’re scanning trees down to, you know, 10 millimetres — that let’s you reconstitute the branching network of single trees and… hundreds or in fact millions of trees in forests where every stage of degradation and regrowth is present.”
“Nobody’s ever — including all the world’s climate scientists — really spent a lot of time and effort really measuring how much carbon is in the world forests,” he goes on to suggest. “And so we’ve done a lot of that work — collaborating with national governments, collaborating with the World Bank, and publishing on that work. So we are the world leaders in getting that calibration and validation data.”
It’s also had to develop a lot of new technologies to be able to interrogate project claims in a way that scales — which means remotely, such as by using machine learning applied to satellite data to analyze what’s happening on the ground in carbon sinks like forests, soils, agricultural systems or mangroves. But, again, Furey voices confidence in its ability to measure and forecast the performance of carbon projects into the future given the foundation of standardized data sampling underpinning its modelling.
“We were the first ones to go very deep and have our own in house set of tests… with a consultative process with a whole market. So what those tests let you do is ingest all the data about all of the projects in a very standard way. They lets you benchmark the projects in a standard way. And then you can start to test whether the claims are accurate — using some independent data,” he says.
“We have the world’s best team there,” he further asserts of Sylvera’s in-house machine learning smarts — before pointing back to the core of quality data they’re working with. “We don’t just sit in rooms and work on computers. It turns out that to actually make those models work you need to go into those ecosystems and measure them more accurately than they’ve ever been done before. Because it’s basically like anything with AI. It’s never about the algorithms — it’s always about the data.
“Unless you have very, very accurate data, the accuracy of your machine learning will be throttled by the training data set that you’ve got. And it turns out the training data set was complete garbage.”
This focus on accuracy via standardizing data capture and showing the workings around carbon accounting are informed at least in part by Sylvera’s target customers — since it’s selling services to big banks, insurance giants and asset managers who have “expectations of rigour”, as Furey puts it — whereas he says the voluntary carbon market was historically more of a “best effort” type thing.
“Fidelity came into the round as the one of the top three asset managers on the planet,” he notes. “Knowing the transition costs is one thing and then being able to be sure that when you say if you’re going to market an ETF [exchange traded fund] that was a Net Zero ETF product, if you don’t have adequate assurances that it’s Zet Zero the SEC will come and sue you! And rightly so.”
Managing uncertainty, loss and risk
Despite that high expectation bar from clients, uncertainly around carbon offsetting project performance — say if a forest ends up burning down years hence or more trees than expected succumb to disease in a fast heating world, or the fact of a regenerative agriculture project being undertaken in one region leading to increased demand for carbon intensive agricultural processes in another region boosting emissions there — is not something to be afraid of, per Furey.
“There’s this strange obsession with perfection in the climate space that basically stops you doing anything,” he argues, discussing how Sylvera thinks about modelling uncertainty and risk in a changing climate. “We deal with uncertainty all the time — a bond default, or houses sometimes burn down and cars crash or whatever. But that doesn’t mean we don’t have houses or cars or stock markets, right. And so you need mechanisms to account for that loss and get trued up — again with a high level of rigour and integrity.
“So this is totally possible. You can basically observe that the forest is lost and make sure that whoever’s used those credits has to ‘true up’ retrospectively — or the system of issuance does that. So these are completely solvable things and they’re often thrown up as straw men for people that don’t want to spend the money [offsetting their carbon emissions].”
“This idea that normal mechanisms of managing risks somehow [can’t be applied] — because it’s climate, it’s specialist — no, no; this accounting and risk management is fully tractable,” he adds. “Especially once you’ve got the monitoring data and that’s accurate. It’s highly solvable.”
While Sylvera’s faith in robust data sampling and methodical data science is certainly infectious, carbon credits as a tool for addressing climate change do continue to suffer from regular reputation knocks. Such as when projects turn out to have drastically overestimated how much carbon is being sequestered. (An investigation of the Verra carbon standard the Guardian reported on in January found the forest carbon offsets approved by that major project certifier, and used by corporate giants like Disney and Shell, were largely worthless — and could even make global heating worse.)
Enthusiastic uptake of offsetting by high polluting sectors like the air travel industry — or, indeed, oil and gas giants — may also add to the suspicion that too many of these projects still sum to greenwashing (at best).
While poorly conceived projects, such as harmful monoculture tree planting, can demonstrably cause more harm than good — such as by disrupting existing ecosystems or if there’s a high rate of tree failure that risks more carbon being emitted than taken up. So it’s fair to say an off-putting haze of negativity continues to fug up the carbon offsetting space.
Furey argues such credibility and quality issues are all the more reason to rigorously rate carbon offsetting projects — to ensure the highest quality and greatest amount of carbon is indeed being sequestered so that good faith carbon credit buyers inside companies that want to do the right thing and deliver on Net Zero targets will have the confidence to sell the necessary investment (and, most likely, profit write-down) to the board. But he also agrees policymakers are going to need to crank the incentives up.
And while a degree of cynicism vis-a-vis carbon offsetting may still be merited, certainly if it enables companies (or even whole industries) to delay efforts to actually decarbonize their businesses, Furey claims the contrary is true — arguing companies buying carbon credits are decarbonizing faster than those who aren’t. (Albeit Sylvera’s website lists Shell as a customer and the oil and gas giant is not exactly a posterchild on that front — but he declines comment on individual customer cases.)
Engineering carbon removal
He also rightly points out that carbon offsetting can bring other benefits funded through the sale of carbon credits, such as biodiversity (as a result of successful forest conservation) or reduced indoor air pollution (from swapping people to using clean cookstoves). So sucking up as much CO2 as they sell credits for is not the only reason why we should want funds to flow to good quality carbon offsetting projects.
At the same time, some of the projects Sylvera is producing assessment methodologies for are less obviously holistically beneficial — such as biochar for DAC, which is a relatively novel carbon removal technology. So the utility of these sorts of technology-based solutions may be narrower — i.e if their carbon accounting fails to stack up.
On this Furey says the reason it’s assessing a broad church of solutions is that humanity simply doesn’t have the luxury of time vs the scale of carbon sequestering required to reach Net Zero by 2050 to only invest in popular nature-based solutions like tree planting. (If only we’d decarbonized decades ago it could be a far more picturesque picture… )
Simply put, we’ll need to throw everything at drawing truly massive amounts of carbon down — and do that blisteringly fast. (“Net Zero… is the only plan for the world to manage climate change — there’s no alternative plan,” Furey points out. “And the net part of Net Zero basically requires 10 Giga tonnes of carbon sequestration in 25 years — and that’s, like, the most insane amount of stuff to be moved.”)
Albeit, for now, when it comes to tackling (unavoidable) emissions it’s nature-based solutions (not carbon capture moonshots) that are absolutely where the investment action is needed — since they’re, y’know, proven to work. Plus we have — or are getting there, thanks to efforts like Sylvera’s — the data analysis to make the carbon accounting stack up.
“Nature based solutions [are] available now. You can scale them. They’re relatively cost effective. But they can’t get you to 10 Giga tonnes [of CO2 removal] a year in 2050. You kind of run out of the ability to do that. So then you do need to replace it with some engineered solution,” he says, suggesting currently nascent (and/or non-existent) technologies will need to be ready to come on-stream to draw down carbon at scale in around 15-20 years, i.e. when nature-based solutions are likely to be tapped out. Which means there are a bunch of simultaneous investment races that must be won if humanity is to survive climate disaster.
“People are selling those things right now but it’s not needle moving in terms of atmospheric CO2. The forest can do that. And the soils and some ocean stuff can pull the weight for the next 20 years. But you’re gonna run out of the ability to do that and it’s quite terrifying, in the back half of this century, how much CO2 needs to be removed. So you have to have those technologies in place,” he adds. “The norm of transparency and disclosure around all the numbers is yet to be established in those [novel carbon removal technologies], I think it’s fair to say. But it will get there.”
