Why we invested: Cervest — the climate intelligence platform
2020 was foreshadowing when it comes to our climate crisis. Despite Covid-19-related economic shutdowns, and more commitments to decarbonize our economy, climate-related extreme events accelerated as we experienced a record year of heavy storms, extensive flooding, severe droughts, widespread wildfires and extreme heat waves. Economic losses resulting from those damages add up to hundreds of billions of dollars. To build a resilient economy in the face of our accelerating climate crisis, we need actionable “Climate Intelligence” that helps us understand and anticipate climate volatility and all of its potential impacts.
This is why we’re thrilled to announce our investment in Cervest, a London-based company that built a cutting-edge AI platform to understand, and plan for, the effects of climate risk on physical assets. We led the company’s seed round in July 2019, investing alongside Astanor Ventures, and my co-founder Sofia Hmich joined the board. Today, the company announced a US$30m Series A, led by Draper Esprit. Other investors included Chris Sacca’s Lowercarbon Capital and TIME Ventures (the venture fund of Marc Benioff, the founder and CEO of Salesforce). We re-invested in the company and are happy to share that FPC played a meaningful role in making many of those introductions.
The context for asset needing asset-level risk assessment
Extreme climate events are accelerating. The average global temperature has risen by 1.1°C since pre-industrial times, causing extreme rainfall and floods, extreme temperatures, acute and long-term droughts, and rising sea levels.
The impact of these climate-related events is universal, pervasive and expensive. It affects everyone, everywhere, in every industry. Think about the last time you’ve read about strained (or blackout) power grids, disrupted supply chains, melting roads, wildfires, or flooded buildings. Historic and current economic loss due to those events is quite simply massive. A few numbers suffice to paint the picture. Between 1980 and 2019, weather and climate-related extremes accounted for around 81% of total economic losses caused by natural hazards in the EEA member countries, amounting to EUR 446 billion. As recently as 2018, the WEF estimated that worldwide economic damage from natural disasters added up to US$165 billion. 50% of that total was uninsured.
If we take an event-specific view to make it more concrete, the nine most destructive climate disasters of 2020 caused damage worth at least $5 billion. Some of these disasters hit fast, like Cyclone Amphan, which struck the Bay of Bengal in May and caused losses valued at $13 billion in just a few days; others unfolded over months, like floods in China and India, which had an estimated cost of $32 billion and $10 billion respectively. In the US, the hurricane season and a record-breaking fire season caused more than $60 billion in damages. In Europe, Storm Ciara struck the UK, Ireland and other European countries in February 2020, causing damage worth $2.7 billion. These estimates are usually based on insured losses only, meaning the true financial costs are likely to be higher. But it’s not just economic damage –climate-related events cause tremendous secondary social damage.
Heat stress resulting from global warming is projected to cause productivity losses equal to 80 million full-time jobs in 2030. As heat and humidity increase in India, for example, 160 to 200 million people could live in regions experiencing a heat wave by 2030, with a 5% probability that these regions exceed the survivability threshold for a healthy human being.
Future loss assessments point to the existential threat that climate change poses to our global economy. 200 of the world’s largest companies estimated that climate change would cost them a combined total of nearly US$1 trillion in the case of non-action. Global reinsurer Swiss Re, which developed a leading climate practice, asserted that “climate change poses the biggest long-term threat to the global economy”. If no mitigating action is taken, global GDP could shrink by 18% in the next 30 years.
Why do we need new intelligence
The problem is this: while understanding climate risk and the related pressures on capital is becoming critical to the way enterprises, producers, and governments operate, there is a large gap between climate science data and the individual risks faced by asset owners and those that price risk on assets. In other words, it’s immensely challenging for asset owners and managers to make sense of, and take decisions from, whatever data we do have.
Data scarcity. Many parts of the world suffer from data scarcity, making it difficult to provide useful and timely analysis. Where data exists — and large data sets and valuable methods of scientific measurement do indeed exist in some parts of the world for specific climate signals — complexity and incompleteness make for poor decision-making inputs. Existing data sets do not provide insight at the asset-level and climate data quality is heavily skewed to areas of high population density.
Moreover, not all data is created equal: the spatio-temporal sampling between differing datasets is widely heterogeneous. For example, satellite data may sample a location in relatively high-spatial detail, but have poor temporal coverage, perhaps only returning to the same location once every 16 days as in the case of Landsat 8. Contrast that with weather station data, which may be sampling in-situ at daily, hourly, or even over continuous timescales, but they are effectively only sampling the conditions at a single point in space. Harmonizing these datasets so that the information can contribute to a common pool of data unlocks their deeper value.
Data silos. Regardless of the availability of underlying data, most data sets are siloed. Though comprehensively understanding physical climate risk requires a holistic view, silos make it difficult to understand the interlinked complexity of climate signals and their underlying environmental and biological processes. A single risk signal, for example flooding, or wildfires, is still only one piece of a larger puzzle. The complexity and interconnectedness of the climate system means that risks cluster and co-occur across time and space. The manifestations of risk may be subtle and non-intuitive. For example, shifting climate patterns may cause extreme drought, heat stress, and wildfire risk to manifest in one location, and simultaneously drive excessive rainfall and flood risk in another location. The ultimate impacts of these changes to assets and economic function will vary depending on the types of assets that are exposed to those risks.
The perfect storm of the wrong risk, at the wrong time, over a specific class of sensitive assets can wreak havoc, for example in the case of a recent cold snap that resulted in an estimated $2.4bn of damage to French wine production.
Data engineering. Furthermore, none of this is built for our API-first web 2.0/3.0 world. Data is heterogenous. Very few data sets are built in a way that are easily usable by engineering teams. None of the models scale across use cases.
Risk standards. Finally, no standardized and independent approach has been established to date. Climate risk assessments are forecasted at global or regional scales by institutions and consultants, with one-off models that are company specific but do not scale nor provide collective intelligence to multiple parties.
Meet Cervest: quantifying asset-level climate risk
Cervest built a “Climate Intelligence” platform, capable of automatically detecting and quantifying climate risk at the asset level.
The Cervest platform is built on three layers: (1) assets — with a global and queryable data set of real assets, initially focused on built infrastructure such as factories, commercial and residential buildings, hotels, power plants and ports(2) chronic and acute climate signals — such as coastal and riverine flooding, wildfire, extreme wind and heat stress, and (3) use cases — to help asset owners and managers direct their inquiries and access insights that suits their needs.
Concretely, an asset owner can log onto the platform, build a portfolio of assets from the pre-populated common library, and, within seconds, obtain detailed historical assessments of climate risk exposure across signals back to 1970, as well as anticipated risk for decades to come, out to 2100, under varying future climate scenarios. Accessing climate intelligence in this way radically transforms the current patchworked, project-specific models that consume time (often months), require significant financial resources to build and quickly become outdated. Importantly, the platform is open — anyone can visualize asset-level risk for any physical asset they care about. We’re hoping that making climate intelligence accessible will help anyone, at any level of seniority, in any organization, appreciate their organization’s risk exposure and contribute to building momentum for businesses to proactively adopt strategies that optimize for resiliency.
Experienced and expert team
We met with the Cervest team before we had even closed our fund and were immediately sold on the team. Iggy Bassi, the founder and CEO, is a serial entrepreneur who experienced the impact of climate risk firsthand when he managed a farm in West Africa. He lost a season’s harvest after a drought, and realized that despite having access to precise weather assessments, he had no way of getting reliable and decision-ready information on his exposure to climate risk.
He paired up with Dr. Ben Calderhead, the company’s Chief Scientific Advisor, who was a tenured Assistant Professor at Imperial College London where he taught Bayesian uncertainty quantification and computational statistics. Ben is a global authority on building computationally-efficient Bayesian models, and an early pioneer in earth system modeling. His 2014 paper, co-authored with Microsoft Research, contributed to Cervest’s underlying scientific approach.
Together, they gathered a stellar team, with individuals like Mark Hodgson, who spent 20 years selling enterprise products at companies like Microsoft and Google and was an angel investor in the company before joining full time to lead commercial development; Dr. Maxime Rischard, who holds a PhD in GeoStatistics from Harvard University, as Lead Scientist; Sachin Kapila, who built his career leading biodiversity and climate assessments for companies like Shell, and leads the company’s climate risk function; and Melissa Ayres, who built her career leading global marketing for SaaS companies in US and EU, as Chief Marketing Officer. And Dr. Benjamin Laken, who previously worked to build pioneering ESG analytics platforms, as Head of Innovation.
Cervest also works closely with the Alan Turing Institute and Imperial College for AI research. It has appointed several key advisors including Professor Mark Girolami from the Alan Turing Institute and Cambridge University; David Sobotka, the former global head of fixed income and Bank of America/Merrill Lynch; and Dr. Nadia Schadlow, the former Deputy National Security Advisor in the US.
Scientific excellence from Day 1
From Day 1, Cervest’s team has operated at the cutting edge of AI. In contrast to most contemporary ML “black box” approaches, Cervest’s team proceeded with early conviction that providing asset-level and decision-ready insights would require applying a healthy mix of traditional machine learning and mechanistic statistical analysis. In the data chaos we previously described, with siloed, disconnected data sources, incorporating mechanistic knowledge helps decrease noise by integrating known physical processes and relationships into models, all to reduce uncertainty. This is possible because Cervest’s product is based on deep Earth Science expertise. Climate and Earth scientists are a core part of the Science team. They model multiple physical sciences and bring methodologies across disciplinary boundaries to jointly score risk levels at scale.
This foundation also helps homogenize different data sets and bridge the gap between existing data and required insight. To use many of the existing models in the scientific literature, the company had to implement bespoke versions of the models in a probabilistic way, with a focus on data assimilation. Most existing data sets are defined by arbitrary grids superimposed over the surface of the Earth. Assets might not neatly land on those grids, and layers of grids aren’t neatly superimposed. Cervest employs Bayesian non-parametric modelling to make inferences and determine best case probabilities and correlations between risks (signals) at any specified point, especially when information isn’t defined at that particular point. In other words, they integrate downscaled models for signals like heat stress, storms, wildfires, or floods and determine outputs at the particular asset level, for any point on Earth, with specific targeted estimates of all of those risks and their correlations. Most AI teams aren’t built to do this; Cervest’s team is purpose-built.
A clear value proposition in a differentiated product
The result is a queryable global asset inventory offering an independent asset rating solution, with on-demand and personalized insights and analyses on asset-level risks. The platform provides ongoing monitoring across a portfolio of assets that come from a catalog of millions of pre-populated physical assets, as well as advanced climate risk signals and insights on risk exposure.
Cervest also uses the herculean work the team did to build scalable data pipelines to provide climate intelligence through APIs, with a focus on insurers and banks that might want to integrate this data into their own risk models. From what we know, this is the most advanced product on the market.
Regardless of how the insights are accessed, individuals, corporate or government customers can now view — and act — on their specific climate risk, one day integrating it into all core decisions. Cervest, in other words, moves climate risk from the complex and theoretical to actionable for any user, empowering connected ecosystems to use Climate Intelligence to make preemptive decisions on common assets of value for the first time
Ideal market timing
As climate-related impacts increase with rising extreme climatic shocks to assets, the push for independent and standardized analysis is growing. Cervest recently conducted a study on 500 businesses in the UK that showed that while three-quarters of respondents are concerned about climate-related risks, just 10% consider measuring and disclosing their climate-related risks a priority. This, however, is fast-changing. Climate volatility threatens our global economic, social, and ecological stability, and the business world is now mobilizing to get a better appreciation of climate risk. The Financial Stability Board’s Taskforce on Climate-related Financial Disclosures (TCFD)’s recommendations have pushed board-level engagement at close to one thousand of the world’s largest companies, which are now looking to better assess physical risk and resulting financial exposure. Governments are moving too: the UK became the world’s first country to propose making climate-related financial disclosures mandatory for public and large privately-held businesses. We expect other countries to follow suit. We can see the momentum building — though Cervest fundamentally offers a platform to make better strategic decisions, customer conversations actually revolve around disclosure and resilience. Asset managers and consultancies are trying to baseline their assets, but they don’t have the methodology, technology or people resources to do this at scale.
To conclude, Cervest built an automated platform around a core set of intelligence capabilities to help enterprises and governments plan for a climate-resilient and low carbon economy — generating and delivering science-based insights that are needed by all major actors in the economy, and doing so in a user-friendly way. They are pioneering and defining a new “Climate Intelligence” category. We’re delighted to welcome Draper Esprit, Lowercarbon Capital, TIME Ventures and other new investors to the table.
At Future Positive, we are seeking to support 20 iconic European companies that apply technology to solve a global need. Cervest is one of those. If you’re interested in what we’re doing, please get in touch.
