This year’s scorching summer of heatwaves has yet again underlined the dramatic impact of climate volatility on global crops. And analysis of the latest season has highlighted the rising pressure on the global brewing industry in particular.

At the end of August, UK barley prices were up 37% year on year, while in France they spiked 23%, according to a report from analysts at investment bank Berenberg. The report also cites US Department of Agriculture data, which puts world barley stocks at their lowest levels since 1984.

Meanwhile, although prices and supply of the other core beer ingredient, hops, were broadly stable this year, a drought in the Czech Republic reduced prized Saaz hops production by more than 30%. And with on-trend craft beer typically using 30 times more hops that traditional brews, the shortage has been especially keenly felt by this flourishing sector.

Developing El Niño conditions are likely to contribute to warmer global average temperatures again next year – and lead to a repeat of summer 2018. While, in the longer-term, projections are that climate change and increases in extreme weather events are likely to lead to more dramatic price hikes and supply chain challenges. Indeed, extreme weather could reduce global barley yield by up to 17%, according to research published in the journal Nature Plants in October.

This is naturally a significant concern for growers keen to protect their livelihoods and buyers struggling to secure their supply each season.

But advances in machine learning and statistical science make it possible to create powerful solutions that address fears at both ends of the supply chain. By drawing on the relevant data points from the billions available, AI can help provide early and accurate signals that can enable more informed decisions to be made much earlier each season.

Here’s how it works using a real-life scenario:

Global brewer feels the heat when severe drought affects multi-million dollar hops order

In 2015, high temperatures and low rainfall led to drought conditions in key growing areas for Saaz hops. In August, a fortnight before scheduled delivery, a hop producer informed one of its brewery buyers of a 31% supply loss and a 140% price increase. This led to significant supply issues and cost implications that had to be passed onto their customers. It also affected the future relationship between the producer and buyer.

In this scenario the shortfall was highlighted only two weeks out – leaving little time to make contingency plans.

We have developed technology that enables us to predict yields and potential disruptions to crop growth early and accurately, so we put our science to the test. We asked our machine learning and statistical scientists to run the hops scenario through our platform to see how much earlier we could have surfaced early warnings about the shortfall. We then analysed the difference the early warning information would have made to growers and buyers.

Combining statistical science, computational sustainability, geospatial, and agronomic data, our platform would have predicted this disruption an invaluable six times earlier and three months out from expected crop delivery – in May, rather than August – with 91% accuracy.

This would have given buyers significantly more time to secure additional supply, at a lower price. They would have been informed in time to make changes to their own purchasing plans, promotions, and factory production settings, and recipes could have been changed according to the predicted supply. Difficult decisions could be made earlier and with more confidence, leading to far greater business efficiencies.

Meanwhile for growers, it would have exponentially increased their ability to plan; for crop management, financially and for harvesting. Early yield disruption warnings also enable growers to manage relationships with their customers; buyers or intermediate processors, adding value and giving all parties confidence to continue working together in the future.

In an alternative scenario, where there was a surplus of supply: this type of early signal is equally useful for farmers to optimise their harvest and minimise waste.


With climate volatility growing, livelihoods are increasingly vulnerable to extreme weather, but embracing new technology is a powerful way to manage the risks better. We are already putting this into practice in real time for a number of forward-thinking Fortune 500 companies.

Smart buyers and their growers are starting to capitalise on advances in AI and machine learning, and taking advantage of pioneering software that is empowering them to make better-informed, earlier, and more confident, decisions that drive efficiency and reduce stress. And into the future, long-range predictions and scenario planning tools are enabling them to both protect food (and beer) supplies for generations to come, and help to regenerate our planet.

Photography is by Markus Spiske on Unsplash.


André DuBuisson 
Chief Product Officer

Andre is a product specialist with 16 years’ experience delivering and growing global B2C / B2B products and Marketplaces across multiple sectors.

He’s successfully serviced diverse clients in product management, business analysis, UX and consulting. Recent engagements have included the innovative use of data science & technology in dynamic marketplaces and developing a large-scale customer engagement platform.

He has previously held product leadership positions at Victor Limited and Collinson Group, developing customer loyalty solutions for well-known international brands.

Ruiao Hu
Research Resident

A computer scientist by trade and a mathematician at heart, Ruiao graduated from Imperial College London with an MEng degree in Joint Mathematics and Computer Science.

With industry experience in data engineering gained with world-leading finance firms, Ruiao has joined Cervest for the summer to construct data pipelines for a number of our upcoming projects.

Just as he joined Cervest, Ruiao achieved a first class MEng degree with the Governor’s Prize and Corporate Partnership Programme award for his final year project. After his time at Cervest he will pursue a PhD in stochastic geometric mechanics under the supervision of Prof Darryl Holm at Imperial College London.

James Walsh
Research Resident

James is a Statistical Scientist with a bachelor’s degree in the subject awarded from the Statistics department at the University of Warwick.

James joins Cervest as part of his work with the Warwick Manufacturing Group and The Alan Turing Institute where he is a Research Assistant developing Physically Informed Hybrid models.

He will be spending his time at Cervest applying this expertise to simulating individual crop growth and constructing bulk processing technologies for modelling organic carbon content for soil quality monitoring. He has also worked in the risk team of financial consultants at Albourne Partners building simulations.

Mike Zotov
Research Resident

Mike has recently graduated with a first class honours degree in Mathematics from the University of Warwick.

Having studied Machine Learning as part of his degree, he has joined Cervest for the summer to work on deep generative models for de-noising remote-sensing data.

After his time at Cervest he will continue to pursue his passion for Machine Learning with a Master’s degree at Imperial College London.

Anna Moses
Business Process Lead

Anna has broad experience helping organisations to define and operationalise their processes in order to achieve ambitious business objectives.

Anna has worked across the public, private, and charity sectors with a focus on organisational transformation through business analysis and behavioural science-focused change management, at organisations including the UK House of Commons, the Parliamentary Digital Service and leading US healthcare consultancy The Advisory Board Company.

Inspired by Cervest’s commitment to using technology for good, through both the delivery of game-changing products and an inclusive working environment, Anna is passionate about being part of the solution to the climate crisis.

She has a BA in Political Science and History from Wellesley College, Massachusetts, and also attended the Massachusetts Institute of Technology.

Michael Griffiths
Senior Engineer

Michael has been building software for over a decade as both a developer and analyst. He has worked and consulted in a broad spectrum of domains – including insurance, payments and e-commerce – for both early-stage startups and established companies. He is also a passionate open-source developer, being a co-maintainer of a number of developer tools for the Clojure programming language. Further, he has been a teacher for the ClojureBridge organisation, which provides free programming workshops for underrepresented groups in tech.

A dedicated environmentalist, having worked for the National Laboratory Service arm of the UK Environment Agency before entering the software industry, he joined Cervest in 2019 with a commitment to use his expertise to build flexible and scalable distributed software systems to help our world adapt to the growing climate crisis, while promoting sustainability and science-driven decision making.

Michael holds a B.Sc. in Theoretical Physics from University College London.

Agnes Schim van der Loeff
Research Resident

Agnes recently graduated with a first class honours BA in Arabic and Development Studies from SOAS University of London. Alongside her studies she has also worked as Environment Officer in the SOAS students’ union. Passionate about environmental justice, she plans to pursue a master’s degree in political ecology.

Agnes has joined Cervest to research the ethical implications of AI in relation to climate change and food security, and thus help ensure Cervest’s pioneering technology adheres to the highest ethical standards. With her social science background she is interested in the social, economic and political dimensions of climate change and will therefore also conduct policy-related research to inform Cervest’s engagement with policy makers.