Monday, 05 February, 2024

Jua, the startup developing the world’s first AI foundation model for the natural world, raises $16M

Jua, the startup developing the world’s first AI foundation model for the natural world, has raised $16 million to continue the research, design and customer launch of its platform. The company is developing the first commercially available AI to help humans understand, respond to and predict events in nature.

With a planned launch shortly, Jua’s technology marks a major step forward in the development of AI by bringing together machine learning and physics. Initially it will predict weather with unrivalled precision, before rapidly being able to spot and understand much broader events, such as natural disasters and environmental changes, and inform the most advanced research about the atmosphere. 

With the potential to provide unparalleled market intelligence and improve supply chain management, Jua’s foundation model will initially help those working in energy trading. In the future it will help businesses in sectors such as energy management, insurance, chemicals, agriculture and air travel, and could help governments and civil society organisations tackle challenges from civil planning and food distribution to public service delivery. 

Following the development of foundation models of AI, trained on language to create large language models (LLMs), the Jua platform trains AI on physics and other natural sciences.

Unlike existing models trained to apply to specific industrial verticals and narrow problems, such as weather predictors using a ring-fenced set of data, Jua’s platform is a true foundation and agnostic model for the natural world, able to make sense of data and phenomena it has never seen before to understand events, reveal deep insights and make accurate predictions. It is also the largest AI model in environmentally focused computing – 20 times larger than GraphCast by Google and significantly more powerful than FourCastNet by Nvidia and Pangu Weather by Huawei.

Open AI’s Sam Altman has said that artificial general intelligence can only be achieved if AI understands physics, and Jua’s foundation model will help overcome this frontier. 

The firm was founded by Marvin Gabler, whose experience includes leading research at one of Germany’s earliest weather forecasting firms and developing deep learning technology for Germany’s Ministry with backing from Economic Affairs and Climate Action. He co-leads the business with Andreas Brenner, who cut his teeth selling biomass to major energy organisations before founding vehicle fleet management software firm

Andreas Brenner, Jua Co-Founder and CEO, said: “The first foundation model for the natural world will transform the way humans understand our environment – from the furthest reaches of the earth’s atmosphere to the deepest forests.” 

Marvin Gabler, Jua Co-Founder and Chief Science Officer, added: “As the platform receives more and more data it will rapidly develop from being the most advanced predictor of the weather to helping tackle challenges such as famine and drought. But more than that – it will play a huge role in the development of artificial general intelligence, dovetailing with large language models to allow AI to surpass the final frontier when it comes to cognitive ability: understanding physics.”

Ludwig Ensthaler, General Partner, 468 Capital, said: “Andreas, Marvin and the team are building the world’s first foundation AI for physics and the natural world, which will be capable of providing powerful insights for a wide range of industries dependent on true understanding of nature – from insurance companies and chemical and energy providers, to disaster planning teams, organisations in agriculture, airlines and aid charities.” 

Janna Ensthaler, Founding Partner, Green Generation Management, said: “As we confront an increase in extreme weather events that threaten livelihoods and lead to multi-billion dollar losses, enhancing our weather forecasting capabilities and deepening our understanding of the natural world emerges as a defining challenge. 

“From the outset, we’ve been impressed by Jua’s foundation model and equally by the exceptional team, fusing expertise in AI, meteorology, and business. The Green Generation Fund looks forward to supporting Jua’s mission in delivering predictions of natural events with unmatched speed and accuracy, contributing to the development of a more climate-resilient society.”

At launch, Jua will help power and energy traders accurately model specific energy market problems in the complicated world of weather and energy systems. For example, wind power traders working on production forecasts for specific sites or liquid natural gas traders wanting to model household demands across territories. As the tool develops, companies will be able to combine their own proprietary data with Jua’s foundation model for AI, removing the need for a mix of disjointed data sources such as satellite images and energy forecasts.

The market for accurate energy information has exploded. Oliver Wyman estimates commodity traders’ profits increased 60 per cent in 2022 to $115 billion. Bernstein, a research firm, said by the first half of 2023, in-house traders in major energy companies such as BP, Shell and TotalEnergies had generated a combined $20 billion of profits.