Jua, an AI platform to better predict and understand weather patterns picks $2.5M

Jua, Zurich-based, an AI platform to better predict and understand weather patterns have raised $2.5M in pre-seed funding, with Promus Ventures serving as the lead investor (Promus Ventures previously supported the Finnish satellite imaging business ICEYE).

Jua has launched a platform that will enable the meteorological industry to produce customized high-resolution weather models that outperform all existing approaches, including numerical models and machine learning-enhanced numerical models.

Prominent investors Siraj Khaliq (Co-Founder of the Climate Corporation & previous Partner at Atomico), Mehdi Ghissassi (Head of Product at Alphabet-acquired Deepmind), and Swiss company Session. vc, are among those taking part in the round (founded by seasoned entrepreneurs and first-check investors of companies like On Running, Bexio, or Nutmeg).

Pierre Festal, a partner at Promus Ventures, has joined the AI platform board in addition to the investment.

“The team at Jua has constructed one of the most ambitious AI applications that gives accuracy much beyond what the weather business has to a date deemed possible,” continued Pierre Festal, Partner at Promus Ventures. We are honored to assist them with the introduction of this industry-defining technology.

Serial businessmen Andreas Brenner and Marvin Gabler created Jua together in 2022 in Zurich. Jua has centers in Zurich, Berlin, and Cape Town. The first worldwide high-resolution weather forecast model has been launched.

Jua wants to make it possible for researchers at startups, corporations, and governmental organizations to quickly build custom weather models. The new platform features a training framework that enables non-technical users to customize their models with proprietary data and houses one of the most considerable weather geospatial data sets currently accessible.

Jua’s weather model offers traditional alternatives with 25 times the higher spatial resolution and 10 times the higher temporal resolution. Tens of millions of sensors, as opposed to the hundreds of thousands of sensors utilized by current standard models, are used in an end-to-end deep learning technique to achieve this enormous resolution gain. The energy efficiency of the new deep learning-powered model, which employs nearly a thousand times less computer power than any existing numerical weather model, is a bonus.

Currently available only upon request to a small group of customers, Jua’s platform will be made available to a larger group of users in the early months of 2023.

We are aiming to forever alter perceptions of weather forecasting, said Andreas Brenner, co-founder, and CEO of Jua. Our first model already beats all other numerical models in terms of spatial resolution, temporal resolution, and update frequency by several orders of magnitude. We now make it possible for everyone to get weather data that is much better than they have ever seen, from small businesses to major corporations.

“It is vital to understand that we took the risk of entirely reinventing the technical approach to weather forecasting and constructed a new system from the ground up,” continued Marvin Gabler, co-founder, and CTO. In addition to accuracy, our strategy offers wholly novel capabilities to the market. It introduces the first probabilistic short-term projection to the energy sector, which has the potential to greatly increase its profitability. We’ll continue from here. This is our first platform, and we are excited to share all we have been working on in the upcoming months and years.

Image Credit: Jua