Swiss startup Jua has introduced an AI‑driven weather model called EPT‑2 that it says surpasses leading systems from Microsoft and Google. In a new report, the company compares EPT‑2 head‑to‑head with Microsoft’s Aurora, ECMWF’s ENS and IFS HRES models, finding it more accurate on variables such as wind speed and temperature, 25% faster, and using 75% less computing power than Aurora. Jua emphasizes that its approach builds a native physics simulation rather than retrofitting AI onto legacy infrastructure, positioning it as a potentially world‑leading forecaster. Read more →