Business Development

WindBorne Systems advances AI weather forecast tech. Published By A U.S.-based weather technology startup says its latest artificial intelligence-driven forecasting model is surpassing traditional government forecasts from established meteorological agencies, underscoring how machine learning is rapidly transforming weather prediction. WindBorne Systems, founded in 2019 by Stanford alumni, today released WeatherMesh-6, an AI model that it claims produces hourly weather predictions and matches the five-day accuracy of conventional models used by major government centers one day ahead of time. WeatherMesh-6 forecasts multiple variables including surface temperature, at a 3 kilometer resolution across Europe and the continental United States, a level of detail equivalent to or finer than many official public weather systems. Unlike traditional forecast models that run physics-based simulations on supercomputers and update output every six hours, WindBorne's system ingests direct sensor data and runs continuous predictions, enabling hourly updates that adapt more dynamically to changing conditions. Startup chips away at traditional forecasting. Government meteorological agencies such as the European Centre for Medium-Range Weather Forecasts (ECMWF) and the National Weather Service (NWS), part of the U.S. National Oceanic and Atmospheric Administration, have long held the gold standard for accuracy. ECMWF models, based on decades of numerical weather prediction expertise, use physics-based equations to simulate atmospheric behavior. The NWS's models and supercomputer systems have historically been essential for public safety and national weather services across the United States. WindBorne's model, by contrast, blends advanced deep learning with a global network of sensors including data from roughly 400 weather balloons launched from 15 sites worldwide that feed direct real-time observations into its transformer-based forecasting engine. According to industry trackers, the startup has raised around $25 million at an $85 million valuation and licenses balloon data to organizations including NOAA and the U.S. military while selling forecasts to commercial traders and investors. What AI models bring to weather prediction. Weather forecasting is one of the most computationally intensive scientific problems, involving vast data streams from satellites, radars, surface stations and atmospheric probes. Traditional approaches solve complex physics equations at large scale, which can take hours on government supercomputers and typically update forecasts at intervals such as six hours or more. AI systems like WeatherMesh-6 use machine learning to approximate atmospheric dynamics and generate rapid updates with lower latency. AI weather models are improving quickly, and academic benchmarks show that deep learning systems can outperform older statistical approaches on many forecast accuracy metrics, although they still face challenges on rare extremes and very long-range forecasts. Research published in arXiv finds that while AI models excel at short-to-medium range predictions, traditional numerical methods still outperform them for unprecedented extreme events and long horizons, underlining why hybrid or ensemble systems remain important for operational use. Commercial and public sector implications. WindBorne's progress arrives amid growing interest in AI weather forecasting across public and private sectors. Startups such as Tomorrow.io are building AI-native satellite constellations and real-time forecasting networks validated by agencies such as NOAA, showing that commercial systems can complement official models. Government agencies themselves are incorporating AI into their forecasting pipelines to accelerate predictions and improve data assimilation, though they stress that AI tools are additive rather than replacements for established physics-based models. The U.S. National Oceanic and Atmospheric Administration launched its own AI-driven forecast suite while continuing traditional forecasting as part of ensemble approaches. Industry observers see these developments as part of a broader trend where AI enhances forecasting speed and resolution for sectors such as agriculture, logistics, emergency planning and energy, areas that depend on timely and precise weather information to manage risk and optimize operations.

Object that hit United flight's windshield may have been weather balloon, company says. (NEW YORK) - A United Airlines flight diverted to Salt Lake City last week after an object struck the plane's windshield at 36,000 feet, causing it to crack and injuring the pilot, according to the airline and officials. Amid the mystery of what could have hit the plane's windshield, on Monday night, WindBorne Systems, a long-duration smart weather balloon company, released a statement saying the object that hit and cracked United flight's windshield may have been a weather balloon from the company. The company said it is working with FAA and the NTSB on the investigation. "We are working closely with the FAA on this matter. We immediately rolled out changes to minimize time spent between 30,000 and 40,000 feet. These changes are already live with immediate effect. Additionally, we are further accelerating our plans to use live flight data to autonomously avoid planes, even if the planes are at a non-standard altitude. We are also actively working on new hardware designs to further reduce impact force magnitude and concentration," WindBorne said in a statement. The windshield is being transported to the National Transportation Safety Board's laboratory as the investigation continues. Data from flight tracking website Flight Radar24 shows the plane was 36,000 feet in the air when an object hit the windshield. The flight then descended to a lower altitude, following standard protocol, before making an emergency landing at Utah's Salt Lake City International Airport. "This is an extraordinary situation in terms of the glass being able to create any damage at all to the people in the cockpit, and what it might have hit at 36,000 feet. That's really the great puzzle," said ABC News aviation analyst John Nance. Aircraft windshields are designed with multiple layers to be able to sustain damage caused by things like a bird strike, weather or even debris, but experts say it's rare for it to be a bird strike that high in the sky. "You're talking about a bird at that altitude. It's very, very rare to say the least, you're talking about maybe a drone, a weather balloon, anything of that nature that has enough mass to be able to cause this kind of shattering," said Nance. United Airlines said the Boeing 737-MAX 8 with 134 passengers landed safely in Utah "to address damage to its multilayered windshield." Officials said the pilot was treated for minor injuries. Heather Ramsey, a college student and a passenger onboard, said she first noticed something was weird about 50 minutes into the flight, even before any announcements, when she overheard one of the flight attendants sharply raising her voice and telling the other to stop the service and get to the back of the cabin. Shortly after, Ramsey said the pilot made an announcement of the flight diverting. "The aircraft has collided with an object and a window in the cockpit has shattered, so we need to make an emergency landing in Salt Lake City," Ramsey told ABC News, recalling the pilot's message. The airline said passengers were accommodated on another aircraft to Los Angeles later that day and United is working with its team to return the plane to service.