IBM, in collaboration with NASA and Oak Ridge National Laboratory, has launched a new open source artificial intelligence (AI) model to optimize weather and climate predictions. This model, available to the scientific community, developers, and businesses, seeks to provide a more flexible and scalable tool to address various weather-related challenges. The platform is designed to exceed the capabilities of current models by providing localized forecasts, detection of severe weather patterns, and refinement of global climate simulations.
The AI model, described in a recent paper in arXiv, has demonstrated exceptional performance. For example, it can accurately reconstruct global surface temperatures using only 5% of the original data. This suggests great potential for its application in data assimilation, a key technique in weather forecasting.
The model has been trained with more than 40 years of observational data from NASA's MERRA-2 project, allowing it to adapt to global and local scales. IBM has released two specialized versions of the Hugging Face platform: one focused on downscaling weather and climate data, capable of generating data with 12 times higher resolution for local predictions, and another dedicated to gravity wave parameterization, which improves the accuracy of numerical models.
Karen M. St. Germain, director of NASA's Earth Science Division, underscored the importance of these developments in providing applied science that addresses rapid changes on the planet. The IBM and NASA models will enable more accurate and useful tools in the field of climate projections.
Applications
IBM has also begun collaborating with the Canadian ministry Environment and Climate Change Canada (ECCC), testing the model in weather forecasting, especially in short-term forecasting of precipitation and downscaling techniques. From IBM Research Europe, Juan Bernabe-Moreno highlighted the model's flexibility, which can be adjusted for different inputs and use cases, helping to understand weather phenomena and extreme weather events.
This initiative is part of a broader collaboration between IBM Research and NASA, which includes the Prithvi geospatial AI model, the largest of its kind in Hugging Face. Prithvi is used in various organizations to study geophysical processes, contributing significantly to research and development in this area.
The weather and climate AI model and the specialized version for gravity wave parameterization, is available through the NASA-IBM page on Hugging Face. The downscaling version is on the IBM Granite page on the same platform, allowing the scientific community and developers to access these advanced tools.
In addition to developments in artificial intelligence, IBM has made significant progress in its business operations. The company recently acquired Kubecost, a company specializing in Kubernetes cost monitoring and optimization, bolstering its investment in FinOps and improving cost management in hybrid cloud environments. This move seeks to maximize the business value derived from cloud investments.
IBM has also announced the purchase of Accelalpha, an Oracle Cloud Applications service provider, with the intent to strengthen its consulting capabilities, particularly in supply chain, logistics, finance, and customer transformation. These acquisitions reflect IBM's strategy to expand its presence in key sectors and optimize its service offerings.
Prospects
On a financial level, IBM announced a quarterly cash dividend of $1.67 per share, reinforcing its commitment to shareholders. In addition, Goldman Sachs maintained its "Conviction Buy" rating for IBM, noting strong demand for infrastructure software and opportunities for growth through mergers and acquisitions.
IBM also introduced its new Telum II processor and IBM Spyre accelerator at the Hot Chips 2024 conference. These technologies are designed to enhance the capabilities of IBM Z mainframe systems, especially in large-scale AI implementation. With these innovations, IBM continues establishing itself as a key player in the technology sector.
Despite staff reductions, a common trend in the industry, IBM is investing in hiring specialized talent in artificial intelligence, seeking to strengthen its capabilities in this field. The company's long-term outlook points to sustained growth, with projected revenue increases between 5% and 9%, driven by growing demand for advanced technology solutions.
