Building the intelligence foundation for the energy transition

The energy transition is rewriting the rules of infrastructure management. As renewable generation expands and electricity systems grow more distributed and variable, the ability to anticipate weather with precision has moved from operational convenience to strategic necessity. Wind and solar generation are inherently weather-dependent. Grid operators, infrastructure planners, and energy managers need to work with environmental data at a depth and speed that traditional numerical forecasting methods were not designed to support.

BKW, an international energy and infrastructure company rooted in Switzerland, recognized this inflection point early. Guided by a purpose captured in the statement “Wir machen Lebensräume lebenswert” (creating living spaces worth living in), BKW plans, builds, operates, and optimizes infrastructure supporting the energy transition and resilient communities across Europe. To meet this moment, BKW began working with Microsoft Energy & Resources Industry, bringing together enterprise cloud infrastructure, geospatial intelligence platforms, and advanced AI weather models developed by Microsoft Research AI for Science and Microsoft AI Weather.

The challenge

BKW operates at the intersection of energy production, grid management, and infrastructure services. Its portfolio spans energy generation, distribution networks, engineering, and building technology, and its customers range from households to large industrial and public-sector organizations across Europe. That breadth means decisions are complex, data-intensive, and increasingly shaped by environmental conditions that are difficult to forecast with precision.

The energy landscape BKW navigates is in rapid transformation. Growing electricity demand, rising renewable penetration, climate volatility, and the decentralization of infrastructure are compounding in ways that require new analytical capabilities. The challenge BKW faced was not a shortage of data: the signals exist. The challenge was working with complex environmental, geospatial, and operational data at the scale and speed required to support better decisions across disciplines.

Numerical weather prediction models remain the state of the art in operational weather and energy forecasting, providing the scientific foundation for many high-quality forecasts used across the energy sector. At the same time, the field is undergoing a significant transformation. New AI-driven weather models are increasingly reaching comparable or even better accuracy in selected forecasting tasks, while enabling forecasts to be generated faster, iterated more easily, and integrated more flexibly into modern data and analytics platforms. For an organization such as BKW, which aims to accelerate the energy transition and scale its analytical capabilities, this evolution opens the door to more responsive, data-driven forecasting workflows that complement established meteorological approaches.

BKW wanted to explore whether AI-driven foundational models could offer a fundamentally different path: more scalable, more integrated, and more accessible to multidisciplinary teams. The goal was not to run an isolated technology experiment, but to understand whether a trusted, enterprise-grade AI and data foundation could help teams work more effectively with the complexity they face every day.

The solution

The collaboration with Microsoft brought together three capabilities that form a new kind of data and AI foundation for BKW’s energy and infrastructure work: an AI weather foundation model developed by Microsoft Research and Microsoft AI Weather, a geospatial data platform built for enterprise scale, and an Azure-native operational infrastructure connecting it all.

Aurora: AI weather forecasting from Microsoft Research and Microsoft AI Weather

At the center of the collaboration is Aurora, the state-of-the-art AI weather foundation model originally developed by Microsoft Research AI for Science and further advanced by Microsoft AI Weather. Aurora represents a significant departure from conventional numerical forecasting. Rather than solving physical atmospheric equations through expensive computation, it uses a large-scale AI model trained on decades of atmospheric data to generate ensemble weather forecasts with strong accuracy and more efficient compute requirements compared to traditional numerical approaches. For BKW, Aurora provides a foundation model that can support regional weather and solar energy forecasting with precision and at a speed well-suited to modern, cloud-native data workflows. Accessed through Azure AI Models, Aurora’s outputs are stored in Microsoft Planetary Computer Pro, making AI-generated atmospheric forecasts directly available to BKW’s data and engineering teams.

Grid HD: High-resolution regional AI weather model

Grid HD is one of Microsoft AI Weather’s fit-for-purpose AI models, designed as a high-resolution, regional forecasting system that blends observational data and analysis to generate precise, localized forecasts. It complements traditional Numerical Prediction Models by leveraging curated “ground truth” datasets from surface, satellite, and grid observations, and serves as a flexible domain-specific layer that enhances global models such as Aurora with detailed regional accuracy for short- to medium-range forecasting. BKW is using Grid HD to evaluate high-resolution, regional forecast performance, particularly for energy use cases such as solar generation and residual load modeling, where hourly temporal resolution and improved local accuracy are critical for back testing, diagnostics, and operational decision-making.

Microsoft Planetary Computer Pro: Geospatial intelligence at enterprise scale

Microsoft Planetary Computer Pro provides the geospatial data layer for the collaboration. Two specialized data catalogs are central to the work: Microsoft AI Weather Grid HD EU, which delivers high-resolution regional weather forecasting data for the European region, and Microsoft AI Weather GridHD EU Solar, which provides regional solar generation forecasting data. A third catalog stores Aurora’s AI-generated weather outputs, creating a unified, queryable repository for AI-derived atmospheric intelligence. These catalogs give BKW teams access to scientifically grounded, AI-ready geospatial data without requiring the bespoke ingestion pipelines or specialist infrastructure that traditional geospatial workflows demand.

Agentic workflows driven by weather insights

Built on Azure-native services such as Azure Container Registry, Azure Container Apps, Azure Application Insights and other Azure components, a Model Context Protocol (MCP) server empowers BKW to unlock weather insights from agentic workflows. This capability will strengthen operational intelligence, supporting the ability to convert insights into coordinated actions with appropriate human review and oversight. The architecture is designed to complement BKW’s existing systems rather than replace them. Microsoft services support scalable data processing and AI-driven insight generation, while BKW maintains governance, domain expertise, and control over how insights are applied.

Microsoft Research, Microsoft AI Weather and Microsoft Energy Industry Expertise at the table

A defining feature of the collaboration is not just the technology but the people behind it. Microsoft engaged researchers, applied scientists, and industry experts from multiple teams and disciplines across different geographies to work alongside BKW’s domain experts. This cross-disciplinary engagement helped BKW not only stand up an initial capability but develop a clearer understanding of the transformation now underway in weather forecasting: the shift from computationally intensive numerical methods toward AI-driven foundational models that can be embedded in modern data platforms and iterated upon by multidisciplinary teams. For BKW, this kind of close collaboration matters. Having access not just to tools but to the expertise behind them is what makes it possible to move from implementation to genuine understanding of what the technology can do.

The impact

The collaboration is in its exploratory phase, and BKW has been deliberate about framing it as a foundation-building engagement rather than a production deployment. The goal from the outset was not to claim outcomes before they have been earned, but to build a reusable platform and develop the organizational understanding needed to deploy AI capabilities responsibly and at scale.

A stronger digital foundation for complex data

BKW now has access to a data and AI platform that can ingest, process, and make complex environmental and geospatial data usable for its teams, without requiring bespoke infrastructure or specialist tooling. The integration of Aurora and Planetary Computer Pro into Azure-native workflows means that data scientists, domain experts, and technology teams can access AI-ready geospatial and atmospheric data from a shared, governed environment. A strong foundation for scalable work with this kind of data is being established.

Faster iteration between disciplines

The collaboration is designed to support a faster pace of iteration between disciplines. Domain experts, data specialists, and technology teams can now test ideas, evaluate uncertainty, and translate insights into decisions through a shared, structured platform. The shift is from fragmented, ad hoc experimentation toward a more coherent, reusable approach to AI-enabled analysis. The practical day-to-day change is not only about automation: it is about better collaboration across disciplines and a clearer path from prototype to operationally relevant insight.

A clearer understanding of what AI-driven forecasting can do

Perhaps most strategically, BKW aims to develop a clearer view of what AI-driven foundational models can contribute to energy forecasting, and what it takes to build the organizational capability to use them effectively. Working directly with Microsoft teams, BKW’s experts have gained a firsthand understanding of the shift from numerical to AI-driven methods: what it means for data quality, for forecast accuracy, for integration with operational systems, and for the teams who need to interpret and act on the results. This understanding is a prerequisite for responsible deployment at scale.

For customers and the communities BKW serves, the value is indirect but important: better data foundations and AI-supported decision-making are what allow BKW to continue delivering reliable, sustainable, and forward-looking energy and infrastructure solutions. This supports BKW’s broader ambition to invest in renewable and efficient energy infrastructure and contribute to climate and sustainability goals.