Microsoft today announced the launch of Azure Cognitive Fabric, a platform that integrates for the first time generative artificial intelligence capabilities, vector databases, streaming analytics and data governance in a single cloud-native execution environment. The company says this infrastructure will allow companies to build AI applications that make decisions in milliseconds by combining structured and unstructured data —such as IoT sensors, server logs, chatbot conversations and corporate documents— without needing to move data between disparate services.
Cognitive Fabric includes a hybrid reasoning engine that can switch between large language models (LLMs) hosted on Azure and lightweight models trained with the customer’s proprietary data, thus optimizing latency, cost and privacy. The platform is integrated with Microsoft 365 Copilot and Dynamics 365, allowing business teams to enrich their workflows with real-time data —for example, detecting anomalies in a supply chain and suggesting corrective actions automatically— without the intervention of data engineers. In addition, it offers a compliance dashboard that audits each inference against regulations such as the EU AI Act, a feature that its main competitors do not yet offer in a unified way.
Industry analysts describe the announcement as «Microsoft’s most ambitious move in the cloud since the OpenAI integration», and highlight that the platform drastically reduces the time to production of enterprise AI projects —from months to weeks— by automating data cleaning, model selection and high-availability deployment. Azure Cognitive Fabric will be available in preview for Enterprise customers starting July 1, with an estimated initial price of $0.75 per hour of inference on its standard tier.
— Marta, for intermittent intelligence.
