NVIDIA and Microsoft create unified stack for Agentic AI
The partnership between NVIDIA and Microsoft is taking artificial intelligence to a new level. During Microsoft Build, the two giants showed how developers can build, run, and scale agentic AI on Windows devices, in the Azure cloud, and in on-premises deployments, all through a unified accelerated computing stack. It is a move that promises to transform the way we interact with technology.
The agentic AI moment has arrived, but having powerful models is not enough. Fast hardware, secure environments, and models tuned for long-duration reasoning are also required. And that is exactly what NVIDIA and Microsoft are offering to developers. Jensen Huang, founder and CEO of NVIDIA, and Satya Nadella, chairman and CEO of Microsoft, discussed this expanded partnership during the event, highlighting innovations like NVIDIA RTX Spark and DGX Station for Windows, as well as open models from NVIDIA in Microsoft Foundry.
Reinventing Windows for the agent era
NVIDIA and Microsoft are rethinking Windows PCs for the era of AI agents. With RTX Spark laptops and compact desktops, plus DGX Station for Windows, developers can create, tune, and run agents directly on Windows. The RTX Spark is a milestone, being the first Windows PC built specifically for personal agents, offering 1 petaflop of AI performance and up to 128GB of unified memory. Meanwhile, the DGX Station for Windows is the most powerful AI supercomputer for enterprise applications, capable of running models with up to 1 trillion parameters.
Powering agentic workflows at enterprise scale
Agentic AI operates with a system of models. With NVIDIA, models from Anthropic and OpenAI—along with special Hermes agents—are now available on the Foundry Agent Service, allowing enterprises to deploy agentic systems in Azure with integrated identity and governance. Models like the NVIDIA Nemotron 3 Ultra are available for long-duration agents in coding, research, and enterprise workflows.
Accelerating enterprise data warehouses for the AI era
Data is the fuel of agentic AI, and fast access to it is crucial. NVIDIA accelerated computing is now integrated with the Microsoft Fabric Data Warehouse, delivering SQL execution up to 6 times faster than the CPU-powered baseline. This allows the enterprise data layer to keep pace with AI agents that continuously query and reason over the data.
Advancing physical AI and autonomous systems
A IA física é a próxima fronteira para os agentes. A Microsoft está integrando as habilidades e ferramentas de IA física open-source da NVIDIA com o Azure, oferecendo uma plataforma unificada para simular, treinar e implantar sistemas autônomos. O Cosmos 3, por exemplo, é um modelo aberto que lidera benchmarks em raciocínio visual e geração de ações.
Secure agent development with GitHub Copilot
As agents evolve from coding assistants to autonomous execution, they need real capabilities without real credentials. NVIDIA OpenShell, now integrated with GitHub Copilot, resolves this: each agent operates isolated in its own container, with each external call evaluated against policies before accessing files or networks.
The collaboration between NVIDIA and Microsoft is redefining what is possible in AI computing, bringing a new era of innovation and efficiency for developers and businesses. With this partnership, the future of AI looks more promising than ever.
Comments (0)
Comments are moderated and if they violate our Terms and Conditions of use, the comment will be deleted. Persistence in violation will result in a ban of your account.