Microsoft and NVIDIA are collaborating to extend confidential computing to bare-metal and rack-scale AI infrastructure.
The companies said NVIDIA Confidential Computing and attestation let customers verify the computing environment before releasing sensitive models, data or keys. The work supports organizations that need to process proprietary or regulated information.
To improve power efficiency, the companies are also deploying NVIDIA MaxLPS across data center infrastructure.
Foundry Adds Nemotron 3.5 Models and Customization Tools
Through Microsoft Foundry, enterprises gain access to the expanded Nemotron 3.5 lineup. Nemotron 3.5 Super targets long-running agents across coding, research and business workflows, while Lightning serves applications needing fast responses and tool use.
Complementary models include Nemotron Embed and Rerank for information retrieval, Nemotron Parse for structured document extraction, and Nemotron Content Safety for harmful content detection. Nemotron 3.5 Bolt, designed for high-throughput, low-latency reasoning, and Transcribe for speech recognition are listed as coming soon.
Microsoft’s serverless Programmed Training offering gives developers control over post-training workflows while Foundry manages infrastructure. Foundry Managed Compute adds support for deploying custom model weights.
NVIDIA Blog reported that EY used Programmed Training to adapt Nemotron open models for classifying corporate purchasing records, improving accuracy on its evaluation dataset by 55% and lowering cost by 15x compared with proprietary frontier models.

Azure Local is expanding support to NVIDIA Jetson Thor and Jetson Orin edge AI platforms for industrial applications such as machine vision, robotics and autonomous operations. NVIDIA physical AI models, including Cosmos, are also being brought to Foundry Local.
Azure Local will support NVIDIA RTX PRO 6000 Blackwell GPUs from hardware partners for on-premises deployments. Expanded support for Nemotron 3.5 in Foundry Local brings reasoning, coding and tool-using agents to Azure Local environments.
RTX Spark Redefines Windows PCs for Local Agents
NVIDIA unveiled RTX Spark, a superchip delivering 1 petaflop of AI performance with up to 128GB of unified memory. The chip features a Blackwell RTX GPU with 6,144 CUDA cores and fifth-generation Tensor Cores with FP4 precision, connected via NVLink-C2C to a 20-core Grace CPU designed with MediaTek.
ASUS, Dell, HP, Lenovo, Microsoft Surface and MSI will offer RTX Spark-powered slim laptops and compact desktops this fall, with models from Acer and GIGABYTE to follow.
NVIDIA and Microsoft are delivering new Windows security primitives and the NVIDIA OpenShell runtime to run agents securely on primary devices.
OpenShell provides policy capabilities for users to define agent permissions, route queries to local models based on privacy policies, and disguise personal information in cloud queries.
Adobe is rearchitecting Photoshop and Premiere for RTX Spark to deliver 2x faster AI and graphics performance.
The superchip enables rendering 90GB+ 3D scenes, editing 12K 4:2:2 video, generating 4K AI videos, running 120B-parameter LLMs with up to 1 million tokens of context, and playing AAA games at 1440p over 100 frames per second.
Microsoft, founded in 1975 by Bill Gates and Paul Allen to sell BASIC interpreters for the Altair 8800, rose to dominate personal computing with MS-DOS and later Windows. The company launched its first in-house PC line, Surface, in 2012, expanding a hardware portfolio that now includes the devices slated to carry the new RTX Spark superchip.
Under CEO Satya Nadella, appointed in 2014, Microsoft pivoted toward cloud computing with Azure, the platform underpinning the confidential computing infrastructure and Foundry services detailed in this collaboration.

For GPU-accelerated Windows 365 Cloud PCs, Direct Stream capabilities support demanding graphics applications with up to two 4K streams at 60 frames per second and low latency over challenging networks. A limited preview is open for sign-ups, with public preview planned for early 2027.
For scientific research, NVIDIA BioNeMo Agent Toolkit and ALCHEMI Toolkit are expanding into the Microsoft Discovery catalog.
The RFD4-Proteina model, developed with the University of Washington’s Institute for Protein Design, generates candidate protein structures and amino acid sequences for laboratory validation.
Microsoft AI uses NVIDIA Dynamo to support reinforcement learning rollout inference during post-training, coordinating inference across large GPU fleets.