Newshot

Microsoft and Nvidia Upgrade Cloud PC Experience: Dual 4K 60fps Streams, Preview Coming in 2027

AI summary

At Ignite 2026 this month, Microsoft announced a partnership with Nvidia to accelerate local AI experiences, moving AI from cloud services that require an internet connection toward running directly on users' own PCs. The collaboration introduces Direct Stream, a low-latency graphics transmission capability for GPU-accelerated Windows 365 Cloud PCs that supports up to two 4K 60fps streams, with public preview expected in early 2027.

Translation(model translation, not the source's own text)

Microsoft and NVIDIA Upgrade Cloud PC Experience: Two 4K 60fps Streams, Preview Launching in 2027

IT Home, November 20 news: At the Ignite 2026 event held this month, Microsoft announced a collaboration with NVIDIA to accelerate the deployment of local AI experiences, moving AI from a cloud service that "requires an internet connection to use" toward an experience that "can run on your own computer."

According to the cooperation vision announced by both parties, Windows PCs equipped with NVIDIA RTX Spark can run local large language model capabilities, allowing users to directly use AI assistants, generate content, or process documents on their computers without fully relying on the network.

This collaboration introduces Direct Stream, a low-latency graphics transmission capability, to GPU-accelerated Windows 365 Cloud PCs, supporting up to two 4K 60fps streams, with public preview expected in early 2027. This feature can transmit GPU-rendered graphics from cloud Windows computers to users' local devices more efficiently and smoothly, with the goal of making remote work, design, development, virtual desktop, and other scenarios feel closer to a "local computer" experience.

Traditional Cloud PCs typically compress the entire screen image and transmit it back to the user's device over the network, which is prone to stuttering, latency, or quality degradation when the network fluctuates. Direct Stream focuses on optimizing this transmission link so that graphics content rendered by cloud GPUs can reach the user's screen with lower latency and higher frame rates.

In terms of task experience, local computers handle everyday AI applications that are more responsive and offer more controllable privacy, while for complex, compute-intensive tasks, Microsoft provides superior cloud AI responsiveness through close integration of NVIDIA's chips and models with Microsoft's Azure cloud and Windows system.

NVIDIA also released the Nemotron 3.5 series of models, divided into different use scenarios: Super is suited for agents that require continuous reasoning and execution of complex tasks; Lightning emphasizes fast response and tool calling; and there are also companion models for retrieval, document parsing, and content safety.

In terms of "edge AI" deployment, Azure Local will support NVIDIA RTX PRO 6000 Blackwell GPUs and expand to edge devices such as Jetson Thor and Jetson Orin, which can be used for machine vision, robotics, and autonomous operations. This means that factory cameras, warehouse robots, and industrial equipment may all be able to complete AI judgment directly on-site in the future.

In terms of enterprise AI, Microsoft Foundry has added support for NVIDIA's open models and customization capabilities. Developers can control the post-training process through the serverless "Programmed Training" service and deploy custom model weights with Foundry Managed Compute.

EY has already used this solution to fine-tune NVIDIA Nemotron open models for classifying enterprise procurement records, achieving a 55% improvement in accuracy on the evaluation dataset and reducing costs to one-fifteenth of proprietary frontier models.

Related Articles

  • Apple releases iOS 27.3 Beta 2 revision: fixes HomeView pairing issue, Siri AI can read full Safari web page content
  • Apple is developing its own AI agent: can control devices, browse the web, and write code, aiming to compete with OpenAI and Anthropic
  • "Largest IPO in history": Anthropic closes up 17% on its first trading day, market value reaches $1.27 trillion
  • Nothing's first AI device design leaks: screenless design, shaped like a voice recorder
  • AI models significantly reduce the cost of digitizing ancient books: recognizing 1,000 pages now costs less than $0.1
  • Online design and publishing tool Canva launches "AI switch" that can disable most AI tools

The translation is machine-generated and individual terms may be off. Nothing is rewritten or cut; the original is authoritative.

Published by IT之家
This site only summarises and aggregates; copyright stays with the original author. Headlines and summaries follow the source — where they differ, the original wins.
Read the original ↗