Microsoft has teamed up with NVIDIA (NVDA) to launch a fresh offensive in the AI PC race.
On Wednesday, Microsoft unveiled the Surface Laptop Ultra, a flagship notebook powered by NVIDIA (NVDA) chips, in San Francisco. The company claims its AI performance surpasses Apple's MacBook Pro M5 across multiple metrics, and it is entering Apple's core market directly with a lower starting price.
At the San Francisco launch event, Microsoft executives said the Surface Laptop Ultra runs local AI models twice as fast as the MacBook Pro M5, with video generation six times quicker. The product starts at $2,599, roughly $400 less than the entry-level 16-inch MacBook Pro, and will begin shipping soon.
The moves are intended to cement Windows' dominance in the AI era. Apple's Mac computers have seen strong demand in recent years thanks to their outstanding ability to run AI models locally, and Microsoft's counterattack will heat up competition directly.
Meanwhile, at its concurrent Build developer conference, Microsoft announced the launch of AI agent product Microsoft Scout. Scout is positioned as an "always-on personal agent."
Nvidia Enters the PC Market, Shifting the Competitive Landscape
The Surface Laptop Ultra uses NVIDIA (NVDA)'s new RTX Spark chip for Windows laptops, which handles both general computing and AI computing tasks.
NVIDIA (NVDA) CEO Jensen Huang and Microsoft CEO Satya Nadella appeared on stage together to support the new PC chip. Huang said on site:
We have completely reinvented the personal computer.
NVIDIA (NVDA)'s entry directly challenges the traditional dominance of Intel and AMD in general-purpose PC processors. In addition, PC makers such as Dell and HP that also use the Windows operating system have already launched devices equipped with NVIDIA (NVDA)'s new chips.
The launch event focused on building a business logic around local AI computing: by running computing tasks on local devices, developers can save on cloud service costs while strengthening data security. Unlike the earlier consumer-oriented Copilot Plus PC, the RTX Spark platform is clearly aimed at AI developers and creators.
Nadella said that over time, developers and users will expect AI tools to be driven jointly by local hardware and cloud services. Microsoft also demonstrated at the event Windows security features for monitoring and controlling AI agent behavior.
However, the large-scale adoption of AI PCs may face a timing challenge. Industry research firm Gartner previously estimated that rapid increases in memory costs could mean AI PCs will not account for the majority of shipments until 2028.
AI Agent Lineup Expands, Scout Focuses on Enterprise Collaboration
Beyond the hardware launch, Microsoft announced the AI agent product Microsoft Scout at its concurrent Build developer conference.
Scout is positioned as an "always-on personal agent" and belongs to Microsoft's newly created "Autopilots" agent category, capable of executing tasks autonomously without needing a prompt each time.
Scout is deeply integrated into the Microsoft 365 app ecosystem and can connect to Teams, Outlook, OneDrive and SharePoint to automatically handle scheduling coordination, meeting conflict resolution, email drafting and other work.
The product is built on OpenClaw open-source technology. Each agent instance has an independent Microsoft Entra service account and is brought into a unified permission governance system. Combined with Microsoft Purview data protection policies, it meets enterprise compliance requirements in all respects.
Currently, Scout is in private preview with a limited group of customers, while also being opened to subscribers who opt in to "frontier" feature access through a desktop app. Using the desktop app requires an active GitHub Copilot subscription. Microsoft said its internal sales team has become one of the largest and fastest-growing groups using Scout.
While rapidly advancing the rollout of agent products, Microsoft is also facing security challenges. Agent tools such as Scout face the risk of prompt injection attacks, in which malicious actors may induce AI agents to perform unauthorized actions or leak sensitive information.
Microsoft's response strategy includes: a phased rollout at limited scale, tools for administrators to track all agent actions, and human approval steps for sensitive operations.
This challenge is not unique to Microsoft. As early as the summer of 2024, when Microsoft partnered with Qualcomm to promote AI-optimized PCs, it came under public pressure because a flagship feature involved privacy issues, which weighed on the overall launch pace. How to balance automation capabilities with user trust remains a core question Microsoft must continue to answer on the AI agent track.
Comments