Microsoft has unveiled new tools to run artificial intelligence models directly on personal computers, alongside security technology designed to control what increasingly autonomous AI agents can access.
The move includes a new high-powered Surface Laptop Ultra using Nvidia’s RTX Spark chips, as Microsoft seeks to shift more AI workloads from cloud data centres to Windows desktops and laptops.
Microsoft also introduced Execution Containers, or MXC, which are designed to prevent AI agents from accessing data or carrying out unauthorised tasks on PCs. Anthropic, OpenAI and Nvidia will use the technology, Microsoft executive vice president Pavan Davuluri said.
“We needed to make the desktop the most secure place for agents to execute,” Microsoft CEO Satya Nadella said.
The company said the new tools would allow IT departments to establish rules for AI agents, with Windows enforcing those controls on individual machines.
Microsoft is also bringing models that previously required cloud access to high-powered Windows PCs, including Nvidia’s open-source Nemotron model. A version of Chinese AI firm DeepSeek’s V4 model can run on machines with at least 60 gigabytes of memory, Davuluri said.
Microsoft’s Copilot will continue using cloud computing for more demanding tasks while delegating some work to local models where cost or privacy are priorities.
However, the shift towards local AI faces a major cost hurdle as memory chip prices rise. The Surface Laptop Ultra starts at US$2,599 and reaches US$5,899 for a model with 128 gigabytes of memory.
Nvidia has also raised the price of its DGX Spark AI desktop by about 75 per cent since launch to US$6,950 amid higher memory costs.
“The software wasn’t ready, but the hardware was. Now the software is ready and the hardware is too expensive to actually run it locally,” said Anshel Sag, an analyst at Moor Insights & Strategy.
Reuters
