Chips & Supply · United States

Microsoft to sell $2,599 Surface Laptop Ultra containing Nvidia AI chip

Credited to CNBC · cnbc.com · 2h ago

Microsoft said its new laptop containing an Nvidia AI chip will start at $2,599, as the software maker tries to bring more artificial intelligence capabilities to its consumer devices.

The Surface Laptop Ultra is now available on preorder and will begin shipping on Oct. 16, Pavan Davuluri, Microsoft's executive vice president of Windows and devices, said at a company event in San Francisco on Wednesday. The computer has a magnetic USB-C port, similar to Apple's MagSafe charging connector.

While Microsoft Windows is the most popular operating system for PCs, the company isn't among the top six PC vendors in terms of unit shipments, according to an estimate from researcher Gartner. As the AI boom matures and chip technology advances, tech giants including Microsoft, Apple and Google are aiming to allow users to run more AI workloads from their devices rather than relying on the cloud.

Users can install and run a variety of models on the new laptop without having to worry about racking up high monthly bills. Microsoft first talked about the Surface Laptop Ultra in May, but is just now providing specifics.

"Microsoft might get me back to Windows," Max Weinbach, an analyst at Creative Strategies, wrote on X after the announcement.

Davuluri also said that Meta will introduce a version of its popular Muse AI agent for Windows, capitalizing on the surging popularity of the personal agent, which has jumped to the top of Apple's App store, surpassing ChatGPT.

Microsoft is still trying to solidify its role as a key player in AI. The company offers its own AI models, but they're less popular than leading models from OpenAI and Anthropic and aren't at the forefront in terms of performance, according to one benchmark. Developers are more likely to have used alternatives, according to a recent survey.

This isn't the first time Microsoft has sought to use the AI boom as a way to boost sales of its devices.

In 2024, the company introduced the Copilot+ PC standard, requiring the inclusion of neural processing units that deliver a minimum level of AI performance, along with compliant Surface computers. But Microsoft has since "simplified product naming to lead with the Surface brand, while continuing to deliver premium hardware and hybrid AI experiences," a spokesperson said.

The new device can contain up to 128 gigabytes of unified memory and deliver up to one petaflop of computing power. The computer has as much computing power as Nvidia's DGX-1 data center server from 2016, which cost $250 million, Nvidia CEO Jensen Huang said at Wednesday's event.

A presenter showed how apps on the Surface Laptop Ultra can route some AI coding requests run on the computer's local GPU, displaying a green leaf icon, and others to the cloud. Through Windows 11, developers can confine coding agents to Microsoft Execution Containers that will prevent unwanted system access.

Microsoft is updating its Windows 11 device setup experience to help people get started quickly with open-source OpenClaw AI agent software.

Microsoft has been releasing its own computers since 2012, and CEO Satya Nadella indicated on Tuesday that the company's position in the market is evolving.

"I think today is the day we'll start talking and thinking about the frontier ecosystem taking maybe slightly different contours two years from now, five years from now, 10 years from now," Nadella said at the event. "That classic edge that distributed computing will always remain distributed will be more true."

WATCH: Microsoft unveils Muse for Windows

Got a confidential news tip? We want to hear from you.

Get this delivered to your inbox, and more info about our products and services.

Chips & SupplyUnited States

Can AI Be Trusted? Synopsys on Agentic AI and Autonomous Engineering

Credited to EE Times · eetimes.com52m ago

Useful

How can engineers trust AI when chip design and simulation demand precision, repeatability, and verifiable results? In this AI Infra Summit interview, Synopsys explains why the LLM is not the source of engineering truth. Instead, agentic AI serves as an intelligent orchestration layer over trusted, deterministic EDA and simulation engines.

The conversation explores how context-aware, tool-aware agents can plan workflows, invoke the right engineering tools, validate results against design requirements, and adapt when results deviate from constraints. It also examines the importance of token efficiency, including giving models the right context at the right time rather than overwhelming them with unnecessary engineering data.

The interview also looks at what differentiates Synopsys in a crowded agentic AI landscape: domain-specific, long-horizon AgentEngineer solutions and the open, secure Synopsys Autopilot Platform spanning silicon-to-systems workflows. The discussion closes with the practical impact for engineers, from reducing repetitive setup and debug to enabling deeper design exploration, faster productivity, and more connected autonomous engineering workflows.

Watch to learn how Synopsys is advancing engineering-grade AI that is precise, efficient, explainable, and grounded in trusted tools.

Synopsys AgentEngineer Solutions and Autopilot Platform

Synopsys Autonomous Engineering Announcement

Introducing Long-Horizon AgentEngineer Solutions

Original · EE Times

Front Method

Chips & Supply · United States

Why AI’s Limit Is Power, Not Chips: Gopi Sirineni at AI Infra Summit 2026

Credited to EE Times · eetimes.com · 1h ago

Photograph · EE Times

Power, not chips, has become the limiting factor in the artificial intelligence buildout, according to Axiado founder, President, and CEO Gopi Sirineni.

Speaking with EE Times reporter Sally Ward-Foxton at the AI Infra Summit 2026, Sirineni discussed his presentation, “More Tokens Per Dollar: AI-Driven Platform Efficiency Controllers for High-Density AI Racks.” He explained that the solution to data center power constraints lives in the platform management layer rather than better facilities or better silicon.

Axiado addresses this through AI agents running on its Trusted Control Unit (TCU) silicon that sense, decide and act in milliseconds with no human in the loop. Sirineni detailed what Axiado’s dynamic voltage and frequency scaling (DVFS) and dynamic thermal management agents can accomplish that a modern server cannot do for itself, noting that efficiency gains range from 10% out of the box to 30% with tuning.

Sirineni argued that effectively managing and securing modern AI infrastructure requires moving away from decades-old legacy management paradigms toward truly intelligent, autonomous infrastructure management to drive higher efficiency, better security and greater reliability.