Oct 21, 2026 · 2h 48m
RTX Spark’s powerful hardware meets unfinished software
WW 1006: Are You Lying to Me Now, Paul? - OmniBook Ultra 16, ProArt P16, & Lenovo Pro 9n Impressions
The episode tests whether expensive local-AI hardware can deliver a better experience than cheaper, more mature cloud tools.
- 1RTX Spark laptops offer strong gaming and AI potential but require difficult configuration and tolerate little software immaturity.
- 2Cloud-based coding tools currently make Windows application development easier and cheaper than assembling a high-end local-AI setup.
- 3Microsoft’s Windows, OneDrive, agent-security, and Xbox plans all point toward ambitious AI integration constrained by trust and execution.
Don't miss
Paul reports that his extended RTX Spark local-AI experiment failed to produce a usable application despite the system’s impressive hardware.
The brief
Micah Sargent fills in for Leo Laporte as Paul Thurrott and Richard Campbell examine Windows 11’s cleanup effort, Insider builds, Edge privacy features, and Microsoft’s shifting AI priorities.
Paul’s RTX Spark laptops deliver impressive hardware, gaming capability, and local-model potential, but immature software, sparse documentation, manual configuration, and thermal compromises dominate the experience.
A cloud-based coding workflow produces Windows applications with far less friction and cost, exposing the gap between local AI’s theoretical control and its everyday usability.
The discussion broadens to OneDrive Copilot, safer AI agents, Rust rewrites, and Project Helix, where Microsoft’s ambitions meet questions about trust, timing, price, and platform design.
The clearest verdict comes from Paul’s long-running RTX Spark experiment: after extensive setup and model troubleshooting, it still produced no usable application.
Weekly email
State of the Pod, every Monday
The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.