Microsoft is announcing local AI on Windows at an October 7th event in San Francisco, positioning it as the PC's next fundamental shift. This claim arrived two years after the company made an identical one about Windows Copilot. Shipped as a non-functional search alternative and users stopped opening it within months.
The pattern matters. A gap of more than two years between major Windows events suggests Microsoft either abandoned its previous narrative or needed time to fix what it oversold. Neither fact is reassuring.
Windows Copilot in 2023 was positioned identically — a transformative moment for how users work with AI, seamlessly integrated, a genuine capability shift. It landed as a sidebar search box that duplicated web results slower than Bing did. Users tested it, found nothing, and stopped testing it. The feature still exists in the OS. Almost no one knows it's there. This is Microsoft's most recent evidence of what 'local AI on Windows' means in practice.
Jensen Huang's appearance at the October 7th event changes the frame. If this is genuinely about RTX Spark PCs — consumer hardware with the neural processing power to run meaningful models locally, not cloud-dependent Copilot-style token forwarding. Then something has actually shifted. The problem is Microsoft has already spent its credibility on that narrative. A user who watched Copilot get announced, demoed, shipped hollow. Abandoned has no reason to trust the same company's next 'transformative' promise, even if Nvidia's hardware executives are standing onstage.
The test is concrete: Do these AI tools do something on your machine that requires no internet? Do they change how you actually work, or do they exist in the gap between announcement and implementation? Microsoft will answer the first question on October 7th. The second question has already been answered. Users learned to wait and see. Is what happens when a company's last transformative moment transformed nothing.