Newshot

Microsoft patent describes using idle gaming GPUs into paid AI training hardware

Original text(via HN AI)

Microsoft proposes GPU-sharing network with payments to PC owners

Microsoft has filed a patent describing a system that could allow PC gamers to earn money by sharing unused GPU processing power for AI training. The application, published on November 5, 2026, describes how consumer graphics cards could contribute to distributed AI workloads when they are not being used for games or other demanding applications.

The patent, titled “Opportunistic Micro-training on Edge Devices Using Spare GPU Utilization,” was filed by Microsoft Technology Licensing in April 2025. It describes a platform that divides large AI training workloads into smaller tasks and distributes them across gaming PCs, laptops and other compatible devices. The system would monitor available GPU resources and assign work according to each device’s capabilities.

Paid ‘Folding at home’ for AI

The patent also describes compensating users who voluntarily provide their hardware. As quoted by Insider Gaming, the filing states that “the system may track the amount of computations” performed by participating devices, which could “result in a credit being applied” to the user’s account. These credits could reportedly be exchanged for free software, access to AI models or money.

Microsoft patent describes turning idle gaming GPUs into paid AI training hardware
Microsoft patent describes turning idle gaming GPUs into paid AI training hardware

Microsoft’s proposed system would monitor GPU utilization, memory consumption, temperature and power draw. Training tasks would be adjusted according to available resources, with workloads reduced or suspended when the device becomes busy or reaches certain thermal limits. The patent also describes mechanisms to verify completed calculations and prevent unreliable devices from affecting training results.

Microsoft patent describes turning idle gaming GPUs into paid AI training hardware
Microsoft patent describes turning idle gaming GPUs into paid AI training hardware

There is still the question of whether users would actually make money from participating. Running AI workloads increases electricity consumption, even when the PC would otherwise be idle. For example, an additional 250W of power consumption for four hours daily would add 30 kWh to the monthly electricity bill. At $0.35 per kWh, this represents $10.50 in additional costs. Since this is just a patent, not even a real program or product, there is currently no way to determine whether participating would be profitable.

This actually reminds me of cryptocurrency mining software. You’re essentially getting paid for your GPU’s computing power, although I doubt it would ever be as profitable as crypto mining was at its peak. Still, I hope Microsoft doesn’t forget that’s not what GPUs were designed for.

Source: US Patent Office (PDF) PC Games Hardware

The text above was extracted automatically from the source page and may miss images, tables or code blocks. The original is authoritative.

Published by HN AI
This site only summarises and aggregates; copyright stays with the original author. Headlines and summaries follow the source — where they differ, the original wins.
Read the original ↗