Microsoft Auto Super Resolution expands to Intel Core Ultra Series 3

The expansion of Microsoft Auto Super Resolution to Intel Core Ultra Series 3 chips signals a shift toward OS-level AI processing that reduces the burden on game developers. By utilizing the NPU for upscaling, Microsoft allows the GPU to focus on rendering, which is critical for power-constrained devices like gaming handhelds and thin laptops. This move positions Windows as a more capable gaming platform by providing a universal performance boost for legacy titles that lack modern upscaling support. Watch for Microsoft to push Auto SR toward 120fps targets as NPU performance scales in future hardware iterations.

Microsoft's Auto Super Resolution sits within a broader push to make NPUs a standard compute tier in consumer PCs. Intel has been aggressive in positioning its Core Ultra Series 3 architecture around on-device AI workloads, and the company confirmed at Computex 2025 that Arc B390 integrated graphics would ship with dedicated XMX AI engines delivering up to 120 TOPS, a figure that directly determines the headroom available for real-time upscaling tasks like Auto SR. Qualcomm's Snapdragon X Elite, which launched the Copilot+ PC category in mid-2024, was the first platform to receive Auto SR support, and Microsoft expanded the feature to AMD Ryzen AI 300 series processors in early 2025, making the Intel addition the third major silicon vendor to gain compatibility. That three-vendor coverage means Auto SR is now a de facto Windows platform feature rather than a single-vendor showcase.

From a business and ecosystem standpoint, Microsoft is tying Auto SR to its Copilot+ PC certification requirements, which mandate a minimum 40 TOPS NPU. OEMs including Dell, HP, and Lenovo committed to shipping Copilot+ PCs across consumer and commercial lines by late 2025, creating a large installed base that can receive Auto SR updates through Windows Update without any action from game publishers. Intel has also been working with Microsoft on driver-level integration, and the two companies jointly presented NPU-accelerated rendering techniques at Intel Innovation 2025, signaling that Auto SR is one piece of a longer roadmap that includes AI-assisted frame generation and video post-processing on the same silicon.

On the technical side, Auto SR competes with GPU-vendor upscaling solutions that already dominate PC gaming. NVIDIA's DLSS 4, which introduced multi-frame generation capable of producing up to three AI-generated frames per rendered frame at CES 2025, remains the performance benchmark for AI upscaling in games, though it requires explicit developer integration and NVIDIA hardware. AMD's FSR 4, meanwhile, moved to machine-learning-based upscaling with its RDNA 4 architecture in March 2025, narrowing the quality gap with DLSS. Auto SR differentiates by operating at the OS level with zero per-title work, but its NPU-based approach currently targets lower frame-rate uplifts than GPU-native solutions. For streaming and video applications, the same NPU infrastructure that powers Auto SR is being evaluated for real-time video enhancement in Microsoft Teams and Edge browser video playback, suggesting the technology could extend beyond gaming into video delivery workflows on Copilot+ PCs.