NVIDIA announced at GTC 2026 the next generation of its Deep Learning Super Sampling technology: DLSS 5. According to NVIDIA’s official newsroom, DLSS 5 combines traditional 3D rendering pipelines with generative AI to produce photorealistic 4K video in real time, with AI-generated lighting and materials that the company describes as indistinguishable from physically accurate rendering — while using significantly less computing power than traditional rendering approaches. Jensen Huang told the GTC audience that DLSS 5 represents ‘the future of real-time rendering’ and described it as the biggest advance in graphics technology since the introduction of hardware ray tracing in 2018. DLSS 5 is planned for launch in autumn 2026 and will be available for NVIDIA’s current RTX Blackwell GPU generation. The underlying technology uses a neural network to regenerate photorealistic detail in frames that have been rendered at a lower internal resolution, combining a 3D geometric understanding of the scene with AI-generated content to fill in physically accurate details — a process NVIDIA describes as bridging ‘traditional rendering with AI-generated reality.’ The announcement connects directly to NVIDIA’s broader roadmap for AI in computing: just as Vera Rubin brings agentic AI to data centres, DLSS 5 brings generative AI directly into the real-time rendering pipeline that powers gaming, virtual production, simulation, and digital twin environments. The next GPU generation — Rubin-architecture consumer cards, tentatively expected as the RTX 60 series — is expected to bring DLSS 5’s full capability to consumer desktop hardware, though NVIDIA has not confirmed a timeline for the consumer GPU launch beyond 2027.
NVIDIA Announces DLSS 5 — 3D-Guided Neural Rendering Generates Photorealistic 4K Visuals, Launching Autumn 2026
56