Nvidia CEO Jensen Huang is pushing back hard against the wave of criticism surrounding the company’s newly announced DLSS 5 technology, dismissing detractors by declaring they have the tech “completely wrong.” The comments came during a press Q&A session at GTC 2026, where Huang was asked to address the growing backlash that erupted almost immediately after the feature was unveiled earlier in the week. The announcement had been accompanied by demonstration footage showing AI-enhanced character faces in popular games — imagery that players widely mocked for producing unsettling, overly smoothed results reminiscent of cheap AI-generated advertising.
At the heart of Huang’s defense is a distinction he draws between DLSS 5 and conventional generative AI. According to Huang, the technology “fuses controllability of the geometry and textures and everything about the game with generative AI,” operating not as a post-processing filter applied to finished frames but as what he calls “generative control at the geometry level.” He insists this makes DLSS 5 fundamentally different from the kind of generative AI the public has grown skeptical of, which is why Nvidia brands it as “neural rendering” rather than simple AI upscaling.
A key pillar of Huang’s argument — and one echoed by Nvidia’s development partners — is that DLSS 5 keeps artistic control firmly in the hands of game studios. Developers can reportedly fine-tune the AI’s output to align with their intended art direction, and players will have the option to toggle the feature on or off entirely. Bethesda Game Studios, one of the most prominent partners featured in the reveal, was quick to note that what was shown for Starfield represented only a “very early look” and that its art teams would continue refining the lighting and visual effects before any public release.
Despite these assurances, the gaming community remains largely unconvinced. The demonstration footage drew particular ridicule for its treatment of character faces in titles like Starfield and Resident Evil Requiem, where the AI-enhanced results struck many viewers as eerily artificial. Resident Evil Requiem’s Grace Ashcroft, rendered through DLSS 5, quickly became a meme due to her unnervingly doll-like appearance. Critics also took issue with footage from The Elder Scrolls IV: Oblivion Remastered, arguing that applying hyperrealistic facial features to the game’s intentionally stylized characters undermined the charm of the original art direction.
The roster of studios backing the initiative is nonetheless substantial. Alongside Bethesda and Capcom, partners include Ubisoft, Warner Bros. Games, Tencent, NetEase, NCSoft, and several others. Bethesda’s Todd Howard praised the technology for allowing artistic vision to “shine through without being held back by the traditional limits of real-time rendering,” while Capcom’s Jun Takeuchi called it an important step in pushing visual fidelity forward. Ubisoft co-CEO Charlie Guillemot described DLSS 5 as a meaningful advance toward making game worlds feel more real.
With DLSS 5 slated for a broader rollout this autumn, Nvidia faces a significant perception problem. While the technical distinction Huang draws between neural rendering and generic generative AI may hold weight in engineering circles, the court of public opinion has so far judged the technology primarily on its visible output — and that output, at least in its current early state, has not impressed. Whether refined demonstrations and hands-on experiences can turn the tide before launch remains an open question for Nvidia and its partners alike.