Nvidia CEO: ‘The longer term [of GPUs] is neural rendering. That’s the way in which graphics should be’
Love or hate it, upscaling expertise like Nvidia’s DLSS have expanded the definition round gaming efficiency. And whereas {hardware} lovers nonetheless need to know what to anticipate for raster efficiency, freed from any software program methods, we’ll have to attend some time longer for a definitive reply.
When PCWorld’s personal Adam Patrick Murray requested concerning the RTX 5090 and the way forward for AI gaming GPUs at a CES 2026 Q&A session for media and analysts, Nvidia CEO Jensen Huang dove briefly into his present view on GPUs, AI, and gaming—one the place upcoming video video games will home layers and layers of AI:
(Transcript evenly edited for readability.)
PCWorld: Adam, with PCWorld, I’d like to speak about gaming for a second—
Huang: Sure, it’s superior. Me too!
PCWorld: So Nvidia continues to push DLSS to be higher and quicker with what’s been launched—
Huang: Fairly superb, proper? Simply rapidly: Earlier than GeForce introduced CUDA to the world, which introduced AI to the world, after which after that, we used AI to deliver RTX to players and DLSS to players. And so, , with out GeForce, there can be no AI at this time. With out AI, there can be no DLSS at this time. That’s nice.
PCWorld: It’s harmonious, yeah. My query, considered one of my questions, is—is the RTX 5090 the quickest GPU that players will ever see in conventional rasterization? And what does an AI gaming GPU appear like sooner or later?
Huang: I believe that the reply is tough to foretell. Perhaps one other manner of claiming it’s that the longer term is neural rendering. It’s principally DLSS. That’s the way in which graphics should be. And so, I believe you’re going to see increasingly advances of DLSS. We’re engaged on issues within the lab which might be simply completely stunning and unbelievable. And so I’d count on that the flexibility for us to generate imagery of virtually any model from picture realism, excessive picture realism, principally {a photograph} interacting with you at 500 frames a second, all the way in which to cartoon shading, should you like. All that total vary goes to be fairly smart to count on. You also needs to count on that future video video games are primarily AI characters inside them, and so it’s virtually as if each character could have their very own AI, and each character will likely be animated robotically utilizing AI. The realism of those video games goes to essentially, actually climb within the subsequent a number of years, and it’s going to be fairly extraordinary, . And so I believe this can be a nice time to be in video video games, frankly.
With the shock announcement of DLSS 4.5, which is able to provide RTX players additional decision boosts and body charge bumps, Nvidia does look like targeted much less on the place gaming efficiency enhancements come from. Its launch of G-Sync Pulsar show tech (out there in choose displays beginning this Wednesday, January 7) additionally displays this stance, enhancing movement readability with a mixture of high-end panel specs and intelligent software program management.
