NVIDIA unveils Turing GPU architecture, says it includes AI-based capabilities and real-time ray tracing, and announces Quadro RTX GPUs that will ship Q4 2018
Moments ago at NVIDIA's SIGGRAPH 2018 keynote presentation, company CEO Jensen Huang formally unveiled the company's much awaited (and much rumored) Turing GPU architecture.
Context & Ripple Effects
Real-time ray tracing had already been teased in March, when Nvidia put it into the Quadro GV100 as a capability bolted onto Volta. Turing is the moment it becomes an architecture: Jensen Huang is folding ray tracing hardware and dedicated AI compute into the same silicon, and putting it into Quadro RTX parts that ship within months rather than years.
The announcement also sets up the rest of Nvidia's 2018 cadence — a data-center inference part and a flagship consumer card both followed on Turing within months — making this keynote the template for how every later RTX generation has been pitched.
First-order effects
- Professional visualization buyers get a single GPU that renders ray-traced scenes and runs AI models natively, replacing the split between graphics and compute cards for workstation workloads.
- Nvidia converts its SIGGRAPH audience of content-creation software vendors into the launch channel: Quadro RTX shipping in Q4 2018 means ISVs must decide immediately whether to build against the new hybrid pipeline.
Second-order effects
- A month after this keynote Nvidia extended the same tensor-core design to servers with the Tesla T4, claiming a large inference speedup over its predecessor — Turing's AI silicon was aimed at data centers from day one, not just workstations.
- By December the architecture reached enthusiasts at a premium price point with the $2,499 Titan RTX, giving Nvidia a halo product that funds and showcases the new feature set.
Third-order effects
- The pitch structure here — new raster hardware plus an AI feature layered on top — became the durable RTX formula, culminating in the RTX 5090's DLSS 4 generating multiple frames per rendered frame, where AI output is now a headline spec alongside transistor counts.
- Once AI accelerators sit inside every graphics tier from laptops (RTX 500 and RTX 1000) to supercomputers like the DGX GH200, Nvidia's moat shifts from chip manufacturing to the integrated hardware-plus-software stack itself.
The trend: GPU generations are increasingly sold on built-in AI capability rather than raw rendering alone, a playbook that began with Turing's fusion of ray tracing and tensor cores.