A new version of Blender does not usually change how we spec workstations. Blender 5.1 is a bit different. It switches on AMD's HIP-RT hardware ray tracing by default in the Cycles renderer, which is a bigger deal than it sounds. For years, AMD ray tracing in Blender has felt like a feature you had to go looking for. Now it is just on.
Why this matters more than a changelog entry
Cycles has been an Nvidia story for a long time. CUDA and OptiX have been the default, most mature path through the renderer, and that has shaped everything around it: tutorials assume Nvidia, render farms are built around Nvidia, and studio pipelines have Nvidia baked into their habits. When something is the default for that long, it stops being a choice and starts being the assumption nobody questions.
AMD support existed before 5.1, but it was the kind of thing you enabled manually and hoped for the best. Making HIP-RT the default is Blender's way of saying it is confident enough in AMD's hardware ray tracing to put it in front of every user, not just the ones who go digging in preferences.
What has not changed
Nvidia has not stood still, and it still has two advantages that are easy to overlook if you only think about Blender in isolation.
- Software ecosystem: CUDA is not just a Blender thing. It underpins a huge range of other 3D, AI and simulation tools, so an Nvidia card tends to keep working hard long after you have closed Blender for the day.
- VRAM at the top end: cards like the RTX 5090 offer 32GB, and large, complex scenes eat VRAM regardless of which renderer or which brand of GPU you are using. More headroom means fewer compromises on scene complexity.
Neither of those points is about Blender rendering speed specifically. They are about what else the machine needs to do, and that is exactly where this decision actually gets made.
So does Nvidia still win?
It depends on the question you are actually asking. If the question is narrowly "which GPU is more competitive in Blender's Cycles renderer than it used to be", then AMD has made real, genuine progress, and it would be dishonest to pretend otherwise.
If the question is "which GPU should I put in my workstation", the answer gets more interesting. A machine that only ever needs to open Blender has far more room to consider AMD than it did a year ago, especially if you also game and AMD suits that side of your life too.
A machine that splits its time between Blender and other CUDA dependent creative or AI tools is a different case entirely. The moment your pipeline touches software that leans on CUDA, Nvidia's broader ecosystem support stops being a nice-to-have and becomes the thing that keeps your workflow from breaking. The same goes if you routinely work on very large, VRAM hungry scenes, where the 5090's 32GB buys you room that a HIP-RT improvement in Cycles does not replace.
Why we do not give everyone the same answer
This is the part build guides tend to skip. There is no single right GPU for "a Blender workstation" because there is no single Blender workstation. We build dedicated systems for Blender, Cinema 4D and ZBrush, and the right GPU for each one depends on what the machine actually needs to do, not on whichever card had the most exciting changelog that month.
If your pipeline is genuinely Blender only, and you have told us that, AMD becoming credible is useful information and we will factor it in. If your pipeline moves between Blender, other 3D packages and AI tools, we will spec around the software that has the least flexibility, which today is still usually Nvidia. Either way, the spec should follow your pipeline, not a headline.
Build it with us
Have a play with our PC configurator, or just get in touch and tell us how you actually use your machine. Shiv and Matt between them handle most of our 3D and rendering builds, so they're worth asking. Every system is hand-built to order here in our Stevenage workshop, stress-tested before it ships, and backed by our 5-year warranty.
