Is this £3,500 Ryzen 9 9950X3D and RTX 5080 build balanced for gaming and AI work?

For gaming alone, a Ryzen 7 9800X3D and 32GB of memory would probably feel just as fast while saving a significant amount. The 9950X3D and 64GB make more sense if you’ll genuinely use the extra cores and memory for AI or other workloads. Unfortunately, 64GB DDR5 pricing has become unusually high, so the kit you found may not be far outside the current market price.

I would avoid treating your main gaming PC as a 24/7 server if you can. A separate used low-power or enterprise system is often cheaper, easier to maintain, and avoids downtime when the gaming machine needs upgrades or troubleshooting. If serious AI or server workloads are definitely part of the plan, think carefully about expansion requirements such as PCIe lanes, memory capacity, and storage before committing to a mainstream AM5 platform. Threadripper offers more expansion, but it would be a very different and less gaming-focused build.

There are several places to cut costs without changing the overall idea. A Thermalright Phantom Spirit should cool this CPU very well for around a third of the Noctua’s price. A good B850 motherboard should also be enough unless you specifically need the extra connectivity or expansion features of X870E. DDR5-6000 CL36 is usually cheaper than CL30 and the gaming difference is tiny. You could also consider an RTX 5070 Ti: it is only modestly slower in many gaming and productivity workloads, still has 16GB of VRAM, and could save more than £250. A quality 1000W supply is unnecessary for that card but gives plenty of upgrade headroom. These changes could bring the build close to £3,000.

For local AI, VRAM can matter more than system RAM. Depending on the models you want to run, a used RTX 4090 with 24GB of VRAM could be more practical than pairing 64GB of system memory with a 16GB or similar GPU, provided you find one at a safe price from a reputable seller. That isn’t necessarily the best choice for everyone, but it’s worth deciding which AI workloads you actually expect before spending heavily on memory.

The 9950X3D is powerful, but make sure its extra cores match your actual workload. It can be a good hybrid gaming and productivity chip, especially for shader compilation and CPU-heavy tasks, but it isn’t automatically the best value. If you keep it, check the relevant cache-priority settings in the motherboard firmware. Also research current Nvidia driver and CUDA support on Fedora, since Linux graphics support has historically required more attention than Windows.