Custom local AI PC assembled in France
"Custom PC," "assembled in France," "local AI": three overused expressions that many sellers employ without them meaning much. Here's what they truly encompass, received idea by received idea, followed by a concrete account of an assembly as practiced at Radiance Systems.
Five Received Ideas About Custom AI PCs
The priorities are reversed. A gaming PC aims for frames per second; an AI PC aims for VRAM, memory bandwidth, and stability over long calculations. A card with a lot of video memory but average graphics performance will be excellent for AI and mediocre for gaming. Designing for AI means making different trade-offs for each component.
True customization focuses on VRAM, RAM, storage, and cooling, adjusted to your specific use. A 14B LLM user, an AI video creator, and an engineer fine-tuning models don't need the same machine, even at the same budget. Color is the last question, not the first.
It's the best way to pay for power you'll never use. If you're running 14 billion parameter models, a 16GB card is sufficient, and a 32GB card will be half idle. Customization is as much about not oversizing as it is about sizing adequately. The goal is suitability, not overkill.
The difference is concrete when something goes wrong. With local assembly, it's the same team that built the machine, tested it, and answers when you call — in French, knowing your configuration. Not a hotline disconnected from production. And the machine remains upgradeable: you can send it back for an upgrade.
That was true two years ago. Today, a machine can be delivered with everything pre-installed: Ollama, Open WebUI or LM Studio, and models already downloaded. The user opens their browser and interacts with their AI like a consumer tool, without typing a single command.
What Assembly Looks Like, Concretely
Customization is not an abstract promise. Here's how the manufacturing of a machine unfolds, step by real step, taking the example of a creator who wants to do AI image and video generation.
Example: a workstation for AI visual creation
The need. The client generates images with Flux and wants to start with AI video. They need a lot of VRAM and bandwidth, but not necessarily an extreme processing CPU. We start from there, not from a catalog.
The core. We select an RTX 5090 32 GB: its memory and throughput are what AI video demands most. This decision structures everything else.
The balance. Around it, we size 64 GB of RAM for encoders and the pipeline, a fast NVMe for large files, a power supply with headroom, and cooling calibrated for long renders without throttling.
Preparation. Upon request, we install ComfyUI, the desired models, and the CUDA environment, so the first generation comes out in minutes, not after a day of configuration.
The workshop. The machine is hand-assembled in Auriol, properly cabled, then tested under real load on AI renders to check temperatures and stability before departure.
Delivery. It ships packaged and insured, anywhere in the EU, ready to plug in. And if the client later wants a second card or more storage, the machine is designed to evolve.
Change the initial need, and the whole chain adapts: a doctor will want a silent, GDPR-compliant server, an ML engineer an ECC card for long training sessions, a gamer a versatile gaming and AI base. The common point: we always start from the use, never from an imposed product.
Sovereignty, Taken Seriously
Running AI locally already keeps your data with you, outside any foreign cloud. Assembling it in France adds end-to-end consistency: design, assembly, and support are local.
For an individual, it's the assurance of an accessible contact person and a durable machine. For a professional subject to secrecy — doctor, lawyer, accountant — it's a direct response to GDPR requirements: no data leaves the office, on locally controlled infrastructure.
Some Basics to Start
All these configurations are starting points, adjustable component by component.
What to Remember
A custom local AI PC is neither a rebadged gaming PC nor a fixed catalog. It's a machine sized for what you actually do, hand-assembled and tested in France, delivered ready to use, and designed to last and evolve. We start from your use, adjust each component, and your data never leaves your hardware.
The rest — case color, lighting — comes last. That's the right order of priorities.
