Ce guide n'est disponible qu'en anglais pour le moment.
Which workstation do I need for data science?
Data science is a mix of loading data, cleaning it, exploring it and training models. A laptop slows down once datasets reach a few gigabytes. A workstation keeps the whole dataset in memory and uses the GPU to speed up the slow steps.
Dernière vérification : 13 Oct 2026
Vous hésitez ? Demandez à notre conseiller IA
Indiquez-lui la taille de vos jeux de données et les outils que vous utilisez. Il ne propose que des machines que nous vendons, avec des prix et des délais de livraison réels.
RAM comes first
Most tools load the whole dataset into memory, then make copies while they work. As a rough guide, have at least two to three times your largest dataset in RAM.
| Largest dataset | Suggested RAM |
|---|---|
| Up to 10 GB | 32 to 64 GB |
| 10 to 50 GB | 128 GB |
| 50 to 150 GB | 256 GB or more |
When a GPU helps
- Deep learning: training neural networks is much faster on an NVIDIA GPU.
- Data frames: GPU libraries such as NVIDIA RAPIDS can speed up pandas-style work on large tables.
- Classic machine learning: many tools (for example XGBoost) can train on a GPU.
If you only use spreadsheets and small tables, spend more on RAM and CPU cores instead.
Workstations in our catalogue
Featured machines. Prices, stock and lead times are live.
- Unified memory: 48 / 64 / 128 GB
- Memory bandwidth: Up to 614 GB/s
- Chip: Apple M5 Max — 18-core CPU (6 super + 12 performance cores)
- Unified memory: 96 / 256 GB (512 GB: see pre-order)
- Memory bandwidth: Up to 1.2 TB/s
- Chip: Apple M5 Ultra — 30- or 36-core CPU
- Platform: AMD WRX90
- CPU socket: sTR5
- CPU options: Threadripper PRO 9965WX / 9975WX / 9985WX / 9995WX
- Warranty: 3-year depot
- Storage: 16 TB (4× 4 TB Samsung 990 PRO NVMe)
- GPUs: 3 or 4× NVIDIA RTX 6000 Ada Generation 48 GB
- Platform: AMD AM5
- Form factor: Mid tower
- CPU: AMD Ryzen 9 9950X (16 cores)
- OS: Ubuntu 24.04 LTS + NVIDIA AI stack
- GPUs: 2× NVIDIA RTX PRO 6000 Blackwell Workstation Edition
- Boot drive: 4 TB PCIe 5.0 NVMe
Questions fréquentes
How many CPU cores do I need?
For most data science, 16 or more cores is plenty. More cores help if you run many jobs at once or use tools that spread work across cores.
Are your prices shown with VAT?
Yes. Every price is shown without VAT and with VAT, and there are no hidden fees.
How long will it take to arrive?
Each product shows its real lead time from our stock data. The advisor uses the same data and won't guess a date.
Obtenez une suggestion adaptée à vos besoins
Indiquez-lui la taille de vos jeux de données et les outils que vous utilisez. Il ne propose que des machines que nous vendons, avec des prix et des délais de livraison réels.