Technology & Software
Server infrastructure for AI & Machine Learning
Sized from the workload rather than sold as a plan name: what AI & Machine Learning actually run, what that needs, and which tier it belongs on.
Recommended builds by size
Three sizes of operation, each sized from an assumed traffic figure.
Small
1–10 people · ~5,000 visits a month
VDS
- 2 Cores
- 32 GB RAM
- 450 GB storage
A single site or team, with traffic that arrives in a predictable working-hours pattern.
Medium
11–50 people · ~50,000 visits a month
Dedicated Server
- 6 Cores
- 96 GB RAM
- 1500 GB storage
A growing business running several systems, where an outage now costs real money.
Large
51+ people · ~500,000 visits a month
Bare Metal
- 20 Cores
- 320 GB RAM
- 4000 GB storage
Multiple sites or a busy platform, with traffic that spikes and an obligation to stay up.
Which tiers suit AI & Machine Learning
| Infrastructure | Verdict | Why |
|---|---|---|
| Shared Hosting Managed space on a shared server | Not this one | There is no version of this workload that fits on shared hosting. |
| VPS Flexible virtual infrastructure with full root access | Workable | Useful for development, notebooks and light inference, but not for training or sustained compute. |
| VDS Virtual server with resources pinned to you alone | Workable | Suits CPU-bound preprocessing and data preparation with a guaranteed share of the machine. |
| Dedicated Server An entire physical server, configured to your specification | Recommended | Production compute wants the whole machine, with memory bandwidth that is not being shared. |
| Bare Metal Physical compute for sustained, demanding workloads | Recommended | Training, inference and simulation belong on physical hardware, with GPUs available on request. |
AI & Machine Learning workloads you can host
Training saturates hardware for hours at a time and inference wants it available instantly — two opposite demands usually met with two different machines.
- Model training
- Inference endpoints
- Vector databases
- Dataset storage
- Experiment tracking
- GPU scheduling
AI & Machine Learning hosting by city
Each page names the closest of our 24 data centres and how far away it actually is.
- New York
- Toronto
- Los Angeles
- Montréal
- Chicago
- Vancouver
- Calgary
- Houston
- Edmonton
- Phoenix
- Ottawa
- Philadelphia
- San Antonio
- Winnipeg
- Québec City
- San Diego
- Dallas
- Hamilton
- Austin
- Kitchener
- Halifax
- San Jose
- Jacksonville
- Victoria
- Saskatoon
- Seattle
- Denver
- Regina
- Boston
- Mississauga
- Atlanta
- Brampton
- Miami
- San Francisco
- ‘Ewa Beach
- ‘Ewa Gentry
- Abbey Wood
- Abbotsford
- Aberdare
- Aberdeen
- Aberdeen
- Aberdeen
- Aberdeen
- Aberystwyth
- Abilene
- Abingdon
- Abington
- Abington
- Abu Hayl
- Accrington
- Acocks Green
- Acton
- Acton
- Acworth
- Ada
- Adams Morgan
- Addison
- Addison
- Adelaide city centre
- Adelaide Hills