Affordable
Shared compute
The lowest-cost way to get your pipelines running
Starting at $1,199/yr
Jump to the detailsCompute infrastructure
Shared, dedicated, or on-prem— compared on one page
Three options for three kinds of work: getting started solo, running a lab long-term, and keeping sensitive data in-house.
Shared from $1,199/yrDedicated from $8,600/yrOn-prem: one-time investment
Positioning
Three options, three kinds of work
Choose with the trade-offs in mind: shared solves affordability, dedicated solves speed and stability, on-prem keeps data in your own hands.
Powerful
Dedicated servers
A whole machine to yourself. Fast, no queues.
Starting at $8,600/yr
Jump to the detailsPrivate
On-premise infrastructure
Compute next to your data. Data never leaves your network.
Starting at $30,000
Jump to the detailsAt a glance
The differences, side by side
Shared compute
Affordable- Resource model
- Shared physical server, isolated quota
- Typical specs
- From 20 cores · 256GB
- Service & support
- Preconfigured environments + technical help
- Best for
- Individuals, experiments, short phases
- Starting price
- $1,199/yr · subscription
Dedicated servers
Powerful- Resource model
- A whole machine, exclusively yours
- Typical specs
- From 12 cores · 128GB · 5TB
- Service & support
- Real-time support + full migration
- Best for
- Labs and long-running projects
- Starting price
- From $8,600/yr · subscription
On-premise infrastructure
Private- Resource model
- Deployed in your own facility
- Typical specs
- 8 × A100 / H100 / H200
- Service & support
- Selection, deployment, training + remote support
- Best for
- Sensitive data and heavy 24/7 workloads
- Starting price
- From $30,000 · one-time purchase
Powerful · Dedicated servers
A whole machine to yourself. Fast, no queues.
The entire server belongs to one lab — roughly 2× the speed of shared.
What’s included
- The whole machine — no waiting on resources
- Real-time online support for environment issues
- Code, environments, and data migrate as a whole
- One place to manage members and resources
Where it shines
- Long-term use by a whole lab
- Long-running projects with deadlines to hit
- No queues at workload peaks
Pricing & specs
Dedicated Starter
From $8,600/yr
12 cores · 128GB memory · 5TB storage
Dedicated Advanced
From $16,800/yr
24 cores · 384GB memory · 10TB storage
Dedicated Pro
From $33,600/yr
48 cores · 768GB memory · 20TB storage
Private · On-premise infrastructure
Compute next to your data. Data never leaves your network.
Hardware installed in your own machine room or office. For steady, heavy use, buying beats renting cloud year after year.
What’s included
- Configured around your models, data, and budget
- Remote deployment: OS, drivers, and environment set up in one pass
- Migration of existing data and dev environments
- Team training plus post-delivery remote support
Where it shines
- Sensitive data stays inside your network
- Training every day — owning beats renting
- Dedicated capacity for 24/7 long-running jobs
Product lines & specs
Desktop supercomputer
From $30,000 · quoted by configuration
Desk-sized, silent, plugs in and runs
Rack-scale 8-GPU system
From ~$120,000 · quoted by configuration
8 × A100 / H100 / H200, networkable into clusters
How to choose
Find your situation
If you run routine analysis yourself and want a low-cost start
A better fit
Shared compute
From $1,199/yr
If your lab shares one environment and projects run for months
A better fit
Dedicated servers
From $8,600/yr
If your data is sensitive, you need private deployment, and the machine never idles
A better fit
On-premise infrastructure
One-time investment
If your workloads need GPUs but you’re not ready to buy hardware
A better fit
Cloud GPU
From $1,299/quarter
Still deciding
Send us your workloads — we’ll help you choose
Tell us the job, the team size, and the data volume. We’ll pick with you and come back with a configuration and a quote.