Skip to main content
This page is manually updated. For the most up to date information on available instance types, see the dashboard.
Today, we support one instance type: h100v. This is a VM instance with NVIDIA H100s. h100v instances currently don’t have InfiniBand, but will support it in Q2 2026. If you need H200s, B200s, B300s, or other GPU SKUs, talk to us to get access. You can purchase one by running: 
sf nodes create -n 8 -s 'tomorrow at 10am' -d '1d'
See our Nodes docs for more information on purchasing nodes.

​
Available zones

You can purchase nodes on a specific zone. For example, to buy a node on zone pacifica: 
sf nodes create -n 8 -s '+1h' -d '1h' -p 20 -z pacifica
Zones represent physical datacenters and nodes purchased on the same zone will be physically colocated. You can find a list of currently listed zones and their capacities in the dashboard on the zones page.

​
Cluster requirements

San Francisco Compute lists clusters from a number of providers that have passed our vetting process and signed a contract with us that provides a strong SLA, which we then pass onto you. For example, here are some of the requirements for listing a training cluster, such as the h100i instance type. If you’re interested in listing, please contact us for full cluster requirements.
  • 1 TB of RAM minimum
  • 2x AMD EPYC 9004 or Intel 4th gen Xeon CPU
  • Three types of networks:
    • 100 Gbps primary in-band network for orchestration
    • 1 Gbps out of band IPMI management network isolated from the primary network
    • 400 Gbps RDMA compute fabric, typically InfiniBand
  • At least 1 Gbps per node of internet bandwidth
  • Managed switches from an approved list of vendors
  • Air or water cooling, no immersion cooling
  • Proper 48 hour burn-in
  • At least 2 TB of high IOPS local NVMe storage per GPU node

​
Getting a test node

In rare cases, we can provide test nodes before you purchase from San Francisco Compute. However, the advantage of SFC is that you can buy whatever configuration you’d like for a short time period in order to run tests. We would encourage you to try this out first, with the knowledge that you’re covered by an SLA.

​
Hardware failures and refunds

Expect your cluster to break. Failure rates on large scale GPU clusters are far higher than what you may be used to on web servers. When (not if) a portion of your cluster breaks, we will attempt to hot swap you a replacement. If we can’t, we’ll refund the purchase. On a normal GPU cloud, that would be it, you simply have one less node. However, SFC is a market. That means, in many cases, you can just buy another node with your refund. The price is not guaranteed to be the same as when you bought it, but we think this is a better experience than simply being out of luck.