San Francisco Compute Company logo

GPU Cloud Provider · San Francisco

San Francisco Compute Company

The San Francisco Compute Company is a provider specializing in GPU instances for computing, particularly suitable for training large AI models. They offer flexible, short-term access to GPU clusters without long-term contracts, appealing to startups and businesses focusing on large-scale model training.

Founded
Not specified

Company Profile

FoundedNot specified
HeadquartersSan Francisco
Legal EntityThe San Francisco Compute Company

Infrastructure

GPU Fleet8 x H100 80GB HBM3, 8 x H200 141GB HBM3e
Network FabricInfiniBand, Scheduled for future support in VMs by Q2 2026
Connectivity400 Gbps RDMA compute fabric,100 Gbps primary in-band network,1 Gbps out of band IPMI management network
Storage1.5TB+ NVMe, Petabyte-scale storage options
Bare MetalYes (contact sales for bare metal or managed Slurm)
AvailabilityOperational with limited clusters available for immediate deployment
leading labs, startups and enterprises

Compute & Deployment

On-Demandtrue
Spot / Interruptibletrue
Bare Metaltrue
VM-Basedtrue
Spin-Up TimeAbout 5 minutes.

GPU Hardware

Latest GenB300s, H100s, H200s
Multi-GPU NodesNodes with multiple GPUs (8 GPUs per node in listed SKUs)
Max GPUs/Node8
InfiniBandBare metal and managed Slurm clusters support InfiniBand today; InfiniBand on VMs planned for Q3 2026

Pricing Model

Per Hour$2.01/hr
SubscriptionVMs on a flexible contract you can sublease when your needs change
Hidden FeesNever pay to move your data.
Egress ChargesNever pay to move your data.

Performance & Scaling

Multi-Node Training# Buy 32 nodes (256 H100s) for 3 days at market price
Elastic ScalingSF Compute's convenient and easy to use platform is fantastic for elastic, on-demand compute.
InfiniBandContact us for a BM or managed Slurm cluster which both support InfiniBand today. We'll support InfiniBand on VMs in Q3 2026.

Developer Experience

OnboardingBuy on website or via CLI; CLI examples provided
CLI Toolingsf CLI with commands including sf buy, sf sell, sf instances create, sf instances ssh, sf skus list, sf skus get
DocumentationAPI reference, CLI command examples, SKU list and SKU details
API FeaturesCLI, SDK expected (implied integration capabilities)

Security & Compliance

Security
Featured in The Wall Street Journal, Bloomberg, The New York Times and Latent Space PodcastNamed customer testimonials (DatologyAI, Ultravox, Mach9, Roboflow)

Data Center Locations

Coverage

Compliance Regions

Datacenter Locations

Key Strengths

Runs large-scale, vetted GPU clusters and can build custom clusters
Offers VMs via website/CLI and bare metal/managed Slurm by contact
No data egress charges for moving your data
Refunds for failed nodes and can repack to healthy hardware
Dedicated Slack channel and emergency phone escalation for large clusters
Competitive pricing example: H100s from $2.01/hr

Known Limitations

VMs are the default purchase option; bare metal and managed Slurm require contacting sales
InfiniBand on VMs is not yet available (planned Q3 2026)
Hardware failures occur and nodes can go down; failures are expected at scale
No explicit multi-region, free tier, or payment method details provided

Additional Information

Support Options

24/7 support via Slack, phone, and email; dedicated Slack channels for large clusters

Community

["Twitter X account @sfcompute","Dedicated Slack channel for large clusters (customers)"]

Core Proposition

Never pay to move your data.

Notable Customers

DatologyAI
Ultravox
Mach9
Roboflow
Last updated March 2026. Information subject to change.