San Francisco Compute Company favicon

GPU Cloud Provider · San Francisco

San Francisco Compute Company

San Francisco Compute Company offers dynamic, market-driven access to large-scale GPU clusters primarily featuring NVIDIA GPUs, without requiring long-term contracts. This flexibility allows customers to purchase GPU time scaled to their needs, from an hour to longer periods, with transparency in pricing and availability.

Founded
Unknown
Countries
1
Data Centers
2
Team Size
11-50

Company Profile

Company TypeStartup
Provider TypeBaremetal Provider
HeadquartersSan Francisco
FundingSeed
Team Size11-50

Infrastructure

GPU FleetNVIDIA H100 SXM, NVIDIA H100 PCIe, NVIDIA A100 80GB
Network Fabric100 Gbps primary network,1 Gbps IPMI out-of-band management network,400 Gbps RDMA compute fabric
Connectivity400 Gbps
StorageHigh IOPS local NVMe storage
Data Center TierCarrier-neutral colocation
Bare MetalYes
AvailabilityGA
StartupEnterpriseAI/ML Researchers

Compute & Deployment

On-DemandYes
Spot / InterruptibleYes (interruptible spot instances at reduced rates)
Reserved InstancesYes (committed capacity contracts available)
Bare MetalYes (primary offering is bare metal GPU servers)
VM-BasedNo
Container-BasedYes (Docker)
KubernetesNo
Serverless GPUNo

GPU Hardware

Latest GenH100 SXM
Legacy SupportA100
Multi-GPU NodesYes (up to 8x per node)
Max GPUs/Node8
NVLinkYes (NVLink on SXM nodes)
InfiniBandYes (InfiniBand interconnect)
PCIe vs SXMSXM only
HGX PlatformYes (HGX H100 8-GPU)

Pricing Model

Per HourYes (primary billing unit)
Per MinuteNo
SubscriptionYes (forward contracts / committed use plans)
Reserved DiscountYes (discounts via forward contracts for committed capacity)
Public PricingPartial (some GPUs listed)
Hidden FeesNone disclosed
Setup FeesNone disclosed
Pay-as-you-goYes

Performance & Scaling

Multi-Node TrainingYes (multi-node clusters supported with high-speed interconnects)
InfiniBandYes (InfiniBand networking available for GPU clusters)
NVSwitchYes (on SXM nodes such as H100 SXM)
Perf IsolationYes (dedicated bare metal GPU instances)
Noisy NeighborYes (bare metal, no sharing)

Developer Experience

OnboardingSelf-serve signup with fast provisioning via web UI or CLI
SDK LanguagesPython
CLI ToolingBasic CLI for instance management and SSH access
JupyterVia SSH port forwarding
DocumentationBasic getting-started guide with API reference
API FeaturesCLI

Security & Compliance

Security
Focused on SF/Bay Area AI startup ecosystemBare metal GPU access with direct hardware performance

Data Center Locations

Coverage

CountriesUnited States
CitiesSan Francisco, CA, Chicago, IL
North America

Compliance Regions

EU Data ResidencyNo EU presence
US Gov CloudNo
India RegionNo
Datacenter Locations

Key Strengths

Highly competitive H100 pricing targeting cost-sensitive AI startups
Simple, no-frills access to large GPU clusters
San Francisco-based focus on the AI research and startup community
Fast provisioning without enterprise sales friction

Known Limitations

Limited geographic availability (primarily US)
Smaller fleet compared to hyperscalers
Limited managed services or MLOps tooling
Documentation and ecosystem less mature than larger providers
No Windows support

Additional Information

Support Options

["Dedicated Slack channel for large clusters","Emergency phone escalation"]

Community

Limited public community presence; primarily direct customer relationships

Core Proposition

Low-cost, high-density bare metal GPU clusters in San Francisco optimized for AI training workloads with spot and reserved pricing.

Payment Methods

Credit CardWire Transfer
Last updated March 2026. Information subject to change.