# The San Francisco Compute Company

> The San Francisco Compute Company (SF Compute) runs a real-time marketplace for AI compute. It buys and operates large-scale, vetted GPU clusters - primarily Nvidia H100s wired with 3.2Tb/s InfiniBand - and sells them on flexible contracts, from a single hour to multiple years, that buyers can also resell. By turning long-term GPU capacity into a liquid spot market with transparent pricing, SF Compute lets startups, researchers, and enterprises buy exactly the compute they need without the multi-year commitments that dominate the industry, while pointing toward a future of cash-settled GPU futures.

- **Founded:** 2023
- **Headquarters:** San Francisco, California, United States
- **Founders:** Evan Conrad (Co-founder & CEO), Alex Gajewski (Co-founder)
- **Team size:** ~54-57 employees (2026)
- **Products:** GPU Spot Market, On-Demand H100 Clusters, Virtual Machines, Managed Slurm & Bare Metal, CLI
- **Notable:** Raised $40M Series A at a $300M valuation led by DCVC and Wing Venture Capital (2025)., Built what it describes as one of the most liquid GPU markets, with compute tradable by the hour., Manages over $100 million in GPU hardware.

## Products & services

- **GPU Spot Market** — A real-time marketplace for Nvidia H100/H200 clusters where compute is bought and sold by the hour, with programmable price limits and the ability to resell unused capacity.
- **On-Demand H100 Clusters** — Vetted, InfiniBand-connected (3.2Tb/s) GPU clusters available from a single hour to multiple years, run from UEFI on up with burn-in testing and monitoring.
- **Virtual Machines** — GPU VMs with roughly five-minute spin-up times, with InfiniBand support on VMs planned.
- **Managed Slurm & Bare Metal** — Managed Slurm scheduling and bare-metal cluster provisioning available on request for large training runs.
- **CLI** — An open-source command-line tool to buy, sell, and manage compute instances programmatically.
- **Large Scale Inference Batch API** — A high-throughput async API for offline inference (data labeling, summarization, synthetic generation) supporting 20+ models, built with Modular, at up to 80% lower cost than typical alternatives.

## Achievements

- Raised $40M Series A at a $300M valuation led by DCVC and Wing Venture Capital (2025).
- Built what it describes as one of the most liquid GPU markets, with compute tradable by the hour.
- Manages over $100 million in GPU hardware.
- Launched a Large Scale Inference Batch API with Modular offering up to 80% lower cost than typical alternatives.
- Recruited Eric Park (former Voltage Park CEO) as CTO, plus talent from Sun Microsystems and Lambda.

## Latest updates

- **2026-01** — Launched the Large Scale Inference Batch API with Modular, supporting 20+ models at up to 80% lower cost.
- **2025-11** — Closed a $40M Series A at a $300M valuation led by DCVC and Wing Venture Capital.
- **2024-12** — Signed a deal with Soluna Cloud to deliver custom AI cloud solutions.

## Links

- Website: https://sfcompute.com
- LinkedIn: https://www.linkedin.com/company/sfcompute
- Twitter/X: https://twitter.com/sfcompute
- GitHub: https://github.com/sfcompute

---

Profile page: https://yespress.io/san-francisco-compute-company
Published by YesPress — https://yespress.io
Last updated: 2026-07-10