TensorPool Blog · Mar 2026
Announcement of globally distributed S3-compatible object storage with intelligent caching, no ingress/egress fees, and latency/throughput benchmarks versus Cloudflare R2.
Company research
Public company, workplace, funding, and market signals
Updated Jul 29, 2026
Overview
TensorPool is a Y Combinator Winter 2025 startup building a developer-friendly GPU cloud for ML workloads: a CLI-first platform for launching, monitoring, and managing GPU clusters and training jobs from the IDE, with multi-cloud cost optimization and storage/networking tuned for distributed training.
Primary product
CLI-first GPU cloud platform for ML training and multi-node GPU cluster orchestration.
Founded
2025
Headquarters
San Francisco, California, United States
Team size
1-10 employees
Industry
Technology, Information and Internet
Sub-industry
GPU cloud infrastructure / AI infrastructure
Offices
San Francisco, California, United States
Business model
usage-based GPU-hour billing
storage-as-a-service
enterprise reservations and custom contracts
freemium / free credits for early users
Funding
Stage
pre-seed / accelerator
Total raised
$500K
Latest round
Pre-Seed / Accelerator · Feb 2025
Latest amount
$500K
Feb 2025
People & investors
Leadership
CEO & Co-Founder
Co-Founder & CTO
Co-Founder & CPO
Work & culture
Developer-first and product-fast: the company emphasizes making GPU access feel like local development, publishes frequent product updates, and leans on an active Slack/Discord community with credits and early-user outreach.
Compensation
No public employee salary bands or hiring compensation data were found. Publicly visible incentives are product credits/free compute rather than compensation disclosures.
Benefits
24/7 support via Slack
$5/week free compute for early users
$20 credit for starring the GitHub repo
startup program credits ($1,000-$100,000)
academic program credits ($1,000-$100,000)
99.9% uptime SLA on enterprise plans
shared storage and high-performance networking
no ingress/egress fees on object storage
Product & market
Pricing
Usage-based pricing per GPU-hour, plus storage pricing by TB/month and object-storage request charges; enterprise and committed-discount pricing available on request.
Differentiators
- —CLI-first workflow from the IDE
- —No SSH or manual data migration required
- —Real-time multi-cloud price routing to the cheapest provider
- —Spot-node pricing with on-demand-like reliability
- —Git-style jobs with TOML configs and versionable experiments
- —High-performance shared storage and networking for distributed training
- —S3-compatible globally distributed object storage
- —Free early-user compute credits
Technology
TensorPool CLI
TOML job configuration files
git-style GPU job workflows
multi-cloud GPU orchestration
spot-node resummation
S3-compatible object storage
high-performance NFS/shared storage
multi-node GPU clusters
Jupyter/nb.dev notebook workflows
Slack/Discord community support
Customers
ZeroEntropy
ML startups
academic research labs
machine learning engineers
enterprises needing cheaper GPU access
Competitors
AWS SageMaker / EC2
Google Cloud / Vertex AI
Lambda Labs
RunPod
CoreWeave
Vast.ai
Scale & contact
Estimated revenue
≈$330K ARR (estimated)
Estimated monthly visits
9K
Traffic estimate as of Feb 2025
Latest news
TensorPool Blog · Nov 2025
Customer case study showing ZeroEntropy used TensorPool for zerank-1 and zerank-2 training and that TensorPool helped speed experimentation.
TensorPool Blog · Oct 2025
Product announcement for Jobs, a git-style interface for training workloads that spin up clusters, run jobs, save outputs, and bill only for runtime.
Startup Intros · Mar 2025
Third-party funding writeup stating TensorPool raised $500K in a seed round in March 2025.
Y Combinator · Feb 2025
Official YC launch announcement framing TensorPool as an easy way to use cloud GPUs from the IDE with a CLI-first workflow, multi-cloud cost routing, and spot-node resummation.
Jobs and careers at TensorPool
0 jobs at TensorPool
No open positions
Check back later for new openings