spot-instances
(4)
TensorPool
TensorPool (YC W25) is a San Francisco startup that turns cloud GPUs into a git-style command line. Instead of wrestling with SSH sessions, provider dashboards, and idle-instance billing, machine learning engineers describe a training job in a config file and run 'tp job push' - TensorPool spins up a multi-node cluster across partner clouds, streams logs, saves outputs, and shuts everything down, billing only for runtime. It advertises GPU access at roughly half the cost of major cloud providers, with H100s starting around $1.99/hr. Founded by three Stanford dormmates, the company was accepted into Y Combinator's Winter 2025 batch.
Coiled
Coiled is a lightweight cloud compute platform that lets Python users scale data science, machine learning, and AI workloads from a laptop to thousands of cloud machines without Docker or Kubernetes. Founded in 2020 by Matthew Rocklin, the creator of the open-source Dask library, Coiled runs inside a customer's own AWS, GCP, or Azure account, handling provisioning, autoscaling, environment replication, and cost visibility so data teams can run Dask clusters, serverless functions, and batch jobs with minimal friction.
Zesty
Zesty builds an automated Kubernetes and cloud optimization platform that continuously right-sizes compute, storage, and cloud commitments to match real-time workload demand. Its flagship Kompass platform cuts Kubernetes compute and storage costs by up to 70% without compromising performance or SLAs, using multi-dimensional autoscaling, fast node deployment, and predictive automation. Founded in 2019 and backed by roughly $129M in funding, Zesty manages billions in cloud spend across thousands of accounts for engineering and FinOps teams.
nOps
nOps is an AI-powered FinOps platform that automatically optimizes AWS cloud costs for enterprises. Managing over $4 billion in annual cloud spend across 600+ customers, nOps uses machine learning to intelligently provision compute resources, manage AWS commitments, and deliver 50%+ cost savings without operational overhead. The company's Compute Copilot product integrates with Karpenter to optimize Kubernetes workloads, while its Clara AI agent answers cost questions and executes optimization tasks.