About
machine0 builds CLI- and MCP-managed persistent cloud VMs for long-running AI agents, including coding and auto-research workloads. It targets developers and teams operating agent fleets, differentiating through always-on dedicated compute, static IPs, HTTPS endpoints, broad CPU/GPU scaling, and per-minute billing.
Market
machine0 competes in developer cloud infrastructure and AI-agent compute, offering persistent dedicated KVM/QEMU VMs with optional GPUs, reproducible NixOS or Ubuntu images, static IPs, HTTPS endpoints, and CLI/MCP control. It differentiates from general-purpose VM clouds such as Fly.io, DigitalOcean, and Hetzner, GPU-focused platforms such as Runpod and Modal, and agent sandbox platforms such as Daytona through persistent state, dedicated resources, open-source images, suspend/resume, and per-minute billing rather than a primarily serverless or ephemeral execution model.
AI-native startups and small-to-mid-sized software, machine-learning, and infrastructure teams that need persistent, dedicated cloud compute for long-running agents, development, hosting, or GPU workloads. The primary users and buyers are developers, ML engineers, and platform/infrastructure engineers who prefer CLI- and automation-driven workflows.
At a Glance
Problem
AI agents increasingly need persistent, always-on computing rather than short-lived execution environments, but running them on a developer’s laptop is limiting for builds, tests, long jobs, and continuous operation. The pain is both operational and economic: agents need dedicated CPU, memory, storage, networking, and sometimes expensive GPUs, while users want to avoid paying for idle compute or managing infrastructure manually. The central use case is giving an agent its own reliable cloud computer so it can run software projects, experiments, and long-running workflows independently.
Product / Service
machine0 provides persistent cloud virtual machines that agents can create and manage through a CLI or remote MCP server. A command such as machine0 new mybox produces an SSH-ready VM with a static IP and HTTPS endpoint; machines remain available until stopped or suspended, and profiles can inject selected tools, credentials, prompts, and environment variables. Agents can therefore create boxes for builds, tests, and long jobs, host themselves continuously, or clone and manage a fleet without constant human intervention.
The service uses dedicated CPU, RAM, and optional GPU resources, ranging from small CPU machines to systems with up to 60 vCPUs, 240 GB of RAM, and eight high-end GPUs. It is pay-as-you-go and billed by the minute, starting at $0.013 per hour for CPU VMs; suspending a VM stops compute charges while preserving the machine for later resumption. This combines the power and persistence of a cloud server with a usage model designed to reduce the cost of idle infrastructure.
Market
machine0 competes in cloud infrastructure and developer tools for AI agents, particularly persistent agent computers and on-demand compute. Its closest adjacent alternatives include E2B, which provides secure computers for AI agents and agentic workflows, and Modal, which provides on-demand high-performance AI infrastructure and GPU capacity. machine0 is differentiated by offering general-purpose persistent VMs controlled directly by agents through CLI or MCP, rather than focusing only on ephemeral sandboxes or model-serving infrastructure.
The company appears to be at a very early commercial stage. Y Combinator lists it as an active Summer 2026 company based in Paris with a one-person team, and its YC launch announcement is dated July 31, 2026. The available evidence confirms YC backing, public launch, and a live pay-as-you-go product, but does not disclose customer counts, revenue, or usage metrics; therefore, traction beyond launch and availability is not established in the research corpus.
Founders & Leadership
Funding History
Y Combinator
Recent News
machine0 launched as a CLI for creating, provisioning, and snapshotting persistent NixOS and Ubuntu VMs. The service offers static IPs, HTTPS endpoints, dedicated compute, optional GPUs, and lifecycle management through the CLI.
Y Combinator profiled machine0 as a cloud-compute CLI for AI agents. Its VMs are billed by the minute, support up to 60 vCPUs, 240 GB of RAM, and 8×H100 GPUs, and can be managed by agents through the CLI or MCP.
machine0’s documentation describes its persistent cloud VM service, including dedicated CPU, RAM, and GPU resources, static IPs, HTTPS endpoints, and a remote MCP server that lets agents manage infrastructure programmatically. It also documents integrations and profiles for services such as GitHub, Claude Code, Codex, and remote MCP servers.
Active Roles
0No active roles right now.
Get notified when they postBusiness Model
machine0 uses a pay-as-you-go infrastructure model, charging customers per minute for active VM usage, with CPU and GPU instances priced by compute capacity. Suspended VMs stop compute billing while customers pay only for image storage; the service also requires a minimum $5 top-up.