boat vs STACKIT
STACKIT is a European cloud from the Schwarz Group that rents virtual machines you set up yourself. boat also runs full VMs in the EU, but each one comes ready for AI agents, with a desktop, coding agents, minute-by-minute snapshots and HTTPS links built in.
Where boat stands out
Coding agents you drive from outside
Every boat sandbox ships with Claude Code, Codex, OpenCode and other agents, and you drive agents inside from outside with a prompt, steer and events API. boat's credentials proxy keeps real AI keys out of the machine and can switch to another account when one runs out of credits. A STACKIT server is an empty VM: its Server Agent runs scripts, not agents.
A desktop and browser for computer use
boat streams a 60fps desktop with Chromium from every sandbox, with computer use and browser control, and you can embed the stream in your own app. STACKIT servers have no graphical desktop; the portal console is a text console for fixing a broken server.
Snapshots every minute, free stops, one-call forks
boat snapshots each sandbox's disk every minute and on stop, a stopped sandbox costs nothing, and you can fork a running one in a single call. On STACKIT you take snapshots yourself, a stopped server stays reserved until you deallocate it, and a clone means copying the boot volume and creating a new server from it.
The companies and products
boat
boat.dev runs sandboxes for AI agents: each one is a full Linux VM with its own kernel.
You start a machine with one command or one API call, then SSH in, run Docker, open a desktop, or let a built-in agent (Claude Code, Codex, OpenCode and others) work in it. The disk stays between sessions, and a stopped sandbox costs nothing.
- Isolation
- Full VM per sandbox, own kernel
- Regions
- EU: Germany, Finland, France
STACKIT
STACKIT is a general European cloud run by Schwarz Digits, the IT arm of the Schwarz Group, which also owns Lidl and Kaufland.
STACKIT Compute Engine rents virtual machines by the hour, in shared or dedicated vCPU sizes, with Linux, Windows, Arm and Nvidia GPU options. You attach disks, networks, security groups and public IPs yourself through a portal, CLI, API or Terraform. It is not an agent sandbox: it is built for companies that want to run their own servers and over 40 cloud services in European data centers.
Which use cases each one fits
For each use case, Battleships prices every sandbox provider on the same workload and sorts them from the cheapest monthly bill up. “3rd cheapest of 367” means 2 of the other 366 providers run that workload, with every feature it needs, for less. “With limits” means the provider runs it but misses something, so it is listed after every full match. Open a use case for its whole ranking and what each provider misses.
| Use case and its workload | boat: price rank | STACKIT: price rank |
|---|---|---|
| Software factory 100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month Both run Docker on a full VM; on STACKIT you install the agents, the browser and the tooling yourself and wire up your own way to reach each machine. | 3rd cheapest of 367 $634 a month | 238th of 367, with limits $3,226 a month |
| Always-on AI employees 50 machines 24/7 · 2 vCPU / 4 GB Both can run an assistant for months; boat adds snapshots every minute and the agent driving API, which on STACKIT you would build yourself. | 5th cheapest of 366 $657 a month | 194th of 366, with limits $1,873 a month |
| Code interpreter bursts 100 at once · 1 vCPU / 2 GB · 200,000 sessions of 1 min a month Both are a compromise here: STACKIT bills each started hour, so one-minute jobs pay for an hour, and boat's start limits slow down big bursts. | 186th of 345, with limits $2,000 a month | 184th of 345, with limits $1,901 a month |
| Sandboxes for your users 300 at once · 2 vCPU / 4 GB · 30,000 sessions of 23 min a month STACKIT does not publish how many servers a project can run at once, so 300 machines for your users need a quota you confirm with them first. | Cheapest of 361 $209 a month | 133rd of 361, with limits $1,511 a month |
| RL rollouts 1,000 at once · 2 vCPU / 4 GB · 1,000,000 sessions of 3 min a month | 34th of 360, with limits $2,000 a month | 201st of 360, with limits $36,989 a month |
| Agent evals 250 at once · 4 vCPU / 8 GB · 15,000 sessions of 20 min a month | 10th cheapest of 364 $200 a month | 199th of 364, with limits $2,648 a month |
| CI runners 40 at once · 8 vCPU / 16 GB · 6,000 sessions of 12 min a month Both fit; STACKIT sells many dedicated vCPU sizes up to large machines, and boat starts a fresh VM in seconds with Docker ready. | 4th cheapest of 377 $86 a month | 130th cheapest of 377 $2,143 a month |
| 24/7 dev boxes 1 machine 24/7 · 4 vCPU / 8 GB Both fit a box that runs all month; on STACKIT a stopped but still allocated server keeps its resources reserved and billed, so deallocate it to stop paying for compute. | 18th cheapest of 368 $26 a month | 91st cheapest of 368 $131 a month |
| Computer-use agents 50 at once · 4 vCPU / 8 GB · 20,000 sessions of 15 min a month STACKIT has no desktop stream or computer-use API, so you would install a desktop and VNC yourself; boat streams a 60fps desktop with Chromium. | 6th cheapest of 369 $200 a month | 336th of 369, with limits $3,530 a month |
| Browser-use agents 50 at once · 2 vCPU / 4 GB · 30,000 sessions of 10 min a month | 7th cheapest of 367 $200 a month | 331st of 367, with limits $1,509 a month |
| Scraping & bots (startup scale) 50 at once · 1 vCPU / 2 GB · 100,000 sessions of 3 min a month | 6th cheapest of 346 $90 a month | 207th of 346, with limits $151 a month |
| Mass scraping & bots (undetectable) 1,000 at once · 1 vCPU / 2 GB · 3,000,000 sessions of 2 min a month | 71st of 338, with limits $1,800 a month | 111th of 338, with limits $3,017 a month |
| Parallel red teaming 200 at once · 2 vCPU / 4 GB · 5,000 sessions of 30 min a month | 2nd cheapest of 360 $200 a month | 224th of 360, with limits $257 a month |
| GPU eval jobs 10 at once · 8 vCPU / 32 GB · 500 sessions of 30 min a month Only STACKIT fits: it rents Nvidia A100, L40S and H100 machines, and boat has no GPUs. | Can't run it | 32nd cheapest of 98 $5,571 a month |
| macOS agents 5 at once · 4 vCPU / 8 GB · 1,000 sessions of 20 min a month | Can't run it | Can't run it |
Key features
| Feature | boat | STACKIT |
|---|---|---|
| What it is | Sandboxes for AI agents, each a full VM with its own kernel | General cloud VMs you set up and run yourself |
| Docker, root and nested virtualization | Docker, root, systemd and /dev/kvm on every machine | Root on your own VM, so Docker installs like on any Linux server; nested virtualization is not documented |
| CPU | Dedicated AMD Ryzen 9 9950X (Zen 5) hosts | Shared vCPU (overprovisioned) or dedicated vCPU sizes on Intel, AMD EPYC and Arm |
| Desktop and computer use | 60fps desktop and Chromium, computer use and browser control, embeddable stream | None; the portal web console is an emergency text console without copy and paste |
| Coding agents | Claude Code, Codex, OpenCode and more preinstalled, with a prompt, steer and events API | None built in; the Server Agent runs shell and PowerShell scripts on your servers |
| Keeping keys out of the machine | AI credentials proxy for model providers and any HTTPS API; switch a credential to another account when credits run out | Secrets Manager stores secrets that your VM reads, so the key still reaches the machine |
| Files after a stop | Snapshots every minute and on stop; a stopped sandbox costs nothing and keeps its disk | A stopped server stays reserved and billed; deallocate it to stop compute billing, and take disk snapshots by hand or on a backup schedule |
| Fork | Fork a running sandbox in one call; the copy boots from its disk | Clone the boot volume of a running server, then create a new server from it (disk only) |
| Reaching the machine | HTTPS URLs for up to 50 ports, token-protected; raw TCP; dedicated IPv6 or IPv4 | Attach a public IP and open ports in a security group; no managed HTTPS URLs |
| Outbound internet rules | Open internet on every plan; no provider-side allowlist | Security groups limit outbound traffic by protocol, port and IP range |
| GPUs | None | Nvidia A100, L40S and 8x H100 machine types |
| Windows and Arm | Linux x86 only; Wine for Windows apps | Windows Server images and Arm machine types |
| Regions | EU only: Germany, Finland, France | EU only: Germany and Austria |
| Compliance and transparency | SOC 2 is in progress. We're open about where we stand: live status and data retention policy. | SOC 2 |
Every feature Battleships tracks (43)
| Feature | boat | STACKIT |
|---|---|---|
| Full VM (own kernel) | Yes | Yes |
| Docker inside | Yes | Yes |
| Nested virtualization | Yes | Not published |
| Root | Yes | Yes |
| Systemd | Yes | Not published |
| Desktop GUI | Yes | No |
| Browser | Yes | No |
| Computer-use API | Yes | No |
| SSH | Yes | Yes |
| Persistent disk | Yes | Yes |
| Pause/resume | Yes | Yes |
| Automatic snapshots | Yes | No |
| Fork/clone | Yes | Yes |
| Live fork (no pause) | Yes | Yes |
| Memory fork | No | No |
| Custom image (Docker or snapshot) | Yes | Yes |
| ≥24 h sessions | Yes | Yes |
| Idle auto-stop | No | No |
| Wake on request | No | No |
| HTTPS preview URLs | Yes | No |
| Raw TCP inbound | Yes | Not published |
| Public IPv4 | No | Yes |
| Egress allowlist | No | Yes |
| Secret proxy | Yes | No |
| Secret proxy for any API | Yes | No |
| Preinstalled agents | Yes | No |
| Agent harness API | Yes | No |
| MCP server | No | No |
| GPU | No | Yes |
| Windows | No | Yes |
| Arm64 | No | Yes |
| Self-hosting | No | No |
| BYOC | No | No |
| Open source | No | No |
| Scoped API keys | Yes | Not published |
| Spend limits | Yes | Not published |
| Usage per sandbox (API) | Yes | Yes |
| Audit logs | No | Not published |
| SSO | No | Yes |
| SOC 2 | No | Yes |
| HIPAA | No | No |
| EU data residency | Yes | Yes |
| Zero data retention | Yes | Not published |
Performance
There is no comparable measurement for STACKIT: it was not part of the public benchmark run we use. boat did 26.69 runs per second in StarSling's Node.js build-and-test benchmark at 4 vCPU / 8 GB, on dedicated AMD Ryzen 9 9950X (Zen 5) hosts. STACKIT's speed depends on the family you pick, and the shared vCPU sizes in the scenarios below have no published baseline.
Node.js build-and-test loop at 4 vCPU / 8 GB, higher is better. Measured with the open hpc-sandbox-benchmarks suite. STACKIT has no published run in this suite yet, so there is no like-for-like number to compare.
Pricing in five scenarios
Each bill is priced from the providers' public price lists by the open Battleships engine, with the cheapest plan that runs the scenario.
| Scenario | boat, a month | STACKIT, a month |
|---|---|---|
| One coding agent, workdays 4 vCPU / 8 GB, 50 GB disk, 8 hours a day, 22 days a month, files kept between days | $20 · $20/mo | $34 · Pay as you go |
| One agent that never stops 4 vCPU / 8 GB, 50 GB disk, running 24/7 all month | $26 · $20/mo | $131 · Pay as you go |
| 50 agents in parallel 50 machines of 4 vCPU / 8 GB at once, 8-hour sessions, 1,100 sessions a month, Docker inside | $317 · $20/mo × 3 seats | $1,611 · Pay as you go |
| Sandboxes for your users 2 vCPU / 4 GB, 30,000 sessions of 10 minutes a month, up to 300 at once | $200 · $100/mo × 2 seats | $1,509 · Pay as you go · with limits |
| 1,000 short jobs at once 2 vCPU / 4 GB, 1,000 at once, 3 minutes each, 200,000 jobs a month | $2,000 · $2000/mo · with limits | $10,058 · Pay as you go · with limits |
boat: You pick a monthly plan from $20, and all of it becomes usage credit spent per second on running sandboxes. A stopped sandbox costs nothing and keeps its disk.
STACKIT: STACKIT bills each server per started hour at its net list price in euros, plus taxes, and bills disks, snapshots and public IPs on top. A stopped server keeps its resources reserved and billed until you deallocate it, and the scenarios below use its cheapest shared vCPU sizes.
Limits in those scenarios
STACKIT has per-project quotas for vCPU, memory, volumes and public IPs, but it does not publish the defaults for a new project, so 50, 300 or 1,000 servers at once are not confirmed until you ask. It also does not publish an egress price. Its hourly billing means each short job pays for a full hour. On boat the limit is a sandbox count (100 to 2,000 by plan) plus starts per minute, so the 1,000-job burst gets queued and needs more starts per day than the $2,000 plan allows.
| Scenario | boat | STACKIT |
|---|---|---|
| One coding agent, workdays | None that bite |
|
| One agent that never stops | None that bite |
|
| 50 agents in parallel | None that bite |
|
| Sandboxes for your users | None that bite |
|
| 1,000 short jobs at once |
|
|
Which one should you use?
Use STACKIT when
you need GPUs, Windows Server or Arm, or want plain long-lived servers next to managed databases and Kubernetes in a German or Austrian cloud.
- GPUs for model work. Useful if your agents train or test models: STACKIT rents Nvidia A100, L40S and H100 machines in the EU.
- Windows, Arm and many machine sizes. Useful if you need Windows Server, Arm, or very large or memory-heavy VMs from a single European cloud.
- Network rules and a full cloud around the VM. Useful if your servers must sit behind firewall rules next to managed databases, Kubernetes and storage in one EU cloud.
Use boat when
you run AI agents: full VMs with Docker and /dev/kvm, a desktop, built-in coding agents and snapshots every minute, all in the EU and ready in seconds.
- Coding agents you drive from outside. On STACKIT you install each agent, copy your API keys onto the VM and write the code that starts, prompts and watches it. A key on the VM can be read by any code the agent runs.
- A desktop and browser for computer use. An agent that clicks through web apps needs a screen. On STACKIT you install a desktop, a VNC server and a browser on every VM, then build the screenshot and click loop and its access control yourself.
- Snapshots every minute, free stops, one-call forks. Agents break things. Without automatic snapshots a bad command can lose hours of work, and without cheap forks you script volume copies and server creation every time you want to try two ideas side by side.
Questions people ask
Is boat cheaper than STACKIT?
In the scenarios we priced, yes: one coding agent on workdays comes to $20 a month on boat and $34 on STACKIT, and one agent running all month comes to $26 versus $131. 50 agents in parallel come to $317 versus $1,611. For 1,000 short jobs at once neither fits as published: boat lands at $2,000 and STACKIT at $10,058, mostly because it bills each started hour.
Can I run AI agent sandboxes on STACKIT?
Yes, as plain VMs you set up yourself. You install Docker, the agents and any browser, add your API keys, open ports and write your own code to start, stop and watch each server. boat gives you those parts ready on every sandbox.
Are both in the EU?
Yes. STACKIT runs only in its own data centers in Germany and Austria. boat runs only in the EU (Germany, Finland and France), with EU data residency and a zero data retention option.
Does boat have GPUs or Windows like STACKIT?
No. boat has no GPUs and no Windows guests; it runs Linux VMs on dedicated AMD Ryzen 9 9950X hosts and has Wine docs for Windows apps. If you need Nvidia GPUs or Windows Server, STACKIT has both.
What happens when I stop a server?
On STACKIT a stopped server keeps its resources reserved and billed; you deallocate it to stop paying for compute, while its disks stay and are still billed. On boat a stopped sandbox costs nothing, keeps its disk, and has snapshots from every minute it ran.
Is boat SOC 2 compliant?
SOC 2 is in progress. We're open about where we stand: live status at status.ascii.dev and our data retention policy at docs.boat.dev/data-retention. STACKIT says it holds BSI C5 Type 2, ISAE 3000 (SOC 2) and ISO/IEC 27001 certifications.
Start a sandbox
Install the CLI, then run boat new.
7-day free trial