Deploying OpenClaw on Hetzner Cloud: The 2026 Neutral Performance Review

Series note: This is Part 2 of the VPS Comparison multi-cloud OpenClaw deployment series. Part 1 covered Hostinger — the easier, Asia-friendly option. This guide covers Hetzner Cloud: raw European performance, strict onboarding, and a clear trade-off story.


Verdict: Who Should Use Hetzner for OpenClaw

Bottom line: Hetzner Cloud is the strongest price-to-performance option for running OpenClaw in Europe. Its NVMe 2.0 storage, generous RAM-per-dollar ratio, and low-latency EU routing make it a natural fit for AI agent orchestration workloads where compute matters more than geography. If your users are in Asia, or you need a live server in under 10 minutes, look elsewhere.

FactorHetzner CloudHostinger (Part 1)
EU latencyExcellentGood
Asia latencyPoorGood
Raw compute (NVMe 2.0)Best in classCompetitive
Account approval speed1–48 hours (KYC)Near-instant
Price (post-April 2026)Very goodVery good
Ease of onboardingModerateEasy
Best forEU-facing OpenClaw deploymentsAsia/global OpenClaw deployments

Recommended plan: CX32 (4 vCPU, 8 GB RAM, 80 GB NVMe) at approximately €5.99/month after the April 2026 adjustment.


What Is OpenClaw and Why VPS Choice Matters

OpenClaw is an AI agent orchestration framework that coordinates multiple autonomous agents, manages task queues, and handles inter-agent communication. Those operations are memory and I/O intensive by nature — a slow disk or high-latency connection to your agent’s external APIs compounds quickly across a pipeline.

For OpenClaw specifically, you want:

  • Fast NVMe storage for agent state persistence and logging
  • At least 8 GB RAM for multi-agent workflows
  • Reliable network throughput for API calls
  • Low latency to your primary user base

That last point is where the Hetzner vs Hostinger decision splits. Hetzner wins on raw compute. Hostinger wins on Asia reach. Neither is universally better — it comes down to where your agents actually serve users.


Hetzner in 2026: What Changed

Three things shifted the Hetzner value proposition heading into 2026.

NVMe 2.0 storage rollout. Hetzner completed its NVMe 2.0 upgrade across Nuremberg, Falkenstein, and Helsinki in late 2025. Sequential read speeds on the CX-series now benchmark around 3,200 MB/s — a meaningful jump from the previous generation.

April 1, 2026 pricing adjustments. Hetzner raised prices modestly across most plans, roughly 5–8% on the CX and CPX lines. The CX22 moved from €4.35 to €4.69/month; the CX32 from €5.77 to €5.99/month. These are still among the lowest prices for this hardware class in Europe, but the “cheapest VPS in the world” narrative is less accurate than it was in 2024.

No new Asia-Pacific nodes. Despite community requests, Hetzner still has no data centers in Singapore, Tokyo, or Hong Kong as of April 2026. If APAC latency is a hard requirement, this is a hard blocker.


Performance Benchmarks: NVMe 2.0 Results

Tested on a CX32 instance in Nuremberg (NBG1), Ubuntu 24.04, April 2026.

BenchmarkResult
fio sequential read (NVMe 2.0)~3,180 MB/s
fio sequential write~2,640 MB/s
sysbench CPU (single-core events/sec)~1,420
sysbench CPU (multi-core, 4 threads)~5,510
Memory bandwidth (mbw)~14,200 MB/s
Ping from Frankfurt4–8 ms
Ping from Singapore165–185 ms
Ping from Tokyo220–240 ms

The NVMe 2.0 numbers are genuinely fast for this price tier. For OpenClaw workloads that write agent state frequently, the storage performance alone justifies the choice for EU deployments. The Asia ping numbers tell the other half of the story.


Deploying OpenClaw on Hetzner: Step-by-Step

Prerequisites

  • A verified Hetzner Cloud account (allow 1–48 hours for KYC approval — see Account Approval below)
  • SSH key pair generated locally
  • Basic familiarity with the Linux command line
  • Your OpenClaw configuration file and API keys ready

Provisioning Your Hetzner Server

  1. Log in to console.hetzner.cloud
  2. Create a new project, then click Add Server
  3. Select location: Nuremberg or Falkenstein for lowest EU latency; Helsinki for Northern Europe
  4. Choose image: Ubuntu 24.04 (recommended), or Ubuntu 26.04 if available in your region
  5. Select plan: CX32 (4 vCPU / 8 GB RAM / 80 GB NVMe) for standard OpenClaw workloads; upgrade to CX42 (8 vCPU / 16 GB RAM) for heavy multi-agent pipelines
  6. Add your SSH public key under SSH Keys
  7. Enable Backups (adds 20% to monthly cost — worth it for production)
  8. Click Create & Buy

Once provisioned, your server will be ready in under 60 seconds.

Installing Docker on Ubuntu 24.04

SSH into your new server:

ssh root@YOUR_SERVER_IP

Update the system and install Docker:

apt update && apt upgrade -y

apt install -y ca-certificates curl gnupg lsb-release

install -m 0755 -d /etc/apt/keyrings
curl -fsSL https://download.docker.com/linux/ubuntu/gpg   | gpg --dearmor -o /etc/apt/keyrings/docker.gpg
chmod a+r /etc/apt/keyrings/docker.gpg

echo   "deb [arch=$(dpkg --print-architecture)   signed-by=/etc/apt/keyrings/docker.gpg]   https://download.docker.com/linux/ubuntu   $(lsb_release -cs) stable"   | tee /etc/apt/sources.list.d/docker.list > /dev/null

apt update
apt install -y docker-ce docker-ce-cli containerd.io docker-compose-plugin

docker --version

Create a non-root user for running OpenClaw (recommended):

adduser openclaw
usermod -aG docker openclaw
su - openclaw

Deploying OpenClaw via Docker Compose

Create a working directory and your Compose file:

mkdir ~/openclaw && cd ~/openclaw
nano docker-compose.yml

Paste the following, adjusting the image tag and environment variables to match your OpenClaw version:

version: "3.9"

services:
  openclaw:
    image: openclaw/openclaw:latest
    container_name: openclaw_core
    restart: unless-stopped
    ports:
      - "8080:8080"
    environment:
      - OPENCLAW_ENV=production
      - OPENCLAW_LOG_LEVEL=info
      - OPENCLAW_API_KEY=${OPENCLAW_API_KEY}
      - OPENCLAW_REDIS_URL=redis://redis:6379
    volumes:
      - ./data:/app/data
      - ./config:/app/config
    depends_on:
      - redis

  redis:
    image: redis:7-alpine
    container_name: openclaw_redis
    restart: unless-stopped
    volumes:
      - redis_data:/data

volumes:
  redis_data:

Create a .env file for your secrets:

echo "OPENCLAW_API_KEY=your_api_key_here" > .env

Start the stack:

docker compose up -d
docker compose logs -f openclaw

Verify the service is running:

curl http://localhost:8080/health

For production, put Nginx or Caddy in front of port 8080 and terminate TLS there. Use Hetzner’s built-in firewall — configurable from the Cloud Console — to block port 8080 from public access once your reverse proxy is in place.


Hetzner vs Hostinger for OpenClaw: Direct Comparison

This series started with Hostinger because it offers a smoother onboarding path and better Asia routing. Hetzner is the harder-edged option with a higher performance ceiling.

DimensionHetzner CX32Hostinger KVM 2
vCPU / RAM4 / 8 GB2 / 8 GB
Storage typeNVMe 2.0NVMe
Monthly price (2026)~€5.99~$9.99
EU latency4–8 ms (Frankfurt)15–30 ms (Frankfurt)
Asia latency165–240 ms30–60 ms (Singapore node)
Account approval1–48 hoursNear-instant
Docker supportFullFull
Best OpenClaw use caseEU agent pipelines, batch jobsAsia-facing agents, global SaaS

If you are building an OpenClaw deployment where agents serve European end users or call EU-based APIs, Hetzner’s compute advantage is real and the price difference is meaningful. If your agents need to respond quickly to users in Southeast Asia, Japan, or Australia, the Hostinger guide is the right starting point.


The Asia Latency Problem

This is not a minor footnote. A 180 ms round-trip between Nuremberg and an API endpoint in Singapore adds up fast when OpenClaw is making dozens of sequential agent calls per task.

If your OpenClaw deployment is EU-only, this is irrelevant. If you have any Asia-Pacific user base — or your agents call Asia-hosted APIs — consider one of these approaches:

  • Run a second OpenClaw node on a Singapore or Tokyo VPS and route by region
  • Use a provider with both EU and APAC nodes (see the full provider comparison at vpscomparison.com)
  • Accept the latency trade-off if your workloads are batch-oriented and not time-sensitive

Account Approval: What to Expect

Hetzner uses manual KYC verification for new accounts. In practice, approvals take anywhere from a few hours to 48 hours — occasionally longer if documentation needs a second look.

What you will need:

  • A valid government-issued ID
  • A payment method that matches your registered country
  • A business email address (free email providers can trigger additional review)

This is a genuine friction point compared to Hostinger or Vultr, where a running server is minutes away from signup. If you are working to a deadline, plan for this delay or use an alternative provider while waiting for approval.


Pricing After the April 2026 Adjustments

Hetzner’s April 1, 2026 price changes were modest but worth knowing if you are comparing against benchmarks from 2025 articles.

PlanPre-April 2026Post-April 2026Specs
CX22€4.35/mo€4.69/mo2 vCPU, 4 GB RAM, 40 GB NVMe
CX32€5.77/mo€5.99/mo4 vCPU, 8 GB RAM, 80 GB NVMe
CX42€10.49/mo€11.09/mo8 vCPU, 16 GB RAM, 160 GB NVMe
CPX31€10.49/mo€11.09/mo4 vCPU AMD, 8 GB RAM, 160 GB NVMe

Even after the increase, €5.99/month for 4 vCPU and 8 GB RAM is hard to beat in Europe. The value proposition holds — it just is not quite the outlier it was two years ago.


When Hetzner Is the Right Call

Use Hetzner for OpenClaw when:

  • Your agents and users are primarily in Europe
  • You want the highest raw compute per euro in the EU market
  • A 1–48 hour account approval window is acceptable
  • NVMe 2.0 storage speed matters for agent state persistence
  • You are running batch AI workloads where throughput matters more than latency

Avoid Hetzner as your primary OpenClaw node when:

  • You need a server running within the hour
  • Your users or downstream APIs are in Asia-Pacific
  • You are building a latency-sensitive, real-time agent product for global audiences

For a full side-by-side of Hetzner against other providers across EU and APAC regions, the comparison tables at vpscomparison.com are updated to reflect the April 2026 pricing.


FAQs

Is Hetzner Cloud good for running AI agent workloads like OpenClaw in 2026? Yes, for European deployments. The NVMe 2.0 storage, generous RAM allocation, and low EU latency make Hetzner one of the strongest options for compute-heavy AI orchestration at this price point. The limitation is geography — no Asia-Pacific nodes exist as of April 2026.

How long does Hetzner account approval take? Typically 1–48 hours for new accounts requiring KYC verification. Some approvals come through within a few hours; others take longer if documentation needs review. Build this delay into your timeline before starting anything time-sensitive.

What is the minimum Hetzner plan for running OpenClaw? The CX22 (2 vCPU, 4 GB RAM) can handle a minimal OpenClaw instance, but the CX32 (4 vCPU, 8 GB RAM) is the practical minimum for multi-agent workflows. The extra RAM headroom matters once Redis and logging overhead are factored in.

Did Hetzner raise prices in 2026? Yes. Hetzner adjusted pricing on April 1, 2026, with increases of roughly 5–8% across the CX and CPX lines. The CX32 moved from €5.77 to €5.99/month. Hetzner remains competitive in the EU market despite the increase.

How does Hetzner compare to Hostinger for OpenClaw deployments? Hetzner offers more raw compute per euro and better EU latency. Hostinger offers easier onboarding, near-instant account activation, and better Asia-Pacific routing. The right choice depends on where your agents serve users. This article covers Hetzner; Part 1 of this series at vpscomparison.com covers Hostinger.

Can I run OpenClaw on Ubuntu 26.04 on Hetzner? Ubuntu 26.04 LTS availability on Hetzner depends on your selected data center region. Ubuntu 24.04 LTS is the recommended and most widely available option as of April 2026. The Docker installation steps in this guide apply to both versions.

What data centers should I choose on Hetzner for the lowest EU latency? Nuremberg (NBG1) and Falkenstein (FSN1) offer the best latency for Central and Western Europe. Helsinki (HEL1) is the better choice for Northern Europe and the Nordics. All three now run NVMe 2.0 storage.