OpenClaw on Raspberry Pi 5: Build a $100 24/7 Home AI Server with NVMe SSD (2026)

Why Raspberry Pi 5 Is Perfect for OpenClaw

The Raspberry Pi 5 is the most significant hardware upgrade in the Raspberry Pi Foundation's history. Released in 2024, it packs a quad-core Cortex-A76 processor running at 2.4GHz, up to 8GB of LPDDR4X RAM, and — crucially — the first native PCIe 2.0 interface ever included on a Raspberry Pi board. That PCIe slot changes everything for an OpenClaw home server build.

Previous Raspberry Pi models relied on microSD cards for storage, which were notoriously slow (15–30 MB/s read speeds) and prone to corruption under constant read/write workloads. OpenClaw generates significant I/O activity — it maintains persistent memory, writes conversation logs, caches skill data, and processes background cron jobs around the clock. On a microSD-based Pi 4, this would wear out the card within months. On a Pi 5 with an NVMe SSD attached via the PCIe slot, you get 400–600 MB/s read speeds and enterprise-grade reliability.

Consider the economics: running OpenClaw on a cloud VPS costs $5–25 per month, which adds up to $60–300 per year. A Raspberry Pi 5 with NVMe SSD runs on about 6W of electricity, costing roughly $4–6 per year in electricity (at average US rates). The hardware pays for itself in two to three months compared to cloud hosting. And unlike cloud VPS instances, your Pi 5 is physically in your home — your OpenClaw's memory, skills, and all conversation history stays entirely on your hardware, never uploaded to a third-party server.

Hardware Shopping List

You need five components. Every one of them is readily available from Amazon, Adafruit, PiShop, or local electronics retailers. The prices below reflect 2026 market rates.

Step 1 — Install Raspberry Pi OS (Bookworm 64-bit)

OpenClaw requires Node.js 22, which depends on a modern 64-bit OS. Raspberry Pi OS "Bookworm" (Debian 12-based, released 2024) is the official and best-supported choice. The 64-bit version is required for Node.js 22 compatibility.

Download and install the Raspberry Pi Imager on your main computer (Windows/macOS/Linux) from raspberrypi.com/software. You will use a microSD card as a temporary boot medium to flash the NVMe SSD, or you can use the USB-to-NVMe adapter method to write directly to the SSD.

In Raspberry Pi Imager, select Raspberry Pi 5 as the device, then choose Raspberry Pi OS (64-bit) — the Lite (no desktop) version is recommended for a headless server. Before writing, click the gear icon to configure: set a hostname, username, password, and enable SSH. This allows you to connect wirelessly without ever needing a keyboard and monitor attached to your Pi.

Once booted, your Pi 5 should appear on your home network. SSH in from your main computer:

Run the full system update before proceeding — this ensures you have the latest firmware and security patches:

Step 2 — Set Up NVMe SSD Boot

After assembling your M.2 HAT and NVMe SSD, the Pi 5 should detect the drive automatically. Verify it is recognized:

To boot from NVMe (recommended for performance), use the SD Card Copier tool in raspi-config or copy your OS to the NVMe using rpi-clone. Once copied, change the boot order in raspi-config to prioritize NVMe over SD:

Verify you are now booted from NVMe by confirming the root filesystem is on nvme0n1p2:

Step 3 — Install Node.js 22 via NVM

OpenClaw requires Node.js 22 (LTS). The package in the default Raspberry Pi OS repositories is outdated. Use nvm (Node Version Manager) to install the correct version:

Since nvm is user-specific, you also need to make Node.js accessible system-wide for the systemd service that will run OpenClaw at startup:

Step 4 — Install and Configure OpenClaw

Install OpenClaw globally using npm:

Run the onboarding wizard. This is an interactive setup process that configures your AI agent's name, connects it to your Claude API key, and sets up your preferred messaging channel (Telegram is most popular):

During the onboarding process you will be prompted for:

Test that OpenClaw runs correctly by starting it manually:

Step 5 — Create a Systemd Service for 24/7 Autostart

The key to a true 24/7 home server is making OpenClaw start automatically when the Pi boots and restart automatically if it ever crashes. On Raspberry Pi OS (Debian-based), the systemd init system handles this perfectly.

Create a new systemd service file:

Paste the following content (replace pi with your actual username if different):

Enable and start the service:

Reboot your Pi and verify OpenClaw starts automatically — send a message to your Telegram bot and confirm it responds. The Restart=always directive ensures that if OpenClaw ever crashes or encounters an unhandled error, systemd will automatically restart it within 10 seconds.

Step 6 — Remote Access and Port Configuration

Your OpenClaw agent on Pi 5 communicates with the outside world through Telegram's servers — you do not need to open any ports on your home router for basic Telegram-based access. The agent reaches out to Telegram's API (long-polling), so inbound connections from the internet are not required.

However, if you want to access the OpenClaw web interface or SSH into your Pi from outside your home network, you have several options:

For Tailscale setup (the simplest option), install it on your Pi:

Step 7 — Thermal Management and Performance Tuning

The Pi 5 runs warmer than its predecessors due to the more powerful CPU. Without proper cooling, it will throttle CPU speed when the temperature exceeds 80°C. Monitor your Pi's temperature regularly:

If you see temperatures consistently above 70°C, consider upgrading to a better cooled case. The Argon NEO 5 BRED with its integrated heatsink and fan keeps temperatures 15–20°C lower than passive cooling alone.

For power savings during off-peak hours, you can optionally configure the Pi to underclock slightly when idle. This reduces heat and power consumption without affecting OpenClaw's responsiveness — the agent spends most of its time waiting for incoming Telegram messages anyway. Edit /boot/firmware/config.txt to add arm_freq=1800 for a slight underclock, or leave it at the default 2400MHz for maximum performance.

Why Your Pi 5 OpenClaw Needs a VPN — and Why VPN07 Is the Right Choice

Here is a challenge that surprises many Pi 5 OpenClaw builders: your beautiful new home server is on a residential internet connection. Residential ISPs are notorious for several behaviors that directly harm OpenClaw's performance — including traffic shaping, rate-limiting connections to AI API providers, and IP reputation issues that cause Claude API requests to fail unpredictably.

Additionally, if you are running your Pi in a region where direct connections to Anthropic's Claude API are slow or unreliable, the experience degrades significantly. OpenClaw's cron jobs run on schedules — if an API call fails due to network issues, the job silently fails and your agent stops working as expected. This is frustrating when you have built an entire home server around the expectation of reliable 24/7 operation.

VPN07 solves residential ISP limitations by routing your Pi's API traffic through a dedicated 1000Mbps commercial network. This gives your little home server the same network quality as a data-center VPS, without the monthly cloud hosting fees. You keep your data on your Pi while benefiting from enterprise-grade connectivity.

Installing VPN07 on Raspberry Pi OS takes under two minutes. VPN07 supports Linux ARM64 (which is exactly what Pi 5 runs) with a lightweight command-line client that consumes minimal system resources:

VPN07's Linux client also supports split tunneling, allowing you to route only OpenClaw's API traffic through the VPN while keeping local network communication (Tailscale, SSH from within your home) on the direct connection. This prevents any performance overhead for non-AI traffic while ensuring rock-solid AI API connectivity.