Sentinel Core Puts a PCIe x16 Slot on Your Raspberry Pi CM5
Sanctuary Systems launched the Sentinel Core on Crowd Supply — a $129 open-hardware mini-ITX carrier board that gives the CM5 a full-size PCIe x16 expansion slot.
A Full-Size PCIe Slot for Your Compute Module 5
The Raspberry Pi Compute Module 5 is arguably the most versatile Pi ever built — powerful enough for serious applications, compact enough for embedded deployments. What it has always lacked is a direct path to desktop-class expansion hardware. The Sentinel Core from Sanctuary Systems addresses exactly that gap.
Launched on Crowd Supply on April 3, 2026, the Sentinel Core is a mini-ITX carrier board purpose-built for the CM5. The headline feature: a full PCIe x16 mechanical slot wired through the CM5's PCIe interface. Plug in a GPU for AI inference, a GPU for 3D rendering, a capture card, or a multi-port SSD controller — essentially any PCIe card that fits in a standard ATX enclosure.
Spec Rundown
The Sentinel Core delivers a solid I/O complement alongside the PCIe slot:
- 2x HDMI output ports for multi-display configurations
- Gigabit Ethernet
- 2x USB 3.0 ports
- MIPI DSI and CSI connectors for displays and camera modules
- Standard 40-pin GPIO header for full Pi compatibility
- 24-pin ATX power connector
The ATX power connector is a particularly thoughtful inclusion — it means you can use any standard PC power supply, making the board genuinely practical in a full-size mini-ITX case alongside a GPU. This is a real PCIe plus ATX setup, not a hobbyist proof of concept.
Open Hardware All the Way Down
Sanctuary Systems released the Sentinel Core as open-source hardware. The KiCAD design files — full schematics and PCB layout — are published on GitHub. Engineers who want to modify the design, source their own components, or audit the circuit have everything they need available.
The crowdfunding campaign on Crowd Supply has a $1,250 funding goal. The board itself is $129; a CM5 Wireless bundle (8GB/32GB) runs $195. Shipping is expected by end of August 2026.
What It Is Actually For
The most compelling application is GPU-accelerated AI inference. A CM5 paired with even a mid-range discrete GPU can run Stable Diffusion locally, serve quantized large language models, or handle real-time computer vision tasks at a price point impossible with comparable x86 hardware. The CM5's performance-per-watt profile makes it a solid choice for always-on AI workloads in homelab or edge environments.
Video transcoding, multi-display output, and NAS configurations with high-speed storage expansion are equally valid targets. The x16 slot's flexibility means the Sentinel Core's final form depends entirely on what its owner plugs in.
Sources: CNX Software (April 3, 2026), Crowd Supply (April 2026), Hackster.io (April 2026)