Intel-based UP AI Development Kits - Part 1: Unboxing and First Boot to Ubuntu Pro 24.04
AAEON's UP lineup lands three x86 options for edge AI work: the credit card-sized UP TWL, the UP Squared Pro TWL with M.2 expansion and a Hailo-8L, and the UP Xtreme ARL powered by a 14-core Intel Core Ultra 5 225H. Together, they cover entry-level, mid-range, and high-end needs for developers building AI at the edge.
All three ship with Ubuntu Pro 24.04 and support the UP AI Toolkit and Network Optix's Nx Meta platform-useful if you're building vision-driven applications. This first look focuses on what's in the box and a quick first boot.
What's in the box
UP TWL - Entry-level AI dev kit
- Intel Processor N150 (Twin Lake), 8GB RAM, 64GB eMMC
- 12V/5A power adapter (standard DC barrel), four standoffs, safety manual
- USB VGA camera
Bottom side: 40-pin GPIO header, 10-pin wafer (USB 2.0 + UART), 4-pin front wafer, 64GB Kingston eMMC, RTC battery.
Top side: Intel N150 with active fansink for cooling. A fanless sibling (UP TWLS) exists, but not as part of this kit.
Ports: 1x GbE, HDMI, 3x USB 3.0. No onboard AI accelerator and no M.2 socket, so AI runs on the N150's CPU/GPU. A USB 3.0 AI accelerator could be added separately.
UP Squared Pro TWL - Mid-range with Hailo-8L
- Intel Processor N150, 8GB RAM, 64GB eMMC
- Hailo-8L M.2 AI accelerator (13 TOPS) and small heatsink
- 12V/6A power adapter with lockable DC jack, four standoffs + two nuts, safety manual
- USB VGA camera
Bottom side: 40-pin GPIO, multiple wafer headers, MIPI CSI, RTC battery, SATA, M.2 3052 B-key with SIM slot (cellular).
Top side: active cooler, M.2 2280 M-key and M.2 2230 E-key for storage, AI accelerators, and/or Wi-Fi.
Ports: 2x GbE, DisplayPort, HDMI, 2x USB 3.2 Type-A, 1x USB 3.2 Type-C, lockable DC, power button. The platform can scale up to 214 TOPS with a Metis accelerator (not included in this configuration).
UP Xtreme ARL - High-end with Intel Core Ultra 5 225H
- Intel Core Ultra 5 225H (14 cores), 16GB RAM
- 256GB M.2 NVMe SSD (pre-fitted)
- 19V/6.32A power adapter with lockable DC jack, four standoffs + nuts, safety manual
- USB VGA camera
Bottom side: 40-pin GPIO, 3.5mm audio jack, wafer headers, MIPI CSI, RTC battery, M.2 2230 E-key, dual M.2 2280 M-key (PCIe Gen4 x4), M.2 3052 B-key with Nano SIM slot. One M.2 M-key slot is populated with a 256GB Kingston SSD (the E-key sits underneath).
Ports: 1x GbE, 1x 2.5GbE, 2x HDMI, 1x DisplayPort, 2x USB 3.2, lockable DC, power button.
AI capability: up to 83 TOPS via the SoC's CPU, GPU, and Intel AI Boost. Additional third-party accelerators can be added via M.2.
Note: Ubuntu Pro 24.04 may require a license for full support. Details are available from Canonical's documentation. Ubuntu Pro overview
Hardware prep
Setup is minimal: install the standoffs, mount the boards, and connect power. The UP Squared Pro TWL includes the Hailo-8L; attach the small heatsink to the module before use.
Power cords are not included but use a standard PC PSU cable, which most offices already have on hand.
First boot: Ubuntu Pro 24.04
For a quick check, the UP Xtreme ARL was connected to HDMI, USB (touch display), and a wireless keyboard/mouse dongle. The unit powered on and booted into Ubuntu Pro 24.04 without issue.
Default credentials are straightforward: username "devkit", password "devkit". After login, you can proceed with network setup, updates, and installing the UP AI Toolkit or your preferred runtime stack.
Which kit fits your use case?
- UP TWL (Entry): Best for lightweight inference on CPU/GPU, headless services, simple sensors, and cost-sensitive deployments. Add a USB accelerator if needed.
- UP Squared Pro TWL (Mid): Adds M.2 expansion, dual LAN, and the included Hailo-8L (13 TOPS). Solid choice for compact vision projects and edge gateways.
- UP Xtreme ARL (High): For on-device inference with higher throughput, multiple displays, and NVMe storage. Good for multi-stream analytics and local model execution.
Pricing
- UP TWL AI Dev Kit: $279
- UP Squared Pro TWL AI Dev Kit: starting at $469
- UP Xtreme ARL AI Dev Kit: $899
Configurations can be extended with additional M.2 accelerators and accessories. Check current SKUs for RAM, storage, and module options.
What's next
Upcoming parts will validate features on Ubuntu Pro 24.04, run benchmarks, and focus on AI performance. Expect tests with camera inputs, framework runtimes, and potentially Nx Meta for IP video use cases.
Want to sharpen your AI build skills?
If you're mapping workloads to hardware and need practical training, browse curated programs by role. Courses by job
Your membership also unlocks:
