LattePanda Sigma 2 Debuts as a Compact Intel Panther Lake Single-Board Server for Local AI Workloads
LattePanda has unveiled the Sigma 2, a compact single-board computer designed to deliver workstation-like performance in a small form factor. Positioned as a “single-board server,” the new SBC targets developers, AI researchers, homelab users, edge computing projects, and anyone looking to build a powerful mini server without moving to a full-size desktop system.
At the heart of the LattePanda Sigma 2 is Intel’s Core Ultra X7 358H, a processor from the Panther Lake family. It is paired with Intel Arc B390 integrated graphics, giving the board enough graphical and compute capability for demanding local workloads. According to LattePanda, the system can reach up to 180 TOPS of INT8 AI performance, making it a strong candidate for on-device AI processing, local large language model experiments, computer vision tasks, automation systems, and compact edge AI deployments.
Memory is another major highlight. The Sigma 2 supports up to 64GB of LPDDR5X RAM, although LattePanda has not yet detailed every available memory configuration. On the highest-end model, up to 59GB of that memory can reportedly be assigned as VRAM for the integrated GPU. That flexibility could be especially useful for AI workloads, graphics-heavy tasks, and applications that benefit from a large shared memory pool.
Expansion is where the Sigma 2 becomes particularly interesting. Despite its compact design, the board includes two Thunderbolt 4 ports and one PCIe 4.0 x2 OCuLink port. These options open the door to high-speed external storage, external GPUs, accelerator cards, and other performance-focused add-ons. LattePanda describes this as “server-class” expansion, and the connectivity certainly gives the Sigma 2 more flexibility than many typical single-board computers.
Networking is also built with performance in mind. The board features dual 10GbE ports, making it suitable for fast file transfers, network-attached storage projects, virtualization labs, AI model serving, and compact server builds. There is no built-in wireless connectivity, but users can add Wi-Fi and Bluetooth through the onboard M.2 E-key slot.
Storage expansion should be generous as well. The Sigma 2 includes three M.2 M-key slots, giving builders room to install multiple SSDs for fast local storage. This makes the board appealing for users who want to create a compact NAS, development server, media server, or AI data processing machine.
Other features include GPIO support through an RP2350 microcontroller, two USB 3.2 Gen 2 Type-A ports, two USB 2.0 Type-A ports, and an active cooling solution to help manage thermals under sustained workloads. The active cooler is an important addition, especially given the board’s high-performance processor and AI-focused ambitions.
During the pre-sale period, the LattePanda Sigma 2 starts at $2,599. Its regular price is listed at $2,799, so early buyers can get a $200 discount. While this is far from a budget single-board computer, the Sigma 2 is clearly aimed at users who need serious performance, strong expandability, high-speed networking, and local AI computing power in a compact package.
For developers and enthusiasts searching for a powerful Intel Panther Lake SBC, the LattePanda Sigma 2 stands out as one of the more ambitious options in the compact computing space. Its combination of AI performance, dual 10GbE networking, Thunderbolt 4, OCuLink, and multiple M.2 slots makes it a compelling choice for advanced homelab setups, edge AI systems, and small but powerful server builds.
