Fanless server HQ-Box2 AI

GPUAI edgeembedded

Rugged Fanless server for AI applications Specially crafted Quad M.2-M array for SSDs and/or AI accelerators with optional fans. Including the motherboard, the total number of M.2-M slots is 5. 8-Lane PCIe G4 FHHL DW slot for GPU or AI accelerator 8x SFP+ with up to 100Gb Ethernet throughput AST2600-based BMC with Display Port (DP) video On-board storage: M.2 NVMe + M.2 SATA + eMMC (soldered) The power demands of M.2 AI accelerators reach the limits of the M.2 standard and generate significant heat. Compounding this issue is the limited operating temperature of some of these modules. More often than not, you will need more than one AI accelerator, further exacerbating the problem of heat dissipation. The “HQ-Box2 AI” model addresses these challenges with a custom housing for four M.2 AI accelerators, each featuring a dedicated heat sink and a fan tray to handle the most demanding scenarios. Users can also mix M.2 AI accelerators and M.2 SSDs within the same enclosure. Additionally, an 8-lane PCIe G4 slot (full height, half length, double width) enables the addition of a GPU or another AI module, offering enhanced flexibility and performance.