
AI INFRASTRUCTURE PORTFOLIO
Products for Enterprise AI
Deployment-ready infrastructure engineered for modern AI workloads, from accelerated compute to the network, storage and cooling systems around it.
AI INFRASTRUCTURE PORTFOLIO
Infrastructure built to work as one system.
Explore focused B2B product information for planning, integration and direct project consultation.
105 products
Rack ServersTFNEW ZH-4201-8 4U 8-GPU Flagship Compute Server
ZH-4201-8 4U 8-GPU flagship compute server: dual 4th/5th Gen Xeon processors, supporting 8 dual-width GPUs, 32 DDR5 DIMM slots (up to 8 TB) and 24 NVMe SSD bays, designed for large-model training and scientific computing.
Rack ServersTFNEW ZH-4201-4 4U 4-GPU High-Density Compute Server
ZH-4201-4 4U 4-GPU high-density compute server: dual 4th/5th Gen Xeon processors, supporting 4 dual-width GPUs, 7 PCIe 5.0 slots and 8 drive bays, designed for graphics rendering and academic/scientific simulation.
Liquid-Cooled WorkstationsTFNEW W297G5-Q Liquid-Cooled Workstation
W297G5-Q: an enterprise liquid-cooled workstation supporting 2x 4th/5th Gen Intel Xeon Scalable processors (up to 385W), with 16 DDR5 DIMM slots (up to 4 TB) and PCIe 5.0 support.
Liquid-Cooled WorkstationsTFNEW W297G4-Q Liquid-Cooled Workstation
W297G4-Q: an enterprise liquid-cooled workstation supporting 2x 3rd Gen Intel Xeon Scalable processors (IceLake-SP, TDP 270W), with 16 DDR4 RECC memory slots (up to 2 TB) and support for 4 liquid-cooled GPU cards, with cold-plate liquid cooling.
Liquid-Cooled WorkstationsTFNEW W197G5-Q Liquid-Cooled Workstation
W197G5-Q: an enterprise liquid-cooled workstation powered by an Intel Xeon Sapphire Rapids workstation series processor, with 8 DDR5 ECC memory slots and support for 3 liquid-cooled GPU cards (NVIDIA 4090, L20, A6000, etc.).
Industrial Touch DisplaysTFNEW UTC-2105D Industrial Touch Monitor
UTC-2105D: an enterprise-grade industrial touch monitor with a 1920x1080 TFT-LCD display, 300 cd/m² brightness, fanless passive cooling, and IP65 front panel protection.
Industrial Touch DisplaysTFNEW UTC-1900 Industrial Touch Monitor
UTC-1900: an enterprise-grade industrial touch monitor with a 1280x1024 TFT-LCD display, 250 cd/m² brightness, five-wire resistive / 10-point capacitive touch screen, and IP65 front panel protection.
Industrial Touch DisplaysTFNEW UTC-1506 Industrial Touch Monitor
UTC-1506: an enterprise-grade industrial touch screen monitor with a 15.6-inch 1920x1080 TFT-LCD display, five-wire resistive / 10-point capacitive touch screen, and IP65 front panel protection.
Industrial Touch DisplaysTFNEW UTC-1501 Industrial Touch Monitor
UTC-1501: an enterprise-grade industrial touch monitor with a 1024x768 TFT-LCD display, 350 cd/m² brightness, five-wire resistive / 10-point capacitive touch screen, and IP65 front panel protection.
Compute ModulesTFNEW S230I2 AI Compute Module
S230I2: an AI compute module with 20 TOPS INT8 AI compute (10 TFLOPS FP16), 12 GB LPDDR4X memory, H.264/H.265 hardware decode and JPEG encode/decode, in a wide-temperature design (-20°C ~ +85°C).
Compute ModulesTFNEW KP920Lite Compute Module
KP920Lite Compute Module: a compute module powered by the KP920S processor (16/24 cores, 2.6 GHz) with 2-channel DDR4 memory, 3 PCIe4.0 controllers (up to 30 lanes) and 140G direct network.
AI Accelerator CardsTFNEW Atlas300VPro Video Analytics Card
Atlas300VPro Video Analytics Card: an AI video analytics card supporting real-time analysis of 128 HD video streams, with 140 TOPS INT8 AI compute and 48 GB LPDDR4X memory, suitable for smart city and smart transportation scenarios.
AI Accelerator CardsTFNEW Atlas300V Video Analytics Card
Atlas300V Video Analytics Card: an AI video analytics card with 140 TOPS INT8 AI compute, 48 GB LPDDR4X memory and hardware video codec, suitable for smart city, smart transportation, smart finance and smart campus scenarios.
AI Accelerator CardsTFNEW Atlas300IDuo Inference Card
Atlas300IDuo Inference Card: an AI inference card powered by the Ascend 710 chip (140 TOPS INT8, 70 TFLOPS FP16) with 24 GB LPDDR4X memory, H.264/H.265 video codec and a PCIe Gen4.0 interface.
AI Accelerator CardsTFNEW Atlas300IPro Inference Card
Atlas300IPro Inference Card: an AI inference card powered by the Ascend 710 chip (140 TOPS INT8, 70 TFLOPS FP16) with 24 GB LPDDR4X memory, ideal for OCR, speech recognition, search ranking and content moderation scenarios.