The Biscotti E1.S AI Module uses up to two Hailo-8 Edge AI processors, delivering up to 26 tera-operations per second (TOPS) each for a total of 52 TOPS while consuming as little as 10 watts. Built on an architecture optimized for neural networks, the Hailo-8 processors enable edge devices to run deep learning applications efficiently, effectively, and sustainably. Designed in the E1.S form factor, the module can be inserted directly into standard E1.S SSD slots to instantly add AI acceleration to larger server systems. This enables scalable deployments ranging from parallel neural networks processing large camera arrays to combined configurations supporting large AI workloads such as LLM inference. Compared to GPU modules or traditional add-in cards, the Biscotti E1.S delivers high AI performance with significantly lower power consumption, helping improve data center power efficiency. The module also includes a robust software suite with out-of-the-box support for state-of-the-art deep learning models, along with a comprehensive dataflow compiler that allows developers to quickly and easily port their neural network models.
