Two vendors have released new servers this month, signaling the growing potential for AI workloads at the edge. Custom storage vendor Unigen has introduced a compact edge AI server called Cupcake, while Lenovo has released its ThinkEdge SE455 V3 server. Both servers aim to make trained AI models actionable at the edge. Lenovo’s server offers greater computing power with AMD EPYC 8004 CPUs and can support up to six GPUs. It is part of Lenovo’s TruScale infrastructure-as-a-service Opex model. Unigen’s Cupcake server is designed for low power, low latency, and high performance with an Intel Elkhart Lake 4-core Atom processor. It is suitable for computer vision use cases. As the demand for edge AI increases, hardware vendors are developing more complex devices. However, the actual adoption and deployment of these devices have been limited. Edge AI hardware addresses the limitations of cloud-based AI infrastructures, such as overdependence on stable network connectivity. Edge servers need to be resilient, energy-efficient, secure, and scalable to handle the challenging environments they are deployed in. Additionally, self-sustaining mechanisms in hardware could mitigate minor discrepancies without human involvement.
Growth observed in the Edge AI server market
