
Non-Von is a Dartmouth-affiliated startup developing a novel parallel neuromorphic hardware architecture inspired by brain circuitry. Their technology combines custom algorithms with this hardware to reduce price and improve power performance by an order of magnitude, eliminating memory bottlenecks. They have successfully fabricated and deployed their proof-of-concept chip, "novo1", in a fielded commercial product for edge processing, with "novo2" planned for late 2026. Their solution is designed to optimize AI systems, particularly those that are sparse and quantized, outperforming traditional GPUs and TPUs on newer AI models and transformer applications. Their SCULPT toolset further optimizes customer models for their hardware, enabling significant cost savings and efficiency gains for AI applications from edge devices to datacenters.

Non-Von is a Dartmouth-affiliated startup developing a novel parallel neuromorphic hardware architecture inspired by brain circuitry. Their technology combines custom algorithms with this hardware to reduce price and improve power performance by an order of magnitude, eliminating memory bottlenecks. They have successfully fabricated and deployed their proof-of-concept chip, "novo1", in a fielded commercial product for edge processing, with "novo2" planned for late 2026. Their solution is designed to optimize AI systems, particularly those that are sparse and quantized, outperforming traditional GPUs and TPUs on newer AI models and transformer applications. Their SCULPT toolset further optimizes customer models for their hardware, enabling significant cost savings and efficiency gains for AI applications from edge devices to datacenters.