Critical Need
The rapid growth of deep learning has resulted in a corresponding increase in the computing demands of AI applications, which are doubling every two months. In contrast, the performance improvements of digital processors are diminishing, doubling every three years. At this rate, training the state-of-the-art neural network in 2025 is expected to cost more than $1 billion and consume more power than all of New York City. This cost-prohibitive nature of training complex AI models limits developments in critical industries including pharmaceuticals, automotives, and defense.

Technology Vision
Eva is developing high-performance, general-purpose analog AI hardware for large-scale training applications, as well as advanced algorithms that enable accurate computation with these processors. The company is pursuing an end-to-end development philosophy to co-optimize the intrinsic physics of a novel processor with the algorithmic mathematics of linear algebra operations to achieve unprecedented speed and energy-efficiency for deep learning.

Potential for Impact
Eva will use its private data centers to train and license complex, energy-efficient AI models far beyond incumbents' current reach. As a result, these systems will enable a broad range of strategic applications including drug discovery, threat monitoring, autonomous systems, fraud detection, demand forecasting, smart agriculture, and weather intelligence.

Website
Eva