Israeli fabless semiconductor company Neologic said on Monday it took part in VivaTech 2026, a European technology and startup exhibition recently held in Paris, and introduced next-generation AI server CPU technology.

According to the company, Neologic is focusing on improving semiconductor power efficiency and reducing inference costs among bottlenecks in AI infrastructure.

Neologic aims to enable CPUs to be used efficiently for AI inference as well as general-purpose computing. It targets lower power consumption, operating cost burdens and inference costs concentrated in GPU-centric infrastructure.

Neologic said that at VivaTech it met with officials from global and European cloud service providers, server makers and semiconductor OEMs to discuss potential future technology partnerships and commercialization cooperation.

Abby Meshica (애비 메시카), Neologic co-founder and CEO, said about 90 percent of current AI workloads are inference and data centers are facing new limits in terms of power consumption and operating costs. As a result, data center customers are depleting their annual AI budgets much faster than expected, he said. Neologic will help cloud service providers and their customers use cheaper and more sustainable AI inference solutions through cost-efficient and power-efficient AI server CPUs, he said.