Signal Only about 20% of infrastructures are said to be upgraded for AI workloads

Summary

NTT DATA reports, based on Everest Group research, that only about 20 percent of infrastructures have been upgraded to handle AI workloads. The figure appears next to the finding that most generative AI proofs of concept fail because of infrastructure gaps. Together, the two numbers describe a wide readiness gap between AI ambitions and the systems meant to run them. The report says informed infrastructure is strongest where latency, sovereignty and resilience matter, naming manufacturing, the public sector, energy, telecoms, finance and healthcare. It expects hybrid cloud and edge architectures and sovereign cloud zones to spread quickly in the near term. The low upgrade share implies substantial investment ahead before AI can operate at scale.