Simply adding more hardware is not a cure-all solution.

Data published in November by @CloudNativeFdn reveals that 54% of respondents managing on-prem GPU fleets experience average utilization levels of 25% or less. Even while the broader industry continues to invest heavily in brand-new gear, a substantial amount of existing computing power is currently operating well below its full potential.

Read the complete findings here: https://www.cncf.io/reports/virtual-machines-agentic-ai-and-the-mix-of-workloads/