Matt Garman, AWS CEO, announced at re:Invent 2025 that AI inference now accounts for roughly two-thirds of all AI compute demand, up from one-third in 2023. He predicted that 80-90% of enterprise AI value will eventually come from inference-powered autonomous agents. AWS showcased its Trainium3 chips, purpose-built for inference workloads rather than training-optimized GPUs. This shift has significant implications for crypto miners who pivoted toward AI services after Bitcoin mining margins tightened following the April 2024 halving. Mining companies still offering generic GPU access for training may face declining demand as the industry moves toward inference-optimized hardware and lower-latency requirements.
Source: Read the original article
