AI Compute Gap Widens

The AI compute gap is a growing concern for enterprises, as they invest heavily in infrastructure without a clear understanding of the costs. According to VentureBeat AI research, 107 enterprises are spending more on AI infrastructure than they can measure, with most running their AI on hyperscalers and model-provider APIs. The majority plan to switch or add providers within the year, prioritizing integration and total cost of ownership over token price. However, with GPUs sitting at half utilization or less and few enterprises rigorously tracking compute costs, the compute gap is widening.