AI's Biggest Bottleneck Is Cost: Companies Blame Stalled Projects on Overruns Three Times as Often as on Compute Shortages
VentureBeat, Tuesday, November 10th, 2026
A VentureBeat survey found 38% of respondents had AI projects delayed or cut over cost overruns, versus 12% citing compute shortages.
VentureBeat's October VB Pulse survey of 156 respondents found that 38% said their organization delayed, scaled back or canceled an AI project in the past year because it cost more than expected, compared with 12% who cited insufficient computing capacity.
Even so, 79% expect their AI budget to grow, while only 26% say spending is judged by a financial return they could defend to finance.
The article cites Uber reporting a 90% reduction in cost per feature-flag cleanup after benchmarking models against real tasks, and describes Tokenomics Foundation and FinOps Foundation work on measuring AI costs.
It also notes 56% plan to adopt a new or replacement AI compute platform within 12 months.