← All reports

Only 11% of businesses can accurately forecast their AI spending.

AIEconomicsOct 5, 2026score 0.442 posts · 0 replies across 2 instances
The thread discusses two main points: the difficulty businesses face in forecasting AI spending and the inefficiency of using cost per token as a metric for AI model performance. It highlights the unpredictability of AI costs and the potential pitfalls of relying on cheaper models.

Claims

Only 11% of businesses can accurately forecast their AI spending.
Parent: AIEntity: AI model cost predictionImpact: negativeDate: Oct 5, 2026Target: Ability of businesses to forecast AI spending
Cheaper AI models often cost more than higher-priced models on 32% of tasks.
Parent: AIEntity: AI model cost efficiencyImpact: negativeDate: Oct 5, 2026Target: Cost efficiency of cheaper AI models

Source posts

@[email protected]
Only 11% of companies in one survey could accurately forecast their AI spending. A separate study of 6,800+ tasks found cheaper AI models actually cost more 32% of the time. So dollars per token is becomes a less relevant measure than dollars per task. AI is like getting a contractor, you know the hourly rate but not how many hours the job will take. More details below including a cool interactive chart showing how a cheaper Gemini Flash spent $14 on a task and failed while the more expensive Pro finished for $1. https://www.wsj.com/tech/personal-tech/ai-token-spending-businesses-431ee94a?st=1trc2u&reflink=desktopwebshare_permalink
2 boosts · 3 favs · 0 replies · Oct 5, 2026
@[email protected]
Survey: only 11% of ~400 businesses could forecast AI spending; a study finds lower-priced models cost more than higher-priced models on 32% of 6,800+ tasks (Wall Street Journal) https://www.wsj.com/tech/personal-tech/ai-token-spending-businesses-431ee94a?st=1trc2u&reflink=desktopwebshare_permalink http://www.techmeme.com/261005/p15#a261005p15
1 boosts · 1 favs · 0 replies · Oct 5, 2026