The era of “AI tokenmaxxing” may be coming to an end. Leaked audio from consulting firm Accenture reveals that many companies are struggling to measure the effectiveness of their AI investments due to excessive spending on tasks that don’t require AI, such as simple chatbots and converting PDFs into markdown files.
Accenture’s agentic AI strategy lead, Justive Kwak, stated that token spend is increasing exponentially as more people start using AI, but it’s proving difficult for companies to predict how much any task will cost. The consulting firm acknowledges that certain trivial tasks are causing massive token overspend, especially when agentic AI is part of the mix.
The problem has led some companies, like Amazon and Uber, to rethink their AI strategies and limit spending on unnecessary tasks. Some software developers have even resorted to using the “caveman” trick to reduce token spend. The major tech companies’ previous bullishness on AI use seems to be waning, and we may see more companies restrict access and spend through 2026 as they reassess their AI strategy.
The lack of clear ways to measure return on investment in AI is a major challenge. Large language models are proving useful in niche cases, but their effectiveness at broader tasks is nebulous. Without a way to quantify the value of AI investments, companies are struggling to justify their spending.
Accenture is now working on a tool called “Token IQ” to help advise clients on how to manage token economics. The consulting firm’s experience highlights the need for a more nuanced approach to measuring the effectiveness of AI investments and encourages companies to adopt more judicious use of AI tokens.
Source: https://www.tomshardware.com/tech-industry/artificial-intelligence/the-ai-tokenmaxxing-party-is-crashing-over-spiraling-costs-leaked-consulting-firm-audio-suggests-no-one-is-sure-how-to-measure-ai-effectiveness