The growing cost of AI adoption is forcing companies to rethink how they use artificial intelligence across their organizations. For the past two years, businesses raced to integrate AI into coding, customer support, research, productivity, and everyday workflows. But as usage expanded, so did the costs.
Companies including Uber, Meta, Microsoft, Salesforce, and DoorDash are now reportedly tightening controls around AI spending as token consumption and computing expenses continue to rise.
Turns out “use AI for everything” sounds a lot better before the monthly invoice arrives.
TL;DR
- Major companies are starting to limit AI usage as costs rise.
- Uber, Meta, Microsoft, Salesforce, and DoorDash are reportedly tightening AI spending.
- Growing token consumption is driving larger AI bills across the industry.
- Companies are shifting from AI experimentation to more measured deployment.
- The AI boom continues, but businesses are becoming more selective about where AI delivers value.
AI Tokens Are Driving Up Costs
A major contributor to rising expenses is the growing consumption of AI tokens, the units used to measure activity within AI systems.
As employees rely more heavily on advanced AI models, companies are paying significantly more for computing resources.
Google recently revealed that its systems now process more than 3.2 quadrillion AI tokens every month, highlighting how rapidly AI workloads are expanding across the industry.
Companies Are Starting to Tighten Controls
Reports indicate that companies including Uber, Meta, Microsoft, Salesforce, and DoorDash are introducing stricter controls around AI usage.
Some organizations have reportedly exhausted annual AI budgets within months, while others are steering employees toward cheaper AI tools or monitoring whether AI spending is producing measurable business results.
The goal is no longer maximizing AI usage. The goal is maximizing useful AI usage.
The End of Tokenmaxxing?
The shift follows a period many insiders described as tokenmaxxing, where employees were encouraged to use AI as much as possible as companies rushed to demonstrate AI adoption.
In some cases, expensive AI models were used for simple tasks, generating large costs without delivering proportional value.
Now, executives are increasingly focusing on outcomes rather than token counts.
Apparently, AI usage charts look less impressive when they’re attached to a budget spreadsheet.
AI Isn’t Slowing Down
Despite the growing focus on costs, this does not signal a retreat from AI.
Companies continue to invest heavily in AI infrastructure, data centers, chips, and automation systems. The difference is that businesses are moving from experimentation to optimization.
The AI boom is still very much alive. Companies are simply learning that scaling AI and paying for AI are two very different challenges.