Last Updated:
Uber, Meta, Microsoft, Salesforce and DoorDash curb AI use as soaring token costs blow budgets, shifting from tokenmaxxing to stricter, results driven deployment.
Uber, Meta, Microsoft, Salesforce and DoorDash curb AI use as soaring token costs blow budgets, shifting from tokenmaxxing to stricter, results driven deployment.
For the past two years, companies across the United States rushed to embrace artificial intelligence, encouraging employees to use AI tools for everything from coding and research to customer service and productivity. But now, some of America’s biggest corporations are beginning to slow down and place limits on AI usage as costs spiral far beyond expectations.
According to a report by The Wall Street Journalexecutives at major firms including Uber, Meta, Microsoft, Salesforce and DoorDash are increasingly trying to control AI spending after seeing usage costs surge dramatically. Some companies reportedly exhausted annual AI budgets within just a few months, while others saw expenses double or even triple as employees embraced AI tools at scale.
The biggest issue is the soaring cost of so-called AI tokens — the units used to measure computing activity inside AI systems. As millions of workers rely more heavily on AI models, demand for computing power has exploded, pushing costs higher across the industry. Google recently revealed it now processes more than 3.2 quadrillion AI tokens every month, roughly seven times more than a year ago.
In response, companies are beginning to ration access to expensive AI models, direct workers toward cheaper internal tools and monitor whether AI spending is actually producing measurable business results. Meta Chief Technology Officer Andrew Bosworth recently warned employees that using AI simply for the sake of appearing AI-focused was not useful and that token usage alone should not be treated as a measure of productivity.
The trend comes after a period that industry insiders describe as “tokenmaxxing” — employees using as much AI computing power as possible because companies wanted to signal they were aggressively adopting artificial intelligence. Some workers reportedly used premium AI models for simple tasks or even casual conversations, significantly increasing costs.
At Uber, executives acknowledged that spending on advanced autonomous AI systems became so high that the company had already burned through its yearly budget by March. Microsoft has also reportedly reduced access to certain third-party AI coding tools for some employees, while Salesforce introduced systems to track whether AI usage actually leads to business outcomes.
Another challenge is that many companies are still struggling to prove strong returns on their AI investments. Data cited by the report showed that for advanced AI coding tools, only about 18% of token spending ultimately translated into finished software products reaching real users. The rest was consumed by testing, debugging, reviewing and correcting AI-generated work.
Despite growing concerns over costs, most investors and technology executives do not see this as a retreat from artificial intelligence. Instead, they view it as a transition from unrestricted experimentation to more disciplined deployment as companies learn where AI creates the most value.
The broader AI boom continues to accelerate, with businesses worldwide spending hundreds of billions of dollars on chips, data centres and computing infrastructure. But as AI becomes deeply integrated into corporate operations, executives are increasingly discovering that the technology’s biggest challenge may not be adoption — it may be figuring out how to afford it.
United States of America (USA)
Read More
Source link
[ad_3]
