Major tech companies have run into a new problem: the widespread adoption of AI tools has started driving costs up instead of down. Internal programs that incentivize employee AI usage have led to a rapid surge in computing and query processing expenses.
Industry sources say companies are pushing teams to use their in-house AI tools while cutting back on third-party solutions. A key driver is the fast-rising expense of processing queries, especially as AI applications become more sophisticated.
Agentic AI, as it’s called, is especially costly. These systems don’t just generate a single answer; they autonomously carry out a chain of actions, analyze data, and make intermediate decisions. Such processes can use hundreds or even thousands of times more computing power than a standard text query.
Meanwhile, a new internal trend has emerged: employees are turning to AI for even the simplest tasks just to register high usage numbers within the company. The result is a paradox: the more widely available the tools became, the more total consumption soared.
Some teams are already reassessing their AI strategy, looking beyond productivity gains to the actual operational expense. The question that keeps coming up: is automation truly cheaper than manual work when compute costs are rising faster than anticipated?
Experts predict the market will slowly shift focus from counting AI usage to measuring real efficiency and economic returns. For companies, that means the next phase isn’t about blanket AI deployment, but picking the use cases where it truly pays off.