AI costs are rising every day, with more and more employers finding that tokenmaxxing isn’t quite leading to the results they were looking for. Employers are slowly finding it harder to justify AI expenses despite some productivity gains, creating new problems in the workplace that require resolving.
Working with AI presents many exciting opportunities for workers, unlocking new and innovative ways to navigate a problem that might have otherwise taken days to resolve. Coders are no longer working with coding language but with the English language instead, prompting and pushing AI tools into giving them the results they desire, identifying AI errors rather than their own. While there are benefits to be found, there are problems as well.
Uber recently announced a monthly $1,500 cap per employee and per agentic coding tool to cut down on its AI spending. Microsoft has also eliminated parts of its AI spending, routing some of its apps to its own models to cut down expenses on platforms like Anthropic and OpenAI. All while changing GitHub Copilot’s billing method to usage-based billing that left many developers disowning the tool. AI is replacing workers, but it is also creating more expenses via the workers who remain, as they try to maximize the use of AI tools to keep their employers happy.
Regardless of the savings afforded by rampant layoffs, rising AI consumption costs suggest that companies may have to keep eliminating workers in large numbers to support internal AI use, which doesn’t entirely appear to be a sustainable strategy.
AI costs are shooting up as employers remain geared towards maxing out their tokens in fear of falling behind their colleagues. Something needs to change. (Image: Pexels)
As AI Costs Continue to Balloon, Employers Must Pause to Ask What’s Next
Agentic AI may be the next frontier of business operations, the final frontier, some might say, and yet the fast-paced advancements in tech are leaving a large majority of businesses grasping for clarity on what this means for them and how best they can adapt to these changing times. Gartner predicts that spending on AI agent software could reach $207 billion in 2026, a 139% increase from previous years. There is evidence to show that this rise in expenditure is already taking place, with AI token costs spiraling out of control. The Globe and Mail reported that a Canadian company saw its AI spending go up from $20,000 a month to $100,000 a month, all within just a few months.
US AI sector in 2025: $100 revenue, $140 costs.
In my equity research days, we called this “losing money.” Today, it’s called “path to profitability.”
Maybe both are true. But the question is who finally keeps the profits: companies burning cash, or customers enjoying… pic.twitter.com/FjVsfqz2oV
— Vishal Khandelwal (@safalniveshak) July 9, 2026
On a related note, SemiAnalysis revealed that its annualized token spend is already 30% of employee compensation, with the company consuming almost 5 billion tokens per month per employee, with some employees running over 100 billion tokens a month. For another example of extravagant AI use, Meta reportedly saw a “Claudeonomics” leaderboard track 85,000 employees inhaling 60 trillion tokens within 30 days. One employee used an average of 281 billion tokens. Fortune estimated that even with the least expensive version of Claude 4.6, this single user could have cost the company over $1.4 million in just a month.
The leaderboard was generated by employees and then shut down by them after the data was shared externally, but it gives us a look into just how much AI is leading how we work today. It also showcases how Big Tech appears to be willing to bear these AI costs. For some companies, this is a feasible scenario where the expenses don’t outweigh the benefits. For others, this paints an unrealistic picture, especially when employers don’t quite know what to do with the AI tools they are investing in. As we continue to witness AI replacing workers across roles, it is important to pace these operations out more carefully.
AI Costs Stump Employers On Multiple Fronts
A new KPMG report found that 29% of senior executives had no idea where the rising costs associated with AI were coming from. Many hadn’t taken into consideration the considerable economic barriers that come with AI use. With the way things stand, AI token costs could eventually outpace employee salaries. This isn’t just a fear, but a distinct possibility. Gartner predicts that within the next two years, AI costs could meet or even exceed a typical software engineer’s salary.
This is a conservative estimate and may not truly trump the salaries of all workers collectively, however, this is a very real scenario that businesses need to prepare for, particularly those that aren’t certain about the “how” and the “where” of AI use. As AI tools are evolving, so are their billing processes. Where there is some possibility that they could eventually choose to make their plans more affordable, employers shouldn’t count on it as part of their strategy.
Employers Remain Confused About How Best to Balance AI Tools in the Workplace
It is important to acknowledge that while AI is replacing jobs, it is also creating some new ones. Many organizations are now hiring experts into senior positions, not just to supervise teams using AI, but to better reshape their investments in the technology and determine whether there is a good use case in their line of work. Organizations are similarly hiring editors and testers with the sole purpose of reviewing AI-generated work to make it more factual and palatable for its audience.
While this is great news for industry experts, it doesn’t fully cater to the number of people now out of work, and neither does it solve the problem of balancing results with expenses, which requires dedicated effort to navigate.
A Boston Consulting Group (BCG) report found that there is actual evidence of employees saving time with AI. And yet employers don’t entirely know what to do with that saved time. Where 42% of frontline employees and regular users of AI admitted to saving at least eight hours a week through the use of the technology, 66% stated that they received very limited or no guidance on what they should do with that saved time.
Simply investing in AI tools and hoping that employees can find a purpose for them may work in the initial days, but with the costs ramping up, organizations may need to look more closely at redesigning their systems to make full use of the technology that is being made available.
What has been your experience with the rising AI costs and the experience of employees? Share them with us in the comments. Subscribe to The HR Digest for more insights on workplace trends, layoffs, and what to expect with the advent of AI.
