https://www.axios.com/2026/05/28/ai-spending-roi-enterprise-costs
Archive link https://web.archive.org/web/20260528114303/https://www.axios.com/2026/05/28/ai-spending-roi-enterprise-costs
Why it matters: Companies that rushed to embrace AI are now confronting ballooning IT costs, uncertain productivity gains and growing employee skepticism.
Driving the news: Microsoft canceled most of its Claude Code licenses, in part over costs, according to The Verge, and Uber’s COO said AI costs are getting “harder to justify.”
An AI consultant tells Axios one of their clients recently spent half a billion dollars in a single month after failing to put usage limits on Claude licenses for employees.
Companies are citing AI's ability to automate jobs as a cause for layoffs, though Anuj Kapur, CEO of CloudBees, told Axios that workforce cuts may simply be "the only lever they can pull" to offset their AI bills.
Consumer sentiment around AI is also nosediving, and employees are rebelling against the use of the technology at work.
What they’re saying: The enterprise is undergoing a “healthy swing” away from AI overuse — or “tokenmaxxing,” the push to burn as many AI tokens as possible — Ali Ansari, CEO of model training firm Micro1, told Axios.
Ansari hopes this correction will push companies toward more efficient AI use.
While the market views these tools as working equally well across the enterprise, Ansari says "the reality of AI right now is that it only works for coding."
That disconnect can drive up IT bills without leading to high return on investment in agents, he said.
Friction point: Corporate AI adoption is running into four unique problems.
Use cases: "Most people default to automating tasks they dislike rather than tasks most valuable to the company," Sophia Velastegui, CEO of Velastegui Ventures and former chief AI officer at Microsoft told Axios. Instead, they should focus on using AI to drive revenue.
Costs: One CTO told Axios that employees were using AI models to check the weather. That gets expensive fast: Enterprise AI plans are not truly 'all you can eat,' and even simple chatbot queries can carry heavy token costs.
“World calculator” becomes a more apt description for these things by the day when I try to use them at work I swear to god
Why it matters: Companies that rushed to embrace AI are now confronting ballooning IT costs, uncertain productivity gains and growing employee skepticism.
Sounds like companies fear that they are losing control over the employee 🤔
Here’s the thing that a lot of people in corporate don’t understand about deadlines: They are completely arbitrary and often completely fictional. They know this too! Because every time a deadline is missed, nothing happens. We call it a “miss” and designate a scapegoat who takes on a majority of the blame.
When people/the media constantly double down on the productivity gains, what they aren’t mentioning is that productivity in and of itself is vibes based. The economy is vibes based.
I am going to try (but will probably fail) to enjoy my weekend before I come back to this bullshit
sticker price
Data: When enterprises are hesitant to give AI agents unfettered access to proprietary data, those agents become less effective, Josh Pantony, CEO of Boosted.ai, which focuses on AI tools for finance, told Axios.
Most people default to automating tasks they dislike rather than tasks most valuable to the company
CEO complains about class struggle
I don’t get it. Why wouldn’t the employees like doing valuable things? Have you diverged your own interests from theirs?
Me years ago when I spent 2 weeks writing a one off script to accomplish something that would take me 4 hours to do manually because I had nothing better to do at work
Costs: One CTO told Axios that employees were using AI models to check the weather.
fuck yeah
is it warm yet? is it warm yet? is it warm yet? is it warm yet? is it warm yet?
The thermostat is on the other side of the room, ergo I must burn AI tokens to use electricity to increase the global temperature.
it’s too warm, when will it cool down, when will it cool down?
Please, the tokenmaxxing meta can do better than this. Put a temperature sensor outside a data center running the AI then have the AI check temperature changes in real time and write essays on why it might be changing. As it works the data center rises temps and it has to write more as temps change. Self reinforcing token burning.
You need to have another ai agent editing the essays and then another agent writing detailed critiques of each essay
$500 million accidental bill, meanwhile people get fired for a few minutes of “”“time theft”“”
I’m going through a tokenmaxxing phase rn. Profitcels about to get slopmogged, while I’m based and clankerpilled.
chatPPB find a way to say “hell yeah GOOD post” in no fewer that 69420 words
hell yeah GOOD post
This literally made me laugh out loud.
Companies are citing AI’s ability to automate jobs as a cause for layoffs, though Anuj Kapur, CEO of CloudBees, told Axios that workforce cuts may simply be “the only lever they can pull” to offset their AI bills.
“Well, we’re firing you all.”
“To replace us with AI??”
“No it’s because we spent too much on AI and now we can’t afford you or AI”
they could use their money more efficiently and increase productivity by giving it straight to nvidia for GPUs so the employees can game on company time
Many businesses are so unbelievably gullible when it comes to AI. They throw all existing performance indicators out of the window and only chase AI adoption about all else. Doesn’t matter if it makes the employee more productive. Those burning many tokens are commended, those not using enough AI are fired.
Their investors are all in on it most likely. The C-suite answers to the board of directors who is in the bag for AI since it is a monopolizing, union busting, wage and workforce cutting tech – whether it “works” for any single company or not.
It’s a bit like that return to the office push. There are recognizable class interests but there is also a psychological component with people that don’t materially benefit jumping on the cargo cult and just really believe in the idea that is how business is done.
And I guess there might be secondary benefits from posting about your management style on LinkedIn and getting affirmation and reach for following the trends.
The code formatting is difficult to read on mobile when the lines run long.
Not ideal on desktop either
Maybe I should be using AI at work🤔
Don’t you want to know the last digit of pi?
Throwing AI licenses at the wall and seeing what sticks (or what Velastegui calls the “thousand flowers bloom” approach) isn’t leading to tangible returns, she said.
Make sure to not use chinese ai though because that would be unamerican.Watch western AI companies just buy chinese AI and then rebrand it as american.
they don’t even have to buy it, deepseek is free (gratis). i think it only has use based restrictions for military stuff.
deepseek does horny stuff??
3.4 You will not use the Services to generate, express or promote content or a chatbot that:
(5) is pornographic, obscene, or sexually explicit (e.g., sexual chatbots);
Volcelseek
this is the terms of use for deepseek.com people can still download the model and host it themselves and then only the license would apply.
first take a cold shower. second i don’t know but it is not against the license if it does not violate any other laws. the license of deepseek-V3 (based on this) states:
attachment A
You agree not to use the Model or Derivatives of the Model:
- In any way that violates any applicable national or international law or regulation or infringes upon the lawful rights and interests of any third party;
- For military use in any way;
- For the purpose of exploiting, harming or attempting to exploit or harm minors in any way;
- To generate or disseminate verifiably false information and/or content with the purpose of harming others;
- To generate or disseminate inappropriate content subject to applicable regulatory requirements;
- To generate or disseminate personal identifiable information without due authorization or for unreasonable use;
- To defame, disparage or otherwise harass others;
- For fully automated decision making that adversely impacts an individual’s legal rights or otherwise creates or modifies a binding, enforceable obligation;
- For any use intended to or which has the effect of discriminating against or harming individuals or groups based on online or offline social behavior or known or predicted personal or personality characteristics;
- To exploit any of the vulnerabilities of a specific group of persons based on their age, social, physical or mental characteristics, in order to materially distort the behavior of a person pertaining to that group in a manner that causes or is likely to cause that person or another person physical or psychological harm;
- For any use intended to or which has the effect of discriminating against individuals or groups based on legally protected characteristics or categories.
I don’t need to want to use it for horny stuff to be surprised if they allow it, i’d be surprised if they did because of 1) all the liability of keeping it from doing illegal stuff and 2) if you can get deepseek to jerk you off for free then why would people pay $300 a month or whatever for Grok Waifu Edition
the version the deepseek team has hosted explicitly forbids horny stuff, but if someone else is hosting it they are subject to the licence only which doesn’t forbid it explicitly. it won’t be free but it’ll certainly be cheaper than Grok White Power gf™.
The deepseek interface on chat.deepseek.com doesn’t allow horny stuff, but
spoiler
the deepseek API, even the official one, doesn’t seem to mind at all
the API does cost money, but it’s so incredibly cheap that it’s basically free. why anyone would pay for grok waifu edition is beyond me
Probably would have been more environmentally friendly to just set up 500M in dollar bills on fire
In car tires even.
I’ve read that Anthropic’s yearly revenue is in the single/double digit billions (want to say 13.5, but I’m not sure) so this company’s error would account for 4% (if my 13.5b is the number) of their yearly revenue in one month…
this reminds me of when AWS was the hot new thing and people did not understand you have to define a spending limit. how did people fuck this up AGAIN?
they only care about optimizing costs when it comes to paying workers
Because its not about the costs. It’s about making sure workers stay desperate and destitute. So they can’t organize.
Hear me out… running AI models on Lambda
i went broke from just reading your suggestion i want a refund
