Uber hat sein gesamtes KI-Budget für 2026 in vier Monaten aufgebraucht. Jetzt fragt sich der COO, ob sich das lohnt
https://fortune.com/2026/05/26/uber-coo-ai-spending-tokens-claude-code/
Uber hat sein gesamtes KI-Budget für 2026 in vier Monaten aufgebraucht. Jetzt fragt sich der COO, ob sich das lohnt
https://fortune.com/2026/05/26/uber-coo-ai-spending-tokens-claude-code/
45 Kommentare
They miscalculated their expenses, and then realized that AI is too expensive.
First Microslop, now Uber… Who’s next?
You can’t put a price tag on a good Al.
Alfred, Albert, Weird Al; doesn’t matter. Priceless.
The question that people are starting to ask (and probably should’ve years ago) is whether AI is revolutionary or evolutionary, and, well, unfortunately it’s becoming increasingly clear it’s the latter.
Sounds like they will be doing layoffs soon so they can allocate more to the AI budget
We are either at the peak or close to the peak of how good the models are vs the price we pay.
Soon the costs will go up and that will only be for access to cheaper models. The good models will be even higher.
This whole thing is not sustainable.
No one is going to want to pay 10x-20x the price for models that still make mistake and are slop cannons.
Use AI in everything!
Wait, not like that!!
Spectacular example of businesses dramatically increasing their expenses in an attempt to penny pinch. They already had a finely tuned business model that extracted the most amount of profit in the sleaziest ways possible and it still wasn’t enough. Over 10% of their code output had no human supervision and it made them burn through a year’s worth of cash in a third of the time. They’ve possibly introduced vulnerabilities with no cost savings to show for it.
TL, DR: Lmao get fucked
Companies: Use more AI! You should be using more tokens! More more more!
Developers: *Use more tokens*
Companies: *ShockedPikachu.jpg*
The greatest scam ever pulled was convincing these ignoble kunths that the workers can be replaced by large language models.
Just wait until AI isn’t subsidized and it’s astronomically more expensive when required by investors to turn a profit
it’s ok— we’ve 10x productivity— right????
/s in case people actually still need it
I’ve been burning our AI budget and almost at 90% used…and it’s May.
Now imagine everyone doing the same. The company is pushing everyone to adopt and use it. Now they’re learning the cost.
Burn it all down. People did their job just fine before AI showed up.
>The comments follow reports that the firm had already burnt through its entire 2026 AI coding tools budget in just four months after **incentivizing employees to adopt the technology through an internal leaderboard ranking teams by total AI tool usage.**
So, leadership says „here guys, use this in your work. See how it goes“. Employees use it extensively, but it doesn’t save leadership any money?
Sounds like AI is a waste of time…
You hear that? It’s the sound of the massive bubble finally popping.
Use ai for the ceo
AI is artificially priced cheap right now so everyone gets locked into a use case, and then they can start charging the true rates they need to charge to make any of their investment back.
The only people desperate enough to try going with AI seem to be greedy CEOs who are learning through the process some refer to as: FAFO
Just wait until they realise they cannot revert slop code…
Of all the companies I can’t care about being surprised that prices go up once the VC funding dries up, Uber is definitely at the top/bottom depending how you look at it.
all to disrupt taxis
WHY. DOES A RIDESHARE APP. NEED. AI.
My company is heavily invested in the AI agenda because the CTO went for a lunch with other CTOs of nearby companies and now they are all measuring their AI-penises.
This is what’s happening on global scale.
burned the 2026 budget by april and now theyre acting surprised. classic „move fast“ until the invoice shows up
From my experience with AI, i can definitely see this happening. You are looking at problems to solve and you can do it two ways, .
First let AI just do it, hook the data up to the AI, give it some instructions and let it go. Its often quick and out the door, results are often a bit sketch, but you probably met the schedule deadline.
Second you can use the AI to help write the code to do the analysis., but in the end AI is not involved. It takes longer, requires more skilled people to write good prompts, developers to guide/fix the output, QA folks to test the code. Its a longer cycle, you are probably going to be pushing whatever schedule was put in place by the folks hyping AI, but in the end you have a product that works, runs fast and for much less recurring cost that the AI.
If I were a betting man, I would bet Uber went with the first option.
Who would have guessed that incentivizing people to burn tokens would cost a lot of money
Not the first time Uber goes 500 km/h towards a wall. There were stories about how Uber was proud to have thousands of microservices. Inevitably there were follow up stories about how their cloud expenses were out of control and they merged a bunch of services together.
How the F does uber spend so much money??! Their app works already. WTf are they doing?
Here’s a clue. It’s not.
Given that countless studies have shown that AI is in fact, almost always not worth it, this is the epitome of an unforced error.
Bah ha ha ha ha 😂 CEOs are AI rubes
This is the mistake of poor leadership not understanding what AI actually is. Incentivizing ai usage instead of having their employees utilize its implementation when needed. The old “we’ve invested in this so everyone use it to justify our work !”
Uber’s always been stupid and unprofitable, then they used stupid unprofitable AI, well now they’re super stupid and super unprofitable.
And AI providers haven’t started to squeeze their prices yet.
The thing people are slowly starting to realize is that AI has a huge scaling issue, near logarithmic, and isn’t just „train and force more data from employees/complacent in training a bot“. It’s not just make a better model, even people with top tier accounts regularly complain about hitting caps and spending considerable amounts of time to get little down.
They also thought that people would just reactionary to data centers if they’d resist at all. Instead, it’s reignited, agitated and engaged and even organized other people focused on priority separate but related issues (climate change, energy, unemployment, etc) into having a big fat single point of actionable, well understood solutions and resistance.
Free users are cooked. It’s straight up to a point where I honestly am just back to doing everything myself. It’s good as a smart auto correct, but even that’s gotten worse…
COO should be fired for making stupid decisions
was all that cocaine worth it? why, you got a line on more cocaine?
What are they developing? Where are they expanding?
What new features come to uber, the app that serves my destination to a list of drivers chosen by proximity through an already established algorithm, to drive me through the city following a route calculated through an already developed algorithm, on an open source map they pay kack shit, and for this service I pay money calculate through a simple multiplication ?
Bullshit man
AI is the very definition of over-promising and under-delivering.
Ironically, AI is revealing the intelligence absent in CEOs
Get ready to see more of this. I can’t wait to stop having to do the performative AI War Dance.
„AI“ is a dead end word salad generator and code autocomplete with limited use cases. It was a billion dollar idea that a bunch of clowns mistook for a trillion dollar idea, and promptly wasted trillions on it…
LOL, AI is the biggest con of modern times and it really shows that most CEO’s have ***’no fucking idea what they are doing’*** and are certainly not worth the massive pay packets they have scammed.
It seems AI was built entirely around a simple gambit. To use investor capital to subsidize the technology as much as possible with the goal to try and make it good enough to replace every worker possible before the investor capital ran out.
It is seemingly clear this gambit failed, and they seem to have had absolutely no plan B
These people are making millions per year and can’t do simple math. Holy fuck.