AI bubble warning signs? Microsoft, Uber face unexpected coding agent cost surge in 2026

Artificial intelligence revolution is meant to bring unprecedented efficiency and boundless productivity for the tech landscape by cutting the costs.

But the promise of revolution is currently hitting a stark fiscal reality marked by unexpected AI-related costs, leading to the burning of budget tech companies designed for 2026 year.

Of all the companies, Uber and Microsoft are grappling with AI budget overruns. For instance, Microsoft recently has decided to rescind its internal Claude code license along with its Experiences and Devices division ending access by June 30, 2026.

The decision to cancel the pilot has been taken only after 6 months since it launched in December 2025 due to unsustainable token-based billing consumption as reported by Fortune.

The announcement comes after the AI costs went beyond company’s expectations. The engineers are advised to use GitHub Copilot CLI instead of Claude.

Similarly, Uber has decided to scale back on AI technologies due to budget overruns caused by overconsumption.

Uber’s Chief Technology Officer, Paraveen Neppalli Naga, revealed the crisis as the company exhausted its 2026 budget of $3.4 billion by April.

The company introduced Claude code to 5000 engineers. Within four months, the usage surged to unprecedented levels. Monthly usage rate among the engineers soared 85-95 percent and API costs ranged between $500 and $2000 monthly per engineer.

Similarly, Nvidia Vice President of Applied Deep Learning also admitted that the cost of AI has exceeded the human-based costs.

Speaking to Axios, Bryan Catanzaro said, “For my team, the cost of compute is far beyond the costs of employees.”

The current AI costs crisis stems from how these coding agents are being deployed in the tech firms. Of all factors responsible for cost explosion, token-based scaling is the most prominent one as thousands of developers use AI systems to automate a large quantity of their coding.

Given the intensity of the crisis, the debate of AI bubble bursting has once again been reignited. According to experts, we are moving into a reality check phase.

The tech savvy users also shared their opinion on the social media platforms as one wrote, “This is the AI bubble- well AI can help solve your problems partially but humans are very much needed in every phase of life cycle.”

Another commented, “The cost problem is real, and nobody is talking about it loudly enough until now. Everyone has tracking capability, but almost nobody is tracking the bill.”

.As maintenance costs soar, the question is no longer just whether AI will change the world, but whether the current business models powering it are standing on a bubble ready to burst.