26.05.2026
5 min
140
Companies that only recently pushed the idea of replacing workers with artificial intelligence are now starting to sharply cut AI spending. Microsoft has already begun limiting employee access to Anthropic models and revoking Claude Code licenses, while Uber is reconsidering its expenses due to the extremely high cost of AI integration.
The AI boom, which over the past two years was presented as a “future without humans,” is starting to run into a simple problem: neural networks turned out to be extremely expensive to maintain.
According to Western media reports, Microsoft is gradually reducing the use of Claude Code inside the company. Some employees have already lost access to Anthropic models, while certain teams are being moved to internal tools, including GitHub Copilot. The reason is simple: AI costs have started rising rapidly.
A similar situation is unfolding at Uber. The company aggressively integrated generative AI into software development and internal operations, but over time it became clear that the expected savings were not there. On the contrary, AI tools became extremely expensive to maintain, especially for large teams.
Particular attention was drawn to comments made by a top executive at NVIDIA. According to him, some teams are already spending more money on artificial intelligence than on their own employees. For certain engineers, monthly bills for tokens and computing resources are reaching $9,000 to $10,000.
The biggest financial drain comes from so-called AI agents. These systems can independently write code, analyze data, launch tasks, and perform complex multi-step operations. However, they also consume massive amounts of GPU resources and tokens.
Following the popularity of ChatGPT, Claude, and GitHub Copilot, many companies rushed to integrate AI into programming, customer support, analytics, and workflow automation. But together with the growing popularity of neural networks, spending on data centers, GPUs, and cloud computing skyrocketed as well.
Against this backdrop, analysts predict that global AI and IT spending could exceed $6 trillion in 2026. This is already raising serious concerns even among the world’s largest technology companies.
Just a year ago, companies were talking about quickly replacing humans with artificial intelligence. Now the market is increasingly facing a different reality: AI may become so expensive that keeping humans in certain roles is simply cheaper.
