Companies are reducing their use of AI: in some cases, human employees are proving more efficient

Companies are reducing their use of AI: in some cases, human employees are proving more efficient


Companies are reducing their use of AI: in some cases, human employees are proving more efficient

Many companies had implemented artificial intelligence systems to reduce costs and automate tasks. However, it has become clear that neural networks are very expensive due to high computational power and token costs.

Moreover, in many cases, the results do not justify the expenses:

• Microsoft is reducing the use of Claude Code in internal systems — payments for some licenses have been stopped because the service became too expensive;

• Uber executives said the annual budget allocated for Claude Code was already exhausted by April. High token costs are not delivering significant results for users;

• Starbucks shut down the AI system used for product monitoring — the neural network often confused products or failed to identify them at all, while human employees worked more efficiently;

• The vice president of Nvidia openly stated that AI computing costs have now become more expensive for his team than regular employees.