AI’s corporate honeymoon phase may be ending

After spending the past two years treating AI adoption like a corporate arms race, some executives are starting to realize the technology’s benefits may be arriving slower than the invoices.

A handful of major companies have begun scaling back are beginning to scrutinize their AI spending more closely:

• Microsoft reportedly canceled many of its internal Claude Code licenses in part due to rising costs.
• Uber COO Andrew Macdonald in an interview on Saturday said AI spending is getting “harder to justify,” after the company blew through its entire annual Claude Code budget by April without a proportional increase in production.
• An AI consultant told Axios one company accidentally racked up ~$500 million in AI costs in a single month after failing to put usage limits on Claude for employees.

Why are costs ballooning so quickly?

Most enterprise AI systems charge businesses based on “tokens,” or chunks of data processed by the model, meaning every prompt carries a cost.

• The recent rise of agentic AI systems that repeatedly chain together queries means some tasks are now consuming exponentially more tokens.
• Additionally, employees who are incentivized by internal management to use AI tools often inflate usage scores by routing unnecessary tasks through frontier models

While companies across the board have been quick to adopt new AI tools, many are still struggling to determine whether the productivity gains justify the cost.

• Businesses say AI is improving areas like coding, customer service, and internal workflows, though the exact return on investment remains difficult to measure.
• Coinbase, Shopify, Pinterest, and Block have all linked recent layoffs or restructurings to AI adoption in some capacity, though some economists argue they may be using it as cover for broader cost-cutting efforts.

Employees are also increasingly skeptical. More than 70% of Americans say AI is advancing too quickly, recent polling shows, while just 9% of US workers said they trust AI to handle complex business decisions—compared to 61% of executives.

Bottom line: Few experts and business leaders expect the broader AI boom to suddenly disappear. Microsoft, Meta, Amazon, and Google are collectively pouring hundreds of billions of dollars into AI infrastructure, while supporters argue the current AI growing pains resemble previous tech revolutions before their business models fully matured.

📊 Flash poll: In your opinion, is the US economy currently in an AI bubble?

• Yes
• No
• Unsure/other