AI Spending Reality Check: Companies Rein In Costs After Budget Overruns

The era of unlimited AI experimentation inside enterprises appears to be ending as companies confront the financial reality of widespread adoption. After two years of encouraging employees to use AI tools liberally, organizations are now implementing stricter controls and demanding measurable returns on investment.

The spending problem has become acute. Uber burned through its entire 2026 AI budget in just four months, with its COO acknowledging that AI expenditures are becoming harder to justify, according to reporting first published by Fortune. One consultant told Axios that a client spent half a billion dollars in a single month after failing to cap employee AI usage.

Why it matters

This shift marks a critical transition from AI as an experimental technology to AI as a managed business resource. Companies that fail to control costs now risk undermining executive support for AI initiatives altogether, while those that optimize spending strategically may gain competitive advantages. The pullback also signals that enterprises are moving past the hype cycle and demanding practical business value from AI investments.

The optimization phase begins

Peter DeSantis, senior vice president at Amazon, characterized the current moment as a predictable phase in technology adoption. He drew parallels to early cloud computing, when customers initially embraced agility but later woke up to unexpectedly high bills. The pattern requires organizations to develop budgeting frameworks and usage controls for new technologies.

Philippe Rambach, chief AI officer at Schneider Electric, told Fortune his company has shifted toward matching specific use cases with appropriate models rather than defaulting to the most powerful options. The company now emphasizes using cheaper models when frontier capabilities aren't necessary.

"The question of the cost of AI is becoming more and more important," Rambach said. "We need to have that under control. We need to measure it. We need to include that in our business case, business plans, and decisions."

The spending surge stems partly from corporate mandates to adopt AI widely. Many organizations distributed AI tool licenses broadly and encouraged heavy experimentation. Employees responded by using AI for nearly everything—including trivial tasks like checking weather—driving up costs without clear value creation. As one executive noted, people tend to automate what they dislike rather than what delivers business impact.

Market dynamics shift

The cost reckoning coincides with broader changes in the AI landscape. ChatGPT's market share has fallen below 50% for the first time, dropping to 46.4% by the end of May, according to Sensor Tower's State of AI Report for 2026. Google's Gemini now holds 27.7% of the market, while Anthropic's Claude claims approximately 10.3%.

Meanwhile, geopolitical concerns are intensifying. The U.S. government's recent decision to shut down foreign access to Anthropic's Mythos-tier models has heightened European anxieties about technological dependence on American providers. At the VivaTech conference in Paris, discussions centered on achieving meaningful AI sovereignty.

Companies are now entering what Rambach described as "coming back to reality"—ensuring AI delivers measurable business value rather than serving as an expensive experiment. This transition from exploration to optimization will likely define enterprise AI strategy for the next phase of adoption.

The details were first reported by Beatrice Nolan for Fortune.