Spending on enterprise AI is exceeding corporate projections. Microsoft has canceled most of its internal Claude Code licenses, while Uber has admitted to spending its 2026 AI budget in just four months.
Token based costs for AI programming tools are causing bills to rise faster than the personnel savings benefits. Many companies are having to add further financial controls to the AI solutions they are rapidly deploying by the end of 2025.
Microsoft and Uber clarify the trend.
According to The Verge , Microsoft began phasing out most internal Claude Code licenses in mid-May 2026.
Most access to the company's Experiences and Devices division will end on June 30th. Engineers there have used this AI programming tool extensively.
Token -based billing leads to excessively high usage costs when deployed on a large scale, according to Fortune . This contrasts sharply with internal Microsoft reports claiming AI can boost productivity by up to 80%.
Uber went even further. Chief Technology Officer Praveen Neppalli Naga stated that Uber had spent its entire 2026 AI budget by April.
This company had deployed Claude Code to approximately 5,000 engineers in just the previous four months.
According to Forbes , the cost per engineer ranges from $500 to $2,000 per month. Currently, about 70% of Uber's code comes from AI tools, demonstrating the increasing reliance on Claude within its large engineering teams.
Market data confirms increasing pressure.
A 2025 survey by Mavvrik found that 85% of companies actually exceeded their AI spending forecasts by more than 10%. The same survey also indicated that 84% of businesses reported AI spending reducing their gross profit margins by more than 6 percentage points.
AI cost statistics. Source: Mavvrik survey"The AI cost crisis has begun," commented Crypto Rover, a trader and investor.
Follow us on X for the latest news updates as they happen.
Major tech companies' Capital on AI reached $650 billion in the first quarter of 2026 alone . The number of FinOps teams managing AI spending doubled, from 31% to 63% in just one year.
Anthropic could still benefit despite customer complaints. The company's projected second-quarter revenue of $10.9 billion would mark its first profitable quarter . This cost story also offers various interpretations.
Companies have now tightened controls by introducing usage limits, internal performance rankings, and rotating to cheaper models for AI deployments that were previously "wide open" by the end of 2025.
The coming quarter will show whether these controls can curb AI consumption. Similar pressure is likely to emerge in crypto's AI infrastructure projects in the coming months.
