AI Spending: Companies Cut Costs & Set Usage Limits

Companies are now closely managing their AI spending after an initial period of rapid adoption. Prices for popular AI tools have risen, leading to significant budget concerns. For example, Coinbase saw "parabolic" internal usage after a new AI model launched in February. Now, Coinbase has implemented weekly price caps for employees, ranging from $500 to $5,000, depending on job level and role. Other companies are also adjusting. Walmart set usage limits on its internal programming tool, and Amazon closed an internal "tokenmaxxing" leaderboard. OpenAI, Anthropic, and GitHub have all changed their pricing models, charging customers based on token usage rather than flat rates. Salesforce CTO Parker Harris notes the need to balance AI spending with business objectives, stating, "We gotta run a business, we're a public company." The shift is from simply using AI to using it effectively. This new focus on "hard-nosed utility" means the "magical thinking era" is over. This change impacts how companies manage their resources and integrate AI into their operations.