All that enthusiasm has resulted in skyrocketing costs for so-called tokens—the basic unit of measurement for AI computing—as AI model providers seek to balance supply and demand and manage their own costs.
Now, corporate leaders are scrambling to bring down expenses by finding ways to ration AI use in their organizations, steer workers toward cheaper, homegrown tools, and help them hone their skills to improve returns.
Key Highlights from the Article
-
Blown Budgets: Some enterprises have hit their annual AI token budget in just three months, while others report seeing their AI spending bills double or triple.
-
Executive Cutbacks: Top technical executives at companies like Uber, Meta, Microsoft, Salesforce, and DoorDash have initiated efforts to ensure AI use directly contributes to productivity, with some limiting tool availability for certain employees.
-
The “Tokenmaxxing” Problem: Until recently, all-you-can-eat subscriptions amounted to a subsidy by model-makers. Exhorted to embrace change, employees engaged in “tokenmaxxing”—using as much computing as possible to seem AI-forward—a habit that backfired when model companies shifted to usage-based pricing.
-
Low Return on Code: According to data from EntelligenceAI on over 2,000 companies, only 18% of spending on tokens for advanced AI coding tools actually translates into shipped coding products that reach real users, due to the high costs of debugging and rewriting AI-generated code.
-
Market Repercussions: Tech critics point to this spending friction as a sign that the ultrafast pace of AI growth could face a structural slowdown. This shift comes at a sensitive time for major model creators like OpenAI and Anthropic as they navigate steps toward public listings.
The Reality Check: As Meta Chief Technology Officer Andrew Bosworth noted in an April memo to employees: “Nobody should be using AI tools just for the sake of using them. All motion is not progress and token usage alone is not a measure of impact.”
