Loading...
Microsoft caps AI spending, tells engineers to avoid 'tokenmaxxing'
Microsoft has adopted OpenAI GPT-5.6 for internal use

Microsoft caps AI spending, tells engineers to avoid 'tokenmaxxing'

Aug 05, 2026
12:17 pm

What's the story

Microsoft has introduced new restrictions on the use of artificial intelligence (AI) tools by its employees. The move comes as part of a wider effort to curb excessive and costly AI usage, a trend some companies have dubbed "tokenmaxxing." Jay Parikh, Microsoft's Executive Vice President, clarified in an internal email that "tokenmaxxing is not what we are optimizing for." He emphasized the company's focus on maximizing outcomes that benefit customers and the business.

AI model adoption

Adoption of OpenAI GPT-5.6 as default AI model

As part of its strategy to get more value from its token investment, Microsoft is adopting OpenAI GPT-5.6 as the default AI model for internal use.

The decision was made because this model is cheaper than several other models.

The company also shared updated internal Copilot guidelines with employees and announced that as of July 2026, all divisions will have an "AI token budget target."

Financial stability

Microsoft's financial health remains strong amid AI cutbacks

Despite the new restrictions on AI tool usage, Microsoft has assured that these changes are not due to any financial instability.

The company's latest earnings report shows a year-on-year increase in revenue, operating income, and net income that exceeded market expectations.

This indicates that the tech giant is in a stable position even as it implements measures to control AI spending among its employees.

ADVERTISEMENT

Industry alignment

Other companies implementing similar AI usage restrictions

The new policy from Microsoft is in line with a wider trend among companies looking to control unnecessary AI usage.

Other companies like Amazon, Adobe, Atlassian, and Citi have also taken steps to limit excessive AI consumption by their employees.

Microsoft has assured that it will continue to refine its AI policies as models and products evolve without slowing down its journey toward becoming "AI-first."

ADVERTISEMENT