Microsoft is fundamentally changing how businesses pay for artificial intelligence, moving away from traditional per-seat licensing toward a model where customers pay based on how much they actually use AI agents. During the company's latest earnings call, CEO Satya Nadella explained that this shift reflects a broader transformation in how AI delivers value, particularly as AI agents become more capable at handling real work.

What Is Microsoft's New AI Pricing Model?

Microsoft's new approach combines seat-based entitlements with consumption-based pricing, similar to how cloud computing works today. Customers get a base allocation of usage rights bundled into their license, but beyond that threshold, they pay for additional consumption. The company is measuring this consumption in tokens, which are small units of text that AI models process. This shift is already underway across Microsoft's product portfolio, with GitHub Copilot moving to usage-based pricing starting June 1.

The rationale behind this change is straightforward: traditional per-seat pricing doesn't reflect the actual value AI agents create. Some users might barely touch an AI tool, while others rely on it constantly. Under the old model, both paid the same price. Now, heavy users pay more, but they also see measurable returns on that investment.

"Whether it's customer service, whether it's individual productivity, team productivity, a business process, some cost per user is either decreasing because of the use of agents, or some revenue is increasing because of agents because it was able to compress these workflows," said Satya Nadella, Chairman and CEO of Microsoft.

Satya Nadella, Chairman and CEO of Microsoft

Nearly 60 percent of Microsoft's customer service customers are already purchasing usage-based credits, and consumption of Copilot credits nearly doubled quarter over quarter as organizations build custom agents tailored to their specific workflows.

How Are Enterprises Actually Using These AI Agents?

The adoption numbers paint a picture of rapid, widespread deployment. Nearly 90 percent of Fortune 500 companies now have active agents built with Microsoft's low-code and no-code tools, meaning they don't require deep programming expertise to create them. Beyond the Fortune 500, tens of thousands of companies are already managing tens of millions of agents in Agent 365, Microsoft's platform for building and managing AI agents.

The practical applications span multiple business functions. Microsoft's Agent Mode feature in Excel, for example, automatically creates and edits spreadsheets alongside user actions. This feature became the default for Microsoft 365 Copilot and Premium subscribers in Excel, Word, and PowerPoint last week, and Nadella credited the improvement to better underlying AI models finally making the feature reliable enough for widespread use.

• Customer Service: Data security triage agents handled more than 2 million unique alerts in a single quarter, automating the process of identifying and prioritizing security threats
• Developer Productivity: Nearly 140,000 organizations now use GitHub Copilot, with enterprise subscribers nearly tripling year over year, and command-line interface usage nearly doubling month over month
• Enterprise Analytics: Microsoft Fabric, the company's analytics platform, now has 35,000 paid customers, up 60 percent year over year, with customers using it to process and analyze massive datasets

Why This Pricing Shift Matters for Your IT Budget

The move to usage-based pricing has significant implications for how companies budget for AI. Under the old model, IT departments purchased licenses upfront and hoped employees would use them. Under the new model, spending directly correlates with business value. If an AI agent saves a customer service team 10 hours per week, that value is visible in the token consumption metrics. If an agent doesn't deliver measurable returns, usage naturally drops and so does spending.

Microsoft's Chief Financial Officer Amy Hood noted on the earnings call that the company's AI margins have actually been better than its cloud business margins during this transition, suggesting that the usage-based model is proving more profitable than traditional licensing. This could signal that enterprises are willing to pay more for AI when they see clear returns on investment.

Nadella also suggested that IT budgets in the AI era could see reallocation from other line items on company income statements. Rather than replacing existing software budgets, AI spending might come from operational expense savings that agents create.

How to Evaluate AI Agent Value for Your Organization

• Measure Workflow Compression: Track how much time AI agents save on repetitive tasks like data entry, email triage, or report generation, then calculate the cost per hour saved versus the token consumption cost
• Monitor Outcome-Based Metrics: Beyond time savings, measure whether agents increase revenue, improve customer satisfaction scores, or reduce error rates in critical processes
• Start with Pilot Programs: Deploy agents in one department or function first, establish baseline metrics, then scale based on demonstrated ROI before committing to enterprise-wide rollouts
• Review Token Consumption Regularly: Just as cloud teams monitor data transfer and compute usage, AI teams should review token consumption reports monthly to identify underutilized agents or opportunities to optimize prompts

Microsoft's broader financial performance underscores the company's confidence in this AI-driven future. The company reported $82.9 billion in total revenue for the quarter, up 15 percent year over year when excluding foreign exchange effects. Capital expenditures are expected to exceed $40 billion annually to build out the data center capacity required to support these AI workloads.

The company has also made infrastructure improvements that suggest it's solving the capacity constraints that plagued the AI industry earlier. Microsoft reduced the time it takes to deploy new graphics processing units (GPUs) in its largest regions by nearly 20 percent since the beginning of the year, and it brought the Fairwater data center in Wisconsin online six weeks ahead of schedule. These operational improvements matter because they mean Microsoft can actually deliver the computing power customers need to run their AI agents at scale.

One notable milestone: Microsoft Bing, the company's 17-year-old search service, reached 1 billion active monthly users for the first time, driven in part by AI-powered search capabilities. This suggests that AI integration is not just an enterprise phenomenon but is reaching mainstream consumer audiences as well.

The shift to usage-based, outcome-based pricing represents a fundamental change in how the software industry monetizes AI. Rather than selling access to tools, Microsoft is increasingly selling results. For enterprises, this means AI spending becomes more directly tied to business value. For Microsoft, it means revenue scales with customer success rather than with the number of licenses sold. As more Fortune 500 companies build custom agents and measure their impact, this pricing model will likely become the industry standard.