The End of SaaS? What does Satya Nadella’s AI prediction mean for licensing?

In a recent discussion on the BG2 podcast, Microsoft CEO Satya Nadella made a bold prediction: the end of SaaS. The traditional Software as a Service (SaaS) model is on the brink of a significant transformation, driven by the rise of AI agents. He suggested that the conventional structure of business applications—centered around CRUD (Create, Read, Update, Delete) operations with embedded business logic—will soon be supplanted by intelligent agents capable of automating and orchestrating tasks across various systems.

Understanding SaaS

For those of you who are unaware, Software as a Service (SaaS) is a cloud-based delivery model that provides software applications over the internet, allowing users to access them via subscription. Unlike traditional software, which requires local installation and maintenance on individual devices or servers, SaaS eliminates these complexities by hosting applications on the cloud. This approach ensures seamless updates, scalability, and reduced IT management. Common examples include email services, calendaring, and productivity suites like Microsoft 365.

This is why Satya’s statement is bold because Microsoft is heavily invested in SaaS, selling a vast range of cloud services. However, by recognizing the shift toward AI early—demonstrated through their collaboration with OpenAI and the development of Copilot—Microsoft has positioned itself to pivot quickly and capitalize on the transition from SaaS to AI agents.

The Impact of AI Agents

Nadella envisions a future where AI agents manage business rules and processes across multiple databases, automating functions that currently require distinct applications. This shift would streamline operations, reduce reliance on traditional backend systems, and enhance efficiency. For instance, a project manager today might use separate tools for tracking progress, resource allocation, and communication. In the future, an AI agent could handle all these tasks automatically in response to simple commands.

Microsoft plays into this with, for example, Copilot Studio. Copilot Studio allows users to create their own AI agent with just a few clicks – and the result looks exceptional! In their latest earnings call (Q2 2025), Microsoft stated that 70,000 customers worldwide are already using Copilot Studio, highlighting how quickly the transition is gaining momentum.

Potential Impact on Licensing

The emergence of AI agents could significantly alter software licensing models. Traditional SaaS platforms often operate on per-user or per-seat licensing. However, as AI agents take on more tasks, the value proposition shifts from individual user access to overall system efficiency and outcome-based performance. This may lead to new licensing frameworks that focus on the capabilities and outputs of AI agents rather than the number of human users. Companies might adopt consumption-based pricing or performance-based models, aligning costs with the tangible benefits delivered by AI-driven automation.

This is an important consideration, as it could make it more challenging for end users—especially businesses—to accurately predict the costs of using AI agents. For instance, Azure OpenAI services for GPT-based models are billed on a tokenized basis, but the definition of a “token” remains somewhat ambiguous: “a string of letters.” What does this mean exactly? Is it a word, a set of words, or does it exclude whitespace? Also, how do you calculate this correctly?

Unlike the straightforward SaaS pricing model, where costs are based on the number of licensed users, this new model shifts to a consumption-based approach—how many queries will you run? This fundamental change requires businesses to rethink their financial planning and budgeting strategies.