The most valuable intellectual property for companies will be their unique, private evaluation benchmarks. These evals allow them to "hill climb" any model, ensuring they retain control and are not locked into a single AI provider. The ability to switch models and improve performance is the key asset.
Nadella highlights a fundamental shift in work conceptualization. Using the Azure networking team as an example, their job transformed from managing the network to building an agentic system that does it for them. This "meta-work"—designing and overseeing automated systems—is the new frontier of productivity.
Rather than just benefiting specialists, AI provides the greatest leverage to generalists. It allows individuals to translate their knowledge work across different domains and artifacts—from writing a document to building an application—dramatically increasing their scope and impact without deep specialization in each area.
While outcome-based pricing is attractive in theory, customers often prefer the certainty of per-user or consumption-based models. According to Nadella, once a customer achieves a successful outcome, they view sharing that upside as a royalty and quickly ask to revert to predictable pricing structures.
Nadella describes a new frontier strategy: using a large, generalist model to generate initial traces for a specific task. These high-quality traces are then used to fine-tune a much smaller, specialized model, allowing it to achieve superior performance on that single task.
Performance comes from a "harness" surrounding the AI model, which includes curated data, tools, and rich context. This harness, which can be open and multi-model, is where the hard work lies—prepping the context layer so that a model's plan can execute efficiently.
The traditional SaaS model of bundling data, logic, and UI is being challenged. To stay relevant, SaaS companies must unbundle their core assets—like semantic models and business logic—so they can be consumed by AI agents, not just humans via a UI. This creates new agent-driven usage and business models.
Previously, tacit employee knowledge was impossible to quantify. Now, AI agents can capture interaction traces between humans and systems to learn how an enterprise creates value. This learned experience, embodied in a "company veteran agent," could become a quantifiable asset on the corporate balance sheet.