Microsoft (MSFT) investor relations material

Platform advantages and AI strategy

• Enduring moat framed as a full-stack platform spanning infrastructure, management, apps, Work IQ intelligence and an agent layer, with relevance expected for at least the next decade.

• Differentiators include partner distribution, an effective sales engine, customer trust in security and sovereignty, IP protection, reliability and a sophisticated technical stack.

• Platform supports more than 11,000 models, from frontier and open-source models to specialized models, including work from the internal MAI team.

• Frontier ecosystem aims to let entities retain a continuous learning loop and capture AI benefits where IP is created, broadening economic participation and the addressable market.

• Model selection is oriented toward the best-value outcome and ROI, mixing frontier models for frontier problems with other models where they fit; Agent 365 is intended to help measure and manage spend.

Azure capacity and investment

• Teams reduced dock-to-live times by more than 50% over a year, with notable progress in H2, while improving throughput across software layers to bring capacity online and convert it to revenue faster.

• Short-term CapEx is demand- and fill-rate-driven; efficiency can lower supply requirements, and the near-term priority is making available capacity revenue-ready quickly.

• Over the next 12 months, limited new capacity can be built and brought online; long-term focus is securing land and power while preserving flexibility to expand if demand warrants.

• Long-term agreements across the supply chain have been used, with significant investment over the past 12 months to secure complete sets of infrastructure components; community impact also supports continued building.

• Cobalt is expected to be in 25 data centers over the course of the year; a mix of internally developed and partner silicon targets competitive price-performance, with Maia benchmarks released and scaling planned.

Capital allocation and pricing

• Debt-market access remains an option when funding is needed and expected ROIC is high; capital would be raised to pursue outsized results.

• Usage-based billing is expanding in coding and M365 alongside user-based pricing; consumption pricing suits units of work or outcomes, while user licenses preserve budget predictability.

• Outcome-based billing is viewed as additive to the addressable market, supporting new outcomes, businesses and revenue rather than only lowering customer costs.

• GitHub pricing experience indicates customers will pay for high-quality productivity gains when perceived value exceeds cost; predictability, spend control and observability remain important.

• Copilot usage growth is supported by productivity and backend efficiency; user licenses provide predictable value, while additional consumption-based tools such as Cowork can monetize usage beyond the license.