Microsoft tells Wall Street it is stepping up its push with homegrown AI models, aiming to rival OpenAI and Anthropic

Microsoft is indicating that it wants to compete more directly with the leading commercial AI model developers, including OpenAI and Anthropic, as it expands its own stack for building and running artificial intelligence systems.

The remarks, reported by Yahoo Finance on July 29, emphasized Microsoft’s plan to grow by deploying homegrown AI models and the “harnesses” around them, terms used to describe the software and infrastructure designed to connect models to products and workloads. The company also pointed to what it described as a “Mythos” competitor, presented as an additional effort aimed at challenging the incumbent set of AI offerings.

Microsoft’s pitch to investors was framed as an ongoing growth story rather than a single product announcement. In the coverage, the company told Wall Street it expects continued expansion of its AI business, implicitly tying that growth to an approach that combines proprietary model development with broader distribution through its cloud and software ecosystem.

The investor-facing message also underscores how the AI arms race has shifted from early demonstrations to model supply chains and integration. For large cloud and enterprise software vendors like Microsoft, the practical edge often comes from being able to offer not just a model, but also the surrounding tooling that enterprise customers need, including ways to run models reliably, integrate them into applications, and manage costs and performance.

While the report highlights Microsoft’s intent to compete against OpenAI and Anthropic more than before, it does not, in the provided account, specify detailed performance figures, pricing economics, or adoption metrics. It also does not outline the technical benchmarks used to compare model quality, latency, or cost efficiency, which are the kinds of details that typically drive evaluation decisions for buyers choosing between multiple foundation model providers.

For context, the commercial AI market has increasingly consolidated around a small number of foundational model suppliers, each building its own model families and tools. At the same time, buyers have become more cautious about dependence on a single vendor, pushing cloud providers and enterprise software firms to offer alternatives, multilayered support, and more options for deploying AI across different use cases.

Microsoft’s strategy, as characterized in the report, appears aimed at reducing that dependency while capturing value across the AI lifecycle, from model generation to deployment and enterprise integration. That matters because model providers and platform vendors are competing not only on model outputs, but also on availability, compatibility with existing systems, and the overall ability to scale AI into operational workflows.

Still, much remains unclear from the investor account alone. The coverage does not provide specifics on the “Mythos” effort’s scope, whether it is a model, a framework, or a set of applications, nor does it clarify which customers or partners are already using the homegrown models. It also does not disclose timing for additional releases, roadmap milestones, or how Microsoft plans to measure “continued growth” against competitors in publicly comparable terms.