As open-weight artificial intelligence models gain traction, enterprises now have a growing array of model options when deploying AI applications, and market competition is shifting from a single large model's performance battle toward multiple layers including infrastructure, data management, model orchestration, and enterprise applications.
Morgan Stanley believes this trend will not weaken Microsoft's (MSFT) competitive advantage but could further solidify its leadership in the enterprise AI market. In a research note published on October 26, Morgan Stanley pointed out that Microsoft, backed by its Azure cloud computing platform, Foundry and Fabric enterprise AI platforms, and the massive enterprise customer network built through Microsoft 365 and Copilot, is positioned to be a major beneficiary of the multi-model AI era. The firm maintained its "Overweight" rating and $600 price target on Microsoft, arguing that the market still underestimates the company's long-term growth potential in the AI value chain.
The report noted that as enterprises increasingly choose different AI models based on factors such as cost, performance, response speed, and data sovereignty, the competitive advantage of any single model may gradually diminish, while the strategic value of infrastructure and application platforms capable of supporting multiple models simultaneously will rise further. Morgan Stanley identified three key advantages that enable Microsoft to continue scaling its AI business through this transition.
First, the growing diversity of AI model choices will further enhance the strategic value of Azure and the Foundry platform. According to OpenRouter data cited in the report, among the top 50 most-used models on the platform, open-weight models now account for approximately 72% of token usage, up from only about 22% a year ago. Meanwhile, since early 2026, the number of Microsoft customers building applications using products from multiple model providers has grown fivefold, indicating that the trend toward multi-model architectures is accelerating. Morgan Stanley noted that as more frontier models, open-weight models, and small models enter the market, enterprises no longer need to rely on a single vendor for all AI tasks but can dynamically select the most suitable model based on the actual requirements of different workloads. This means model selection will gradually evolve from a one-time technology decision into an ongoing cost and performance optimization process.
Against this backdrop, the value of Microsoft's Azure AI Foundry will become more prominent. The Foundry platform currently supports over 11,000 AI models, providing enterprises with a unified environment for model access, management, evaluation, and deployment. At the same time, regardless of which vendor's model an enterprise ultimately chooses, the related workloads can still run through Azure cloud infrastructure. Morgan Stanley believes this architecture, which does not depend on any single model vendor, allows Microsoft to remain relatively neutral amid competition among different AI models and to continuously benefit from growing enterprise AI usage by providing infrastructure and model management services.
Second, Microsoft 365 and Copilot constitute an enterprise distribution advantage that is difficult to replicate. The report noted that as AI inference costs decline, more high-frequency, low-value daily work tasks will become economically viable for automation, and Microsoft has already embedded itself into core enterprise workflows spanning productivity, collaboration, communication, and knowledge management through Microsoft 365 and Copilot. Unlike standalone AI applications that require enterprises to make additional procurement and deployment efforts, Microsoft can directly introduce AI assistants and intelligent agents into the office environments customers already use every day through its existing software products. This distribution advantage not only helps lower the barrier for enterprises to adopt AI but is also expected to drive rapid adoption of AI features across a vast existing customer base. Currently, paid seats for Microsoft 365 Copilot have exceeded 30 million. Morgan Stanley's latest Chief Information Officer survey shows that 47% of responding CIOs prefer building customized AI applications with Microsoft, while the proportions choosing Amazon AWS and Google are both 10%. The firm believes this gap reflects Microsoft's significant customer relationship and distribution channel advantages in the enterprise AI application market.
Moreover, Microsoft's enterprise AI business has begun to form a mutually reinforcing growth cycle. Currently, the number of customers using both Fabric and Foundry has exceeded 17,000, up 60% year over year. In this system, Fabric integrates enterprise data and provides business context, Foundry selects and orchestrates AI models suited to different tasks, and Microsoft 365 and Copilot bring AI capabilities directly into employees' daily workflows. As enterprises deploy more AI applications, related activity in turn drives growth in Azure infrastructure usage, forming a continuous growth loop from data and models to applications and cloud computing resource consumption. Morgan Stanley believes this layout spanning multiple layers of the AI technology stack is a key advantage distinguishing Microsoft from single-model vendors or standalone AI application developers.
Third, declining AI inference costs are expected to accelerate the development of Microsoft's usage-based business model. As open-weight models continue to improve in performance and model operating costs continue to fall, the economic barrier for enterprises to deploy AI agents is also lowering. Although lower inference costs may reduce revenue per AI call, Morgan Stanley believes lower prices will stimulate enterprises to expand the scope of AI usage and significantly increase overall call volumes. The report cited the "Jevons Paradox," noting that when a technology becomes more efficient and its unit cost declines, its overall consumption may actually increase. This logic is particularly important for the AI business model transformation Microsoft is pursuing. Currently, Microsoft is gradually expanding usage-based charging mechanisms in its AI products. For example, Copilot Studio bills through Copilot Credits based on enterprises' actual use of intelligent agents. Morgan Stanley believes that as AI agents can perform more high-frequency tasks at lower cost, enterprise usage is expected to increase notably, thereby driving growth in Microsoft's related revenue. At the same time, Microsoft can also leverage its multi-model architecture to allocate workloads to the most suitable models based on the cost and performance requirements of different tasks, further optimizing the operating costs of its AI services. Therefore, lower inference costs do not necessarily mean pressure on the profitability of Microsoft's AI business; instead, they may create opportunities for revenue growth and improved unit economics by expanding usage scale and improving resource allocation efficiency.
Morgan Stanley emphasized that as foundation models gradually become cheaper and substitutability among different models increases, the competitive focus of the AI industry may shift from the models themselves to the infrastructure, enterprise data, model orchestration, distribution channels, and application ecosystem built around them. In this process, Microsoft does not need to bet on which single AI model ultimately wins. Regardless of which model enterprises choose, Azure has the opportunity to capture related infrastructure demand, Foundry can provide model management and orchestration services, Fabric handles enterprise data integration, and Microsoft 365 and Copilot can translate AI capabilities into actual enterprise applications and commercial revenue.
On valuation, Morgan Stanley believes Microsoft's current share price has not fully reflected the long-term growth potential of its AI business. Based on the closing price of $524.51 on October 23, Microsoft's forward 12-month expected price-to-earnings ratio is approximately 27 times, below its historical level of about 28 to 29 times. More importantly, Microsoft's current PEG ratio relative to expected earnings growth is approximately 1.3 times, compared with a historical level of about 2 times. Morgan Stanley believes that given the long-term earnings potential from Azure growth, Copilot commercialization, and increased AI service usage, Microsoft's current valuation remains attractive. According to the firm's base-case forecast, Microsoft's earnings per share could reach $24.08 in fiscal 2028, corresponding to a P/E ratio of approximately 25 times, supporting the $600 price target. In a more optimistic scenario, if Azure growth accelerates further, enterprises adopt higher-priced Microsoft 365 products, and Copilot and Azure AI services achieve greater-than-expected adoption, Microsoft's fiscal 2028 earnings per share could reach $27.64, corresponding to a bull-case valuation of $795.
However, Morgan Stanley also noted that if macroeconomic weakness leads to a slowdown in enterprise IT spending, or if AI application adoption progresses more slowly than expected, Microsoft's revenue growth and profitability could be affected. In addition, continuously increasing AI infrastructure investment could also pressure profit margins. The firm expects Microsoft's Azure and other cloud services revenue to grow 44.9% in fiscal 2027 and 43.2% in fiscal 2028, while Microsoft 365 Commercial Cloud revenue growth is projected at 16.3% and 17.2% over the same periods.
Comments