OpenAI ignites Microsoft's AI revenue super engine! The main line of the AI bull market is shifting from "buying chips" to "monetizing AI applications."

According to the latest disclosure from the American tech giant Microsoft (MSFT.US), most of the AI-related revenue for this super giant, which focuses on AI cloud computing infrastructure and software, comes from OpenAI. OpenAI is not only the most important supplier of cutting-edge AI large models for Microsoft's business empire but also a super customer that consumes Azure computing power, pays for model training and massive AI application inference costs, and shares revenue and profits; Microsoft's major shareholder OpenAI utilizes Azure infrastructure, enterprise sales channels, and software entry points like Microsoft 365, GitHub, and Dynamics to package OpenAI models into a scalable enterprise AI application service ecosystem.

It is understood that Microsoft recently stated in a regulatory filing that in the fiscal year ending in June, the tech giant generated approximately $24.1 billion in revenue from this AI startup. Microsoft CEO Satya Nadella has stated that as of the end of the fiscal quarter in March, the company's AI business is expected to reach an annual revenue of $37 billion based on the growth rate at that time. When Microsoft announced its fourth-quarter results last week, it did not update the total sales figures related to its AI business.

The current investment theme in the stock market is gradually shifting from "who can configure and build the largest GPU data centers" to "who can convert Tokens into sustainable cash flow," which is the underlying logic behind Goldman Sachs' optimism about Microsoft becoming a core beneficiary in the "monetization phase of AI applications." Compared to independent AI applications that require reacquiring customers, Microsoft can cross-sell Copilot and intelligent agent services to its vast existing enterprise installations, with lower marginal customer acquisition costs, deeper data and workflow barriers, while simultaneously covering the super ecosystem of models, cloud platforms, development tools, and AI application layers.

The Concentration Truth Behind AI Revenue Prosperity: OpenAI May Contribute About 70% of Microsoft's AI Revenue

This latest disclosure indicates that in the actual AI-related business revenue data achieved by Microsoft in its most recent fiscal year, OpenAI contributed more than half, and it is likely around 70%, highlighting the extent to which Microsoft still relies on this closely-knit partner.

According to the agreement between the two companies, OpenAI is required to pay Microsoft for AI cloud computing infrastructure costs, the related costs of building AI models, and a certain percentage of revenue sharing. Microsoft has been trying to reduce its dependence on this partner, including investing in OpenAI competitor Anthropic and developing its own models. However, when specifically assessing the progress of Microsoft's AI business and overall valuation, investors are always questioning: how much revenue is actually contributed by OpenAI.

Market research firm Bloomberg Intelligence's analysis assumptions show that Microsoft's AI business's annual run rate (ARR) continues to maintain the company's reported 123% high growth rate in March. Based on this growth rate, Microsoft's AI-related business is expected to achieve approximately $34 billion in revenue for the fiscal year ending in June, a figure that can be directly compared to the approximately $24.1 billion revenue that OpenAI contributed to Microsoft in that fiscal year Microsoft has only disclosed the total scale of its artificial intelligence-related business twice. The first time was in the quarter ending December 2024, when the company stated that the AI-related business was expected to exceed $13 billion in annual sales based on the current operating pace. The second time was during the quarterly earnings meeting for the quarter ending March, where Microsoft indicated that its annual AI-related business revenue was expected to exceed $37 billion.

The total amount of Microsoft's AI-related business broadly includes revenue from all AI-related customer groups, as well as revenue data generated from selling Microsoft’s AI-exclusive products to any customer. However, a Microsoft spokesperson has confirmed that the revenue contribution figures from OpenAI mainly include all sales from OpenAI and revenue sharing.

When compared to Microsoft's total revenue scale, the business proportion contributed by OpenAI appears to be much smaller, at less than 10%. Microsoft stated that the company added approximately $51 billion in commercial orders in the most recent quarter, primarily driven by customers outside of AI startups. Nevertheless, OpenAI still contributed to the majority of Microsoft's annual order growth.

Until last week, Microsoft had never explicitly disclosed its total revenue data obtained from OpenAI. Olga Usvyatsky, an accounting researcher and founder of data analytics firm Nonlinear Analytics, wrote in a research report that this latest disclosure may be related to OpenAI's plans for its initial public offering (IPO).

Jackson Adde, a senior analyst from Wall Street financial giant KeyBanc, stated that there remains a significant unanswered question: how much of the revenue data contributed by OpenAI comes from revenue-sharing agreements, and how much comes from AI cloud computing business or other services provided by Microsoft. He said, "The more revenue that comes from providing services to OpenAI, rather than from investment returns, the more positive my view of this business will be."

From "Selling AI Computing Resources" to "Selling Productivity": OpenAI Ignites Microsoft's AI Revenue Engine

Microsoft's heavy reliance on OpenAI is not due to a lack of cloud computing or software engineering capabilities, but rather because the two have built a highly coupled "bilateral business flywheel": OpenAI is both Microsoft's most important supplier of cutting-edge models and a super customer that consumes Azure computing power, pays for model training and inference costs, and shares revenue data; Microsoft, in turn, utilizes Azure infrastructure, enterprise sales channels, and software entry points such as Microsoft 365, GitHub, and Dynamics to package OpenAI models into scalable enterprise services.

The latest disclosure shows that for the year ending June 2026, Microsoft confirmed approximately $24.1 billion in revenue from OpenAI-related business arrangements; according to Bloomberg Intelligence estimates, this could account for about 70% of Microsoft's actual AI sales. Therefore, Microsoft's current AI growth benefits from OpenAI's technological leadership, but also faces risks related to customer concentration, bargaining power, and revenue cyclicality—investors must distinguish between the computing power and revenue sharing from OpenAI itself and the real AI application revenue data from millions of independent enterprise customers Microsoft's latest quarterly revenue reached $90 billion, a year-on-year increase of 18%; Azure and other cloud services grew by 43%, with Azure's annual revenue data exceeding $100 billion for the first time. The paid seats for Microsoft 365 Copilot increased from over 20 million in the previous quarter to over 30 million. In terms of AI applications, Microsoft's core advantage lies not only in its models but also in its mastery of enterprise identity permissions, emails, documents, meetings, code, CRM, and data governance systems, allowing AI to be directly embedded into existing workflows. This creates a high-sticky recurring revenue scale through seat subscriptions, usage billing, and Azure consumption.

The super bull market surrounding AI is shifting from "buying chip stocks" to "buying AI workflows," meaning the current market is re-pricing the investment mainline of the AI bull market from "who invests the most capital expenditure" to "who can quickly convert computing power into ARR, profit margins, and free cash flow." This latest rotation benefits software companies focused on AI application platforms that embed into key enterprise processes, have high renewal rates, data barriers, and monetization capabilities for intelligent agents, but it does not mean that all traditional software stocks will rise simultaneously.

With the stock prices of the three major cloud computing and software giants—Microsoft, Amazon, and Google—surging, especially Amazon's stock price rising 20% since the end of July, pushing its market value past the $3 trillion mark, recent software stocks have shown significant strong rotation relative to the broader market and semiconductors. In July, the S&P 500 index slightly fell by about 0.1%, while the iShares Software ETF (ETF code: IGV) rose by 4.4%; Workday, Accenture, and ServiceNow rose by approximately 31%, 33%, and 43% respectively that month. On certain trading days when chip stocks faced severe deleveraging sell-offs, application software stocks like Adobe, ServiceNow, Workday, and Palantir generally rose by about 7%-10% in a single day, while the software ETF increased by 3.3%, and the semiconductor ETF actually fell.

Goldman Sachs is optimistic about Microsoft becoming a core beneficiary in the "application monetization phase," primarily based on the logic that, compared to independent AI applications that need to reacquire customers, Microsoft can cross-sell Copilot and intelligent agent services to its large existing enterprise installations, with lower marginal customer acquisition costs, deeper data and workflow barriers, while covering models, cloud platforms, development tools, and application layers simultaneously. However, investment judgments cannot solely rely on the total revenue from AI business; whether Microsoft can achieve the next round of valuation leap depends on the growth rate of AI revenue after excluding OpenAI revenue sharing, the number of paid seats for Copilot and revenue per seat, enterprise renewal rates, Azure AI gross margins, and capital expenditure return rates. In other words, OpenAI has helped Microsoft win a ticket to the AI era, but whether Microsoft can become the long-term biggest winner ultimately depends on its ability to transform its reliance on a single partner into multi-model, platform-based AI application revenue covering global enterprise customers