Microsoft: How much does it cost to invest in AI?

The following is a summary of Microsoft's third-quarter earnings conference call in 2024. For an interpretation of the financial report, please refer to " Microsoft: Azure Holding Up Everything, a Solid "Talisman" for US Stocks 》

1. Review of Core Financial Information:

2. Detailed Content of the Earnings Conference Call

2.1. Key Points from Executive Statements:

1. Business Highlights:

① Microsoft Cloud revenue exceeded $35 billion, a year-on-year increase of 23%.

② Azure Services: a. Market share continues to grow, especially in AI solutions, Azure utilizes its extensive AI accelerator array to support customers; b. Deepening cooperation with OpenAI, over 65% of Fortune 500 companies use Azure OpenAI services; c. Azure's database services (such as Azure PostgreSQL and Cosmos DB) are deeply integrated with Azure AI to support the development of intelligent applications; d. Azure Arc helps customers simplify cloud migration, with the number of customers increasing by more than double year-on-year.

③ Data Analysis: The Microsoft Intelligent Data Platform provides a wide range of database, analytics, and AI services, with over half of Azure AI customers simultaneously using Microsoft's data and analytics tools; the Microsoft Fabric analytics platform now has over 11,000 paying customers.

④ Developer Services: a. GitHub Copilot has reached 1.8 million subscribers, with quarterly growth exceeding 35%; b. Continuing to expand its low-code/no-code tools to simplify application development and automate workflows.

⑤ Consumer Business: a. LinkedIn enhances features with AI, such as AI-assisted messaging; b. Bing search and Edge browser improve user search and browsing experiences through AI technology applications; c. In the gaming sector, the Game Pass service has added Activision Blizzard's "Diablo 4".

⑥ Security and Compliance: The company has launched the Security Future Plan, focusing company resources on enhancing network security protection. Copilot for Security integrates a large amount of threat intelligence and security signals to provide practical insights for security teams2) Financial Highlights:

① In the next 12 months, it is expected that approximately 45% of the commercial residual performance obligations will be recognized as revenue.

② Segment Revenue: Revenue growth for each business line is strong, with Microsoft Cloud, Office 365 (commercial version), LinkedIn, Dynamics 365, and Intelligent Cloud growing by 23%, 15%, 10%, 23%, and 21% respectively.

③ Cost Side: Operating expenses increased by 10%, including expenses related to the Activision acquisition. Operating profit increased by 23%, with an operating profit margin improvement of approximately 2 percentage points.

④ One-time Costs: The acquisition of Activision contributed about 4 percentage points to total revenue growth, while dragging down operating revenue growth by 2 percentage points, resulting in approximately $93.5 billion in purchase accounting adjustments and related costs.

1. Guidance Outlook:

① Overall Guidance: a. Expect double-digit revenue growth to continue in the fourth quarter and the next fiscal year (FY'25); b. Capital expenditures for FY'25 are expected to increase due to the growth in demand for cloud services and AI products; c. Despite significant investments in cloud and AI, as well as the full-year impact of the Activision acquisition, the operating profit margin for FY'25 is expected to decrease by only about one percentage point.

② Fourth Quarter Guidance: a. Microsoft Cloud's gross margin percentage is expected to decrease by about two percentage points year-on-year, with changes in cloud service gross margins affected by a shift in sales mix towards Azure; b. Expected foreign exchange fluctuations to result in total revenue and departmental revenue growth decreasing by less than one percentage point; c. Business bookings are expected to achieve robust growth on a steady expiration basis, driven by continued strong business sales execution.

③ Segment Business Guidance:

• Intelligent Cloud: Revenue is expected to grow by 19% to 20% in the fourth quarter.

Revenue growth for Azure is expected to reach 30% to 31%, driven by continued contributions from Azure consumer business and AI.

On-premises server business revenue growth is expected to be in the low to mid-single digits, driven by ongoing hybrid demand including licensing for multi-cloud environments.

• Productivity and Business Processes Division: Expected to grow by 9% to 11% in constant currency in the fourth quarter.

Office: Revenue growth will be driven by Office 365, primarily through ARPU growth of E5, with Office 365 revenue growth expected to be around 14% in constant currency, continuing to drive adoption of Microsoft 365's Copilot.

LinkedIn: Revenue growth is expected to be in the mid to high single digits, driven by continued growth across all businesses.

Dynamics: Revenue growth is expected to be in the low to mid-teens, primarily driven by Dynamics 365

• Personal Computing: Expected revenue growth in constant currency is projected to increase by 10% to 13%.

• Windows OEM: Revenue growth is expected to be in the low to mid single digits, as PC market unit volumes continue to remain at pre-pandemic levels.

• Windows Commercial and Cloud Services: Revenue growth is expected to be in the mid single digits, driven by demand for Microsoft 365 and advanced security solutions.

• Devices: Revenue is expected to decline by mid-teens percentage points, as the company continues to focus on high-margin premium products.

• Search and News Advertising: Revenue growth, excluding traffic acquisition costs, is expected to be in the low to mid teens percentage points, driven by sustained traffic intensity.

• Gaming: Expected revenue growth in the low to mid forties percentage points, with Xbox content and services expected to grow revenue in the high fifties percentage points, including the impact from the acquisition of Activision.

• Tax Guidance: Effective tax rate for the fourth quarter is expected to be approximately 18%.

• Dividend Policy: Will continue to return capital to shareholders through dividends and stock repurchases.

• Foreign Exchange Impact: Foreign exchange is expected to have a minimal impact on revenue growth in future quarters.

• Capital Expenditures: Investing in AI and cloud services.

2.2 Q&A Analyst Q&A

Q: Microsoft's capital expenditure plan for this year is expected to grow by over 50%, exceeding $50 billion, and there are rumors of even more spending, with reports suggesting data centers worth up to $100 billion. Would like to understand how the management team evaluates these investment potentials, and also the authenticity of the $100 billion data center rumor?

A: Our management team discusses two aspects: training and reasoning. In terms of training, we aim to maintain a leading position in this technological transformation, so we invest capital in training large-scale foundational models. At the same time, we also focus on reasoning needs and invest in product innovation and infrastructure business. We have been allocating capital in this way for many years, becoming a leader in the field of artificial intelligence, and plan to continue with this plan. It is important to consider our planning cycle, and we are indeed gradually increasing spending to meet demand.

Regarding opportunities and their scale, I think it is important to consider each business process that may be affected and the opportunities represented by each business process. Therefore, when thinking in this way, the opportunities are enormous. It is very important to drive the next wave of "cloud infrastructure", as we have been a leader in cloud transformation for the past decade, and we are confident in continuing to invest in the second wave, drawing on our successful experience in the first wave. Our spending approach is the same as in the past decade: observe signals, invest in becoming a leader in technology foundation, and then continue to add value for customers.

Q: Azure's bookings accelerated year-over-year this quarter, but other vendors are cautious about future performance. What is your view on customers' budget situation this year?

A: From the demand side, first of all, Azure is doing very well, as it has become the preferred platform for almost everyone doing any AI projects. This has been very helpful for us in acquiring new customersNext, we also see that AI projects are not limited to AI models, but also involve vector databases. Azure Search is one of our fastest-growing services, even ChatGPT is using it. We have integrated it with Azure AI, so Cosmos DB is also included. Therefore, the data layer and development tools are another area where we see great appeal. Finally, I want to mention the migration to Azure, which is not just an AI story. We are also focusing on customers. We always talk about the existence of an optimization cycle: as people optimize, they also spend money on new project launches, these projects grow, and then they optimize again. It's an ongoing process.

Q: When do you think artificial intelligence will reach a mature level and become a net growth in IT spending or beyond IT spending? What are the main maturity indicators? Are some projects being delayed to shift spending from core Azure to Azure AI?

A: First, a good starting point is to observe the standard issues of software teams. In the past, they purchased tools, now it's basically purchasing tools plus Copilot. This shift can actually be described as converting OpEx spending into tool spending, as it provides operational leverage for all OpEx spending. We have seen similar trends in areas such as customer service, sales, and marketing. This knowledge shift can be seen as a lean approach to long-distance work, as it reduces waste, increases speed, and customer value.

However, cultural change is a significant limiting factor within organizations. Companies need to simplify, automate processes, and apply these solutions, which require not only technology but also a change in company culture. We see the impact of this transformation in software development, customer service, and other areas. We see this transformation in customers we closely work with and in deployments within Microsoft. This change takes time to spread throughout the economy, but its spread is faster than ever before. Nevertheless, we still need workflow and process changes.

So, are we seeing project launches shifting from core consumption to AI projects? Based on our results, this is not the case. What we see more of is the migration of growth, work in the data space, and the launch of AI projects, which is also why our growth differs from IT budget spending. This is an improvement in share, possibly in areas that we traditionally do not think would be included in the IT budget spending of CIOs, spending by personnel such as customer service executives and marketing executives.

Q: Copilot has an impact on many different industries. How do you think this will affect industry development? When you provide specific Copilot, and other companies follow suit, how will your partner strategy be affected?

A: What we see is that office use involves various business processes, such as sales driving, customer service, revenue operations, and supply chain management. Now, we can integrate knowledge work tools and business process data, such as integration with ServiceNow, SAP, Salesforce, and DynamicsThis means we are building a Copilot that can act as a coordinator for other Copilots, while connecting to various business applications and knowledge work tools to provide more comprehensive support and integration.

Q: Azure's stability has been mentioned in the past few quarters, can you elaborate on the commercial booking data? Considering its variability, can you describe your findings in cost optimization and core workload growth?

A: We have been monitoring Azure's stability, with AI contributing seven percentage points this quarter, and the remaining 24 percentage points coming from core Azure business. Within the core business, we have seen new workload launches and optimizations, which is a common balancing state in the cloud transformation process. Compared to the second quarter, the acceleration seen in the third quarter mainly comes from the launch of new projects, not just AI projects, but also including migrations from on-premises to the cloud. Overall, this shows a fairly balanced trend with no significant industry or regional differences. As to whether we should continue to emphasize stability, I will evaluate the situation based on each workload. This quarter is quite balanced, and optimizations seem normal, which is crucial as we encourage customers to run their workloads as efficiently as possible, essential for customers to achieve growth and derive value from it.

Q: Can you share your thoughts on the data quality issue that investors often discuss? Have you seen progress in addressing this issue within the company? Does this pose a hindrance to the growth of artificial intelligence?

A: There are two key factors to successfully use these new AI capabilities. Firstly, AI provides a new user experience as a natural language interface; secondly, it is a reasoning engine that requires good data support. People can now use tools like retrieval-augmented generation to fine-tune models and further optimize them. The use of these tools is becoming more common, and developers and system integrators are also helping companies address these issues. So, in the enterprise space, we are very optimistic, while in the consumer space, we may need more improvements to develop more complex scenarios.

Considering GitHub is a complete system that includes not only AI models but also components such as user experience building, editors, chat, interpreters, and debuggers. These parts create inference traces together with the models to help the entire system run. We are creating scenarios similar to GitHub Copilot for every business system through tools like Copilot, Copilot Studio, and connections to all business systems. This will bring commercial value and a better foundation for enterprises. At the same time, we are collaborating with systems like Fabric, Cosmos, Postgres, or SQL to prepare data for integration with these AI projects.

Q: I would like to ask about the artificial intelligence aspect of Microsoft 365 Copilot, you have already mentioned some positive impacts on the Office business in the quarterly report. Explain how the lifting of capacity constraints affects our investments and drives more capacity online, and how much impact this has on enhancing the capabilities of Azure's AI business and Microsoft 365 Copilot business?**

A: I don't think Copilot has capacity constraints. Our primary task is to ensure optimized capacity allocation to support the continuous growth of user businesses. Therefore, when we encounter capacity constraints, they mainly occur in Azure infrastructure and business consumption.

Risk Disclosure and Disclaimer for this article: Dolphin Research Disclaimer and General Disclosure