Microsoft: How to control the pace of AI investment

The following is a summary of Microsoft's second quarter 2024 earnings conference call. For an interpretation of the financial report, please refer to " Microsoft: No miracles for Big Power, AI dream stranded? 》

I. Review of Core Financial Information:

II. Detailed Content of the Financial Report Conference Call

2.1. Key Points from Executive Statements:

1. Business Progress

Azure Business: a. Market share accelerated growth driven by AI, expanded global data center coverage; b. Introduced new AI accelerators from AMD and NVIDIA, launched Azure Maia and the new Cobalt 100; c. Azure Arc has reached 36,000 customers, a 90% year-on-year growth; d. Azure AI has over 60,000 customers, nearly 60% year-on-year growth.

Data Platform: a. The number of customers using Microsoft's intelligent data platform data and analytics tools grew by nearly 50% year-on-year ; b. Microsoft Fabric has over 14,000 paying customers, with a 20% quarterly growth.

GitHub: a. GitHub Copilot adopted by over 77,000 organizations, a 180% year-on-year growth ; b. Copilot is driving overall growth for GitHub. GitHub's annual revenue run rate is now $20 billion. Copilot accounts for over 40% of GitHub's revenue growth.

Power Platform: AI-driven features in Power Platform used by over 480,000 organizations, with a 45% quarterly growth.

Microsoft 365: Microsoft 365's Copilot customer count grew by over 60% quarterly.

Security Business: a. Over 1,000 paying customers using Security Copilot; cloud security solution Defender for Cloud generated over $1 billion in revenue in the past 12 months.

Copilot Business: a. Microsoft 365's Copilot customer count grew by over 60% quarterly, with the number of people using Copilot for work doubling quarterly; Teams Copilot: the new Teams Copilot can facilitate meetings and create and assign tasks; b. Copilot Studio: Copilot Studio's quarterly growth exceeded 70%, with 50,000 organizations using it, including Carnival Corporation, Cognizant, Eaton, KPMG, Majesco, and McKinsey.

c. DAX Copilot in the healthcare industry is used by over 400 healthcare organizations, with a quarterly growth of 40%; the number of AI-generated clinical reports has tripled.

LinkedIn Business: LinkedIn's member growth is accelerating, with record engagement, sharing 1.5 million pieces of content per minute on the platform, and a 34% year-on-year increase in video uploads.

Search Advertising and News: a. Bing, Edge, and Copilot drive more engagement and value for users, publishers, and advertisers; b. Record engagement is driven through Copilot for the web, with consumers creating over 12 billion images and engaging in 13 billion chats using Copilot.

Devices: Introducing the new Copilot+ PC category, the fastest and smartest Windows PCs ever; active devices on Windows 11 increased by 50% year-on-year.

Gaming Business: a. Over 500 million monthly active users across platforms and devices; b. Previewed a record 30 new titles this quarter, with 18 available on Game Pass; c. "Fallout" TV series premiered on Amazon Prime, with nearly a 5x quarterly growth in gameplay time for the "Fallout" series on Game Pass.

2) Financial Performance

① Remaining Performance Obligations (RPO) increased by 20%, or 21% at constant exchange rates, reaching $269 billion, with approximately 40% of revenue expected to be recognized in the next 12 months, an 18% year-on-year growth.

② Activision Blizzard Impact: Activision Blizzard's net impact on revenue growth is approximately 3 percentage points; a 2 percentage point drag on operating income growth, negatively impacting earnings per share by $0.06.

③ Microsoft Cloud: Microsoft Cloud revenue was $36.8 billion, a 21% year-on-year growth, or 22% at constant exchange rates; gross margin decreased by around 2 percentage points to 69%, primarily due to sales shifting towards Azure.

④ Capital Expenditure and Cash Flow

a. Capital expenditure (including finance leases) was $19 billion, with cash outlays of $13.9 billion for property, plants, and equipment purchases.

b. Operating cash flow was $37.2 billion, a 29% year-on-year growth.

c. Free cash flow was $23.3 billion, an 18% year-on-year growth.

3) Future Outlook

① Full Year 2025 Outlook

a. Foreign exchange is not expected to have a significant impact on full-year revenue, cost of sales, or operating expenses.

b. Double-digit revenue and operating income growth are expected, focusing on providing differentiated value to customers.

c. Capital expenditure is expected to be higher than in the 2024 fiscal year, primarily for expanding AI and cloud infrastructure d. It is expected that the growth in operating expenses will remain in single digits, and the operating profit margin for the 2025 fiscal year will decrease by about 1 percentage point year-on-year.

e. The effective tax rate for the 2025 fiscal year is expected to be around 19%.

② Outlook for the first quarter of the 2025 fiscal year

a. It is expected that foreign exchange will result in a growth reduction of less than 1 percentage point in total revenue and departmental revenue.

b. It is expected that foreign exchange will lead to a decrease of less than 1 percentage point in the growth rate of sales costs, with no significant impact on the growth of operating expenses.

c. It is expected that the gross margin of Microsoft Cloud will be approximately 70%, a year-on-year decrease mainly due to the expansion of AI infrastructure.

d. Capital expenditures are expected to increase quarter by quarter to meet the demands of cloud and AI.

③ Sector Guidance

a. Productivity and Business Processes: Revenue is expected to grow by 10% to 11% at constant exchange rates, i.e., $20.3 billion to $20.6 billion.

i. Office 365 revenue is expected to grow by approximately 14% at constant exchange rates.

ii. Office consumer revenue will grow in the low to mid-single digits.

iii. LinkedIn revenue will grow in the high single digits.

iv. Dynamics revenue will grow in the low to mid-single digits.

b. Intelligent Cloud: Revenue is expected to grow by 18% to 20% at constant exchange rates, i.e., $28.6 billion to $28.9 billion.

i. Azure revenue is expected to grow by 28% to 29% at constant exchange rates.

c. Personal Computing: Revenue is expected to grow by 9% to 12% at constant exchange rates, i.e., $14.9 billion to $15.3 billion.

i. Windows OEM revenue growth is expected to be relatively flat.

ii. Windows commercial products and cloud services revenue will grow in the mid-single digits.

iii. Device revenue will grow in the low to mid-single digits.

iv. Search and News advertising revenue (excluding traffic acquisition costs) will grow in the mid to high single digits.

v. Gaming revenue will grow by around 30%, including a net impact of approximately 40 percentage points from the acquisition of Activision Blizzard; Xbox content and services revenue is expected to grow in the low to mid-50%.

2.2 Analyst Q&A

Q: There is currently a debate in the industry about whether the capital expenditure requirements and monetization of generative AI are aligned. Is Microsoft's capital expenditure still an appropriate leading indicator of cloud growth? Can you explain the time relationship between capital expenditure investment and return on investment?

A: We mainly consider capital expenditure from the demand side, focusing on the form of the product portfolio and the realization of customer value. This is also a transformation that is knowledge-intensive and capital-intensive. We need to have the right product mix, including infrastructure, metering, and SaaS applications. This is what we need to consider first. For example, M365 Copilot, GitHub Copilot, and Dynamics have performed well in the context of contact centers, and the growth of Azure AI is our primary focus. These demand signals are driving most of the capital expenditure. Over 60% of capital expenditures are used for inference and other equipment, and purchases are only made when there are demand signals. Our assets are long-term assets, including land and data centers, which can be partially built to flexibly respond to demand. Training expenses will also be adjusted based on demand signals We are satisfied with Microsoft's extensive product portfolio, whether in the consumer sector, commercial per seat charges, or consumption measurement, these are fundamental driving factors.

The timing and some issues you mentioned did lead to the capital expenditures mentioned in my comments today. We talked about half of the total capital expenditures for the 2024 fiscal year and half for the fourth quarter, mainly for land construction and financing leases, which will be monetized over 15 years or longer. Our architecture is unified, first with the commercial cloud, followed by Azure Stack for AI, whether the demand is at the platform level, application level, through third parties and partners, or our first-party SaaS, all use the same infrastructure. This is a long-term flexible asset.

Looking at the issue this way, you will understand the importance of building this network synchronously globally. Due to global demand, we are building globally, with large customers in each region. Hopefully, this form of capital expenditure can help everyone see its short-term and long-term monetization drivers.

Q: From an industry perspective, what else is needed for generative AI to become a more realistic technology? And how can it become more visible in SaaS products?

A: Generative AI is ultimately just software, and it is truly transforming into the growth of our M365 SaaS product, which is the new product Copilot SaaS, growing faster than any generation of software we have introduced before. The data we shared this quarter also indicates this. Seat growth has increased by 60% each quarter, which is very healthy. What is most exciting is that customers continue to repurchase more seats. The number of customers with over 10,000 seats has doubled quarter over quarter, which is a healthy SaaS core business.

In addition, Dynamics is also an area we are very excited about. We are gaining market share in this area. Generative AI embedded in Dynamics is fundamentally changing the business application category, with the contact center being a good example. We have saved hundreds of millions of dollars in customer support and contact center operations, and I believe we can bring this value to customers. On the Azure side, the data is very clear. We started providing this data last quarter, and you can see further acceleration this quarter. Another aspect is that AI does not exist in isolation. In fact, 50% of users using Azure AI are also using data meters, which is very exciting for us because the most important thing in Azure is winning enterprise workloads. This has already started to happen. These are generational changes, once they start, they will stay with you.

Q: Improving the profit margin of cloud services takes time, but AI seems different. Can you talk about the impact of AI investment on short-term and long-term profit margins?

A: Regarding the improvement of profit margins, it is different from the previous cloud cycle mainly because we have a consistent platform. We are building on the Azure AI stack, eliminating the need for multiple infrastructure investments. We make one investment, first for internal use, then customers and ISVs build on it, so the profit margin is better from the start and can expand steadily

Q: How to improve the efficiency of capital expenditure? You mentioned that 50% of capital expenditure is used for infrastructure and the other 50% for technology. Do you need to continue increasing capital expenditure at this high level, or can you reduce capital expenditure while maintaining stable revenue growth for Azure and generative AI?

A: I think there are two aspects to the question you raised. Firstly, can we achieve stable revenue growth without increasing capital expenditure? The answer is yes, because half of the capital expenditure is used for long-term construction, and the other half is used to adjust investments in CPUs or GPUs based on demand signals. If the demand signals change, we can adjust the investments in CPUs. We have been doing this for a long time and will apply the same strategy to GPUs. Additionally, you may have noticed the difference between our capital expenditure figures and cash outlays for property, factories, and equipment due to our use of leasing. Leasing appears as a one-time item on the books, leading to more volatility. I have mentioned this before, but since you brought up this issue, I'll emphasize it again as it may provide more context.

When considering capital expenditure, it is important to differentiate between leasing and construction. For construction, we consider the total cost percentage of each project, including land, networks, construction systems or equipment, and ongoing costs. Thinking in this way helps us adjust capital expenditure based on demand signals.

Q: This quarter's performance exceeded expectations by a smaller margin than before. Are there any abnormalities in terms of sales cycles or conversion rates?

A: Actually, there aren't any. When I talked about this quarter, I mentioned that commercial bookings were much better than we expected, and the execution of core annuity renewals and long-term commitments was also better than expected. So, I don't think there are any abnormalities in our business execution.

Q: Azure's growth of 30% this quarter is at the lower end of your forecast range. Can you provide more details on the reasons? I heard two points from your comments: one is that persistent capacity constraints may ease in the second half of the year; the other is that you mentioned slight softness in Europe, which I believe is more of an economic issue rather than an Azure-specific issue, is that correct?

A: Yes, that's completely correct. The guidance of 30% to 31% for the fourth quarter, we achieved 30%. Indeed, the difference lies in the slight softness in non-AI consumption we see in some regions in Europe, which has impacted the numbers. We expect this situation to continue into the first half of FY25, including my forecast of 28% to 29%. Additionally, persistent capacity constraints, especially in AI and Azure, will continue into the fourth quarter and the first half of the year.

Q: Azure demand has once again exceeded available capacity. When we think about cloud capacity and AI services, can you discuss your short-term and long-term strategies with AI partnership with companies like Oracle?

A: We are indeed facing AI capacity constraints. To address this issue, we are collaborating with third parties to expand the Azure platform to meet the demands of Azure AI. At the same time, we are making significant construction investments to restore balance. For us, this is no different from past leasing arrangements Even sometimes purchasing from Oracle may be more efficient because of shorter lease terms.

Q: With M365 Copilot already launched for several quarters, how do you evaluate Copilot's ability to enhance productivity, especially its impact on knowledge workers? For example, with the mention of a 10,000-seat transaction and repeat purchases, is it possible for Copilot's penetration rate in Office to eventually reach the level of GitHub?

A: The design system from GitHub and GitHub Copilot workspace design system are now also applied to M365 Copilot. For instance, a salesperson receives an email and wants to reply to a customer. The data in this email is actually contextual prompts, but you can expand the context by introducing all CRM data, and this type of workflow is already happening.

Then you can use Copilot Studio to further complete the workflow based on more data. So you can set it up that if an email from a specific customer has specific issues, it can be escalated to another person who will receive a notification in Teams. These workflows can be built by IT or end users themselves, which is the future extension of Copilot for us. Therefore, we see this as a new design system that brings productivity to knowledge and frontline work, just like what we see in software engineering.

Q: Regarding the gaming business, Xbox content service revenue grew by 61%, with Activision Blizzard contributing 58 percentage points, resulting in a net increase of about 3 percentage points. How should investors view the long-term growth potential in this area? You have made significant investments in the gaming sector, including the acquisition of Activision Blizzard. What are the considerations?

A: Our investments in the gaming sector aim to have the right product portfolio, including Xbox and its content, and expanding into PC gaming content. Activision Blizzard's product portfolio provides us with high-quality assets for PC, console, and mobile. We are now able to reach all major gaming platforms, including console, PC, and mobile.

We are also excited about new platforms. For example, in the most recent quarter, we expanded X Cloud to Amazon TV. This new touchpoint helps us reach new players or the same players playing games on different platforms. This will ultimately be reflected in our software and services and transactional revenue, which is our long-term KPI and the direction we are striving for. This is also the strategic significance of Activision Blizzard as an asset.

Our goal is to bring more content to more users and establish models similar to software annuities and subscription services, with enhanced transaction and intellectual property ownership, which is very valuable in the long term. As Satya mentioned, intellectual property ownership can be monetized in multiple ways. We are very encouraged by the progress of Game Pass and some new announcements.

Risk Disclosure and Statement of this article: Dolphin Research Disclaimer and General Disclosure