Microsoft's financial report for the first fiscal quarter of fiscal year 2025: Revenue was 65.585 billion US dollars, up 16% year over year; net profit was 24.667 billion US dollars, up 11% year on year, and 10% year on year without taking into account the impact of exchange rate changes; diluted earnings per share were $3.30, up 10% year on year. Excluding the impact of exchange rate changes, the same increase was 10% year on year (note: Microsoft's fiscal year was not synchronized with the natural year).
After the financial report was released, company executives such as Microsoft CEO Satya Nadella (Satya Nadella), Executive Vice President and CFO Amy Hood (Amy Hood), Chief Accounting Officer Alice Jolla (Alice Jolla), and Deputy General Counsel Keith Dolliver (Keith Dolliver) attended the subsequent earnings conference call to interpret the financial report highlights and answer questions from analysts.
Here is a transcript of the conference call:
Morgan Stanley Analyst Keith Weiss: The expansion of generative AI-related functionality, the speed of innovation, and significant opportunities is the most exciting situation I've seen in my 25 years in the software industry. Judging from what was said in front of management during this conference call, it appears that you feel the same way. However, in my conversations with investors, I also heard two of their concerns. Both questions are related to factors limiting the future development of generative artificial intelligence technology.
The first question is, what are the internal limitations or impediments Microsoft has when it comes to making related investments, particularly in funding future generations of the underlying model? Because we estimate that the amount of capital involved could be as high as tens of billions or even more than 100 billion dollars. Another question is what external constraints have Microsoft encountered in expanding production capacity to meet demand and seize opportunities, particularly in building a sustainable power supply capacity for new data centers.
Satya Nadella: Regarding the first question, when we consider capital expenses for AI training — which should be the general meaning of what you mentioned — this expenditure may be limited by circumstances related to inference business revenue generation. As in the past, we allocate capital to build our cloud services based on the market demand signals we see, then we forecast demand based on this and carry out additional construction accordingly. This is also the case in the training business. As we build next-generation, more powerful models, our products themselves will drive more inference requirements, so even if the investment was very intense during some periods, overall, the investment pace must have returned to normal.
I think the best perspective to think about this question is the effectiveness of Moore's Law in chip and system development, so not only calculation, but also computational efficiency, data, and algorithms. Everyone wants to maintain that curve, update equipment according to Moore's Law every year, and then effectively depreciate during its life cycle. Ultimately, reasoning requirements will determine how much we invest in training, because I think in the end, the limiting factor must be demand.
Regarding the external factors you mentioned, we have clearly encountered many external limitations, because market demand is emerging very fast. If you think about it, the most popular artificial intelligence products of this generation use our cloud services, whether it's ChatGPT, Copilot, GitHub Copilot, or Tax Copilot. These are the most commonly used products in or around our ecosystem. As a result, we have encountered a series of limitations. The data center cannot be built overnight, so there are data center related issues, including power issues, etc., but these are all short-term limitations. For example, in the second quarter, some demand-related issues or problems with the ability to meet customer needs were actually all caused by external and third-party factors, and we are at least gradually solving these problems.
But in the long run, we do need an effective power supply and data centers. Some of these problems will take longer to resolve. The good news is that in the second half of this fiscal year, some supply and demand will be perfectly matched.
Jeffrey Analyst Brent Thill: Amy, I'm happy to hear that the Azure business will accelerate again in the second half of the year. I'm guessing many people have this kind of question. The 34% growth rate in the first fiscal quarter may have fallen to around 30% in the second fiscal quarter. I know this may be related to changes in the growth base, but in addition to the high growth base for the second fiscal quarter of last year, did the company consider other factors that led to the slowdown in growth in the second fiscal quarter?
AMY HOOD: I'll first restate some of the points I've made before, and then combine them a little bit to answer your question. We achieved a 34% growth rate in the first fiscal quarter, and we expected 33%. The excess was mainly due to revenue confirmation. I will look at this issue from the perspective of pure consumption and artificial intelligence. We expect a drop of one or two points. The main reason for this is the supply delay mentioned by me and Satya. In terms of basic consumption growth, it was actually stable from the first fiscal quarter to the second fiscal quarter.
As for some of the details you mentioned, there must have been some ups and downs, but we are confident because our supply will increase dramatically in the second half of this fiscal year, especially in terms of artificial intelligence, which can better match supply and demand, and there will also be an acceleration in growth in the second half of the year that we mentioned. I'll also take this opportunity to say that when talking about the usage of artificial intelligence workloads, people always tend to think that graphics processing units (GPUs) are the most critical issue, but in fact, cooperative operation of GPUs and central processing units (CPUs) is also important, so this is also a factor that accelerates in the second half of the year.
Bernstein Analyst Mark Moerdler: Investors are clearly concerned about capital expenditure growth and spending direction. My understanding is that half of capital expenditure is long-term, so I'd like to ask management to share their views on capital expenditure growth? Will Microsoft's capital expenditure return to the traditional model, where the growth of capital expenditure is roughly in sync with or slightly slower than cloud business revenue? If this is the case, can I introduce an approximate point in time, such as whether we will have enough facilities to be put into use by some point next year?
Amy Hood: Looking back at the cloud business transformation that companies have been pursuing in the past ten years or so may be very useful in understanding the issues you mentioned. In the early stages, you saw the work we did, including now we are doing the same thing, which is to continuously build facilities to meet market demand. Unlike cloud business transformation, due to the nature of the needs, we are currently expanding globally simultaneously rather than sequentially. And as long as we continue to see demand growth, capital expenditure growth will slow, but revenue growth will accelerate.
As you said, the growth rate of these two will get closer and closer as time goes by, and the level of growth actually depends entirely on the speed at which new technology is adopted. As Satya mentioned, some of this expenditure will be used to build the next training infrastructure, but will not be included in the cost, but will be included in the operating expenses. Overall, this is a balanced and reasonable method of capital expenditure. Just like the company's previous development cycle, the two will get closer and closer.
UBS analyst Karl Keirstead: I'd like to ask Satya and Amy a question about OpenAI. Investors have been overwhelmed by numerous media reports about OpenAI and Microsoft since Microsoft announced its investment in the company three months ago. I'd like to ask management to elaborate on the relationship between the two companies. All of us have received some signals from various channels. Perhaps Microsoft wants to diversify to a certain extent at the model level and provide customers with more choices. Also, regarding the digital aspects of financial reports, Amy, how can Microsoft help OpenAI achieve its expansion plans and capital expenditure requirements? And how is the company dealing with the impact on other revenue items you just mentioned?
Satya Nadella: For OpenAI and Microsoft, this partnership has always benefited us a lot. After all, our investment in this company four or five years ago was a real bet on innovation, supporting one of the most valuable private companies on the market today. This investment has been a huge achievement for Microsoft and a huge success for OpenAI.
On this basis, Microsoft has provided world-class infrastructure for OpenAI. Using this infrastructure, OpenAI continues to innovate models. We are also carrying out some training-related work at the model level, as well as innovating in building small models and all product innovations. Products like GitHub Copilot (artificial intelligence programming assistant), or the release of DAX Copilot (medical artificial intelligence assistant) or M365 Copilot (data and artificial intelligence assistant) have strengthened my belief in OpenAI and the business they are doing, and there will be more excellent innovative combinations in the future.
As investors, we are very satisfied with Microsoft being able to participate in the development of OpenAI. The point is that in such a partnership, the two parties always maintain dialogue to ensure mutual success, which means we need to push each other to do more and seize this rare opportunity. This is our cooperation plan, and we plan to continue to develop the relationship between the two parties on this basis.
(Continuously updated...)
