Amazon released its Q2 FY2024 financial report today: net sales were $148 billion, up 10% year-on-year, and up 11% year-on-year excluding exchange rate effects; net income was $13.5 billion, up 100% year-on-year; diluted earnings per share were $1.26, up significantly from $0.65 in the same period last year.

After the release of the financial report, Amazon CEO Andy Jassy, CFO Brian Olsavsky, and Director of Investor Relations Dave Fildes answered questions from investors.

The following is a transcript of the conference call:

Goldman Sachs analyst Eric Sheridan: My question is about Amazon Web Services (AWS). Over the past few weeks, everyone has been discussing whether there is overinvestment in AWS, or whether existing investments are still insufficient. Andy, I am curious about your thoughts on this. What are your thoughts on the investment pace for AWS in the long-term?

My second question is about the custom chips mentioned by management earlier. What are your thoughts on custom chips? From the perspective of investment pace and return on investment, what do you think is the long-term return on investment for custom chips?

Andy Jassy: Regarding investment in artificial intelligence (AI) on AWS.

First of all, one of the most misunderstood issues with AWS over the past 18 years has been that we face huge operational challenges in operating such a large business. AWS's operations cover approximately 35 geographic regions, each of which can be viewed as multiple data center clusters; there are also about 110 availability zones, with each zone being equivalent in size to one or more data centers. From this perspective, it is very difficult to provide accurate services to users in nearly 200 AWS availability zones. If the capacity is too small, it cannot meet the service in the region, and enterprise users in the area may not be able to expand. No one wants this. So, people often provide more capacity than demand. But if the actual capacity provided is too much, the company's economic benefit will be discounted, and it may even have a negative impact on revenue.

From the AWS revenue report we shared with you, we can see that after such a long time of operation, we have found the investment balance point and achieved effective and reasonable management of the business. Through long-term exploration and construction, we have established a suitable algorithm model for AWS and found an investment balance point suitable for business development.

The same is true for AI investment, and we will learn from AWS business. Artificial intelligence is still relatively new for everyone. Indeed, some companies may choose to reduce capacity, but no company will suddenly demand hundreds of thousands of chips. Before making an investment, we need to receive signals from users and the market. With the passage of time, I believe we have accumulated sufficient technology and capabilities to build capacity and models that meet user needs.

We have also received a lot of signals from users and understand their needs. That's why we choose to invest heavily in AI capacity and infrastructure, hoping to further expand existing capacity. The fundamental reason for this is that we see a very large user demand. For us, this is a very large business.

As for custom chips, the progress in this area is very interesting. We also draw on experience from AWS's 18-year operating strategy and methods.

I remember when AWS was just launched, we established a very strong partnership with Intel in the general GPU field, which is still the case today. But from customer feedback, we found that everyone wants to find high-value, high-return products, which may have low unit costs in this process, but the total expenditure has not been reduced much. As spending more and more money, everyone hopes to get products with higher cost performance. However, many times it is difficult for people to get ideal cost performance products in the existing choices, unless you choose to optimize the product on the basis of the existing product, and break through the limit through your own capabilities.

This is exactly what we hope to achieve. With the fourth-generation Amazon Graviton processor, we can help users achieve chip customization. This is very successful for both users and our own AWS business. Compared with other x86 processors, the fourth-generation Amazon Graviton processor can help users save 30% to 40% of costs and effectively improve cost performance. In fact, five years ago, we saw similar situations in the field of accelerators, GPUs, etc.: the products on the market are very good, but the suppliers are scarce, and the products are often in short supply.

Users have always wanted to constantly improve product cost performance. That's why we chose to create Trainium training chips and Inferentia inference chips, and now both chips have been updated to their second generation. In today's market environment where GPUs are in short supply, these two chips have very high cost performance. Users are also very excited about this.

Currently, we are accelerating the production of these custom chips. I also believe that, like the Amazon Graviton processor, these chips will bring us good returns. At the same time, I also believe that our innovations will continue to empower AWS and make it stand out in the market.

Morgan Stanley analyst Brian Nowak: First, I see that Amazon's retail business gross margin performance in the second quarter seems to be weaker than expected. Is the reason behind this discount? Or has the company's investment in the Kuiper project affected the retail business? What are the main driving factors behind the gross margin of the retail business this quarter according to management?

My second question is, Andy, you mentioned before that you want to optimize costs and hope to restore Amazon's business profitability in North America to pre-pandemic levels. Can you share with us what ideas you have to execute and achieve this goal? Have you set specific time points or schedules for achieving this goal? In the next few years, Amazon plans not only to increase profitability, but also to invest in new projects such as Kuiper, and also to reinvest in some profitable projects. How does management find a balance between these works?

Brian Olsavsky: Regarding the North American profit margins you mentioned in your first question.

Compared with the first quarter of this year, our profit margin in the second quarter indeed decreased by 20 basis points. But I need to remind everyone that we raise stock compensation at the end of the first quarter every year. Therefore, compared with the first quarter, our stock compensation costs increased by about $1.8 billion in the second quarter. So to a certain extent, this also affected our profits. However, even if the stock compensation costs increased this quarter, the profitability of our North American stores still increased compared to the previous quarter. We have made progress in cost optimization, speed improvement, and category expansion. It can be said that we have maintained a strong momentum in many key areas.

Compared with the first quarter, in the second quarter, we increased our expenses and investment costs in some areas. This is a normal phenomenon. Generally speaking, the first quarter is often the quarter when we invest the least, and we do not invest much in projects such as Prime Video during this period. One more thing I want to point out is that compared with the first quarter, in the second quarter, we will increase our investment in the Kuiper project. The relevant satellite construction work has begun, and we hope to launch satellites in the third and fourth quarters of this year.

Andy Jassy: Let me answer your second question.

Both I and our team have always believed that we have the opportunity to further improve the profitability of our physical stores business. I have mentioned in several conference calls before that this goal will not be suddenly achieved in a quarter. This is not a one-time process, but requires long-term efforts. The epidemic a year and a half ago forced us to re-examine ourselves, and even forced us to break many long-held ideas.

This epidemic also gave us the opportunity to reduce costs. One of the primary work we have done in the past year and a half is to regionalize the Amazon delivery network in the United States. The impact this brought us was far greater than originally imagined. Of course, this work is not completely finished yet, and there are still many areas that can be continuously optimized. I believe this regionalized management method will effectively help us reduce service costs.

At the same time, we have also found many other areas where we can reduce costs while improving user experience. One of the greatest advantages of regionalization is that it not only helps us reduce service costs, but also significantly improves our product delivery speed. In addition, through regionalization, we are able to bundle different products ordered by the same user into the same box, again reducing costs. Many users praise this delivery method as being more environmentally friendly.

From the perspective of operating strategy and philosophy, reducing service costs is also a good thing for us: reducing service costs means that we have the ability to provide users with more product choices. In the past, we did not have inventory of many low ASP goods because stocking such goods was not profitable for us with the service costs at that time. But with the progress of our service cost reduction work, we are able to provide users with more and richer product choices; and as the number of product choices increases, we also find that more and more users choose to shop on Amazon. This creates a positive cycle.

As for your question about whether the Kuiper project has affected other businesses. What I want to say is that projects are not binary opposites. There has never been a situation where projects are opposed to each other in Amazon's many years of development. I have been working at Amazon for 27 years, and I believe we have the ability to balance both projects, which are not mutually exclusive. We will have different work teams on different projects-the store team is responsible for continuing to expand product categories, maintaining low prices, improving delivery times and reducing service costs; The Kuiper team is responsible for helping 400 million to 500 million homes worldwide without broadband connections receive broadband signals. So in my opinion, the relationship between projects is not binary opposites. We will promote both the Kuiper project and cost reduction work at the same time.

Evercore ISI analyst Mark Mahaney: Andy, you mentioned in the briefing that three macro trends drove the accelerated growth of AWS in the second quarter (1. The cost optimization work is nearing completion and new work has begun; 2. The company's energy is returning to the modernization of infrastructure while migrating on-premises facilities to the cloud. 3. People are excited about the artificial intelligence business). In my opinion, these trends seem to be sustainable. Does management also think so? Does this mean that the AWS business can achieve sustained accelerated growth in the second half of the year?

My second question is about Amazon Pharmacy business. Although this may only account for a small part of Amazon's business, according to surveys, consumers are very interested in this business. Could management talk about how you position Amazon Pharmacy business? At what stage is the business currently developing? Does management plan to expand the audience coverage of this business?

Andy Jassy: Regarding AWS. It's actually really hard to predict exactly what the growth rate is going to be. Right now, our AWS business is relatively large with a revenue run rate of $105 billion.

Most of our cost optimization work is already done. Before the outbreak of the pandemic, most companies were working to modernize their infrastructure, which means everyone wants to migrate their local work to the cloud to save money, speed up innovation, and increase developer productivity. With the outbreak of COVID-19, everyone's life has become more difficult. Many companies have been affected by macroeconomic headwinds and hope to save costs and increase efficiency. Thus, more and more people are beginning to ask themselves why they should spend time running their own data centers, why not choose cloud computing services and fundamentally save costs and speeding up innovation? AWS's rich features, powerful performance, and security are also very important for users. AWS is the best choice for users. As people want to transition into cloud computing, more and more users hope to establish deeper relationships with our ecosystem. Our AWS generative AI components are still in the early stages, but we have achieved several billion dollars in revenue run rates in this area, which is very interesting. Although everything is still too early, we have observed significant user demand.

As outlined in the briefing, the three macro trends I mentioned are likely to continue over time. We also look forward to seeing what kind of revenue run rates our AWS business can achieve under the driving force of these three trends.

I mentioned earlier that the current revenue run rate of Amazon's AWS business is around $105 billion, but it is worth noting that about 90% of global IT spending is still stagnant locally. I believe the situation will definitely change in the future. From this perspective, AWS will have great development potential in the future. Moreover, both generative AI and AI itself have great development prospects in the future. In the non-AI field, users may need to spend a long time building infrastructure locally and then migrate it to the cloud through cooperation with the user. This is a very difficult task. In the generative AI field, all work is basically built in the cloud from the beginning.

As for Amazon's pharmaceutical business. Indeed, our business has achieved sustained growth and has been supported by more and more users. I always think it is a natural thing to extend the retail business to the pharmacy field. Our business has grown very rapidly. In the past 18 months, our team has made a lot of progress in terms of user experience, and the performance now proves their efforts. Users really enjoy the shopping experience of Amazon Pharmacy. For example, compared to walking into an offline physical store pharmacy, in Amazon's online pharmacy, you can easily order the required medications. After all, the shopping experience of most physical pharmacies today is not good. Many medications are locked in cabinets, and you may have to push the call button for someone to come and help open the cabinet. In addition, many pharmacies still have theft incidents.

Furthermore, by drawing on the experience of offline physical pharmacies, we are continuously improving the user experience of Amazon's online pharmacy, attracting more and more users to choose us. I believe everyone has also seen our progress in this area. We have expanded the coverage of the Prime prescription drug subscription plan RxPass and included the Medicare population in the plan. As a result, eligible users can purchase up to 60 common medications for only $5 a month. In addition, we have also launched the "same-day delivery of medicines" service in eight cities including Los Angeles and New York and plan to expand the service to more than a dozen cities by the end of this year. We are very optimistic about the growth prospects of this business.

Jefferies Financial analyst Brent Thill: My first question is about AWS. Can management please share with us the number of backlog orders of this quarter? In addition, Andy, I know that AWS's AI features are still in the early stages. But as more and more companies move to the public cloud, do you see an increase in AWS workloads? Even if many companies are not yet ready to use AI features, they also need to prepare for the day. What trends can you see from AWS's subscriptions and revenues?

Dave Fildes: At the end of the second quarter, our backlog orders were $156.6 billion, up about 19% year-on-year.

Regarding your second question, this is the case both in terms of data and the real significance of the AI field.

If users want to use AI features effectively, the data you provide must be rearranged so that data access, operation, and model building can be implemented in the cloud. So, whenever we communicate with customers, whether it is direct or indirect cooperation, most users are particularly eager to use AI functions and wish to use them as soon as possible. In this case, the first question we often ask is: Where is your data? What is the data pool? How to access the data? We often need to reorganize the data before formally using AI functions. Of course, fortunately, many companies have completed the work of data migration to the cloud. These companies are ready to meet AI. A large part of AWS's business growth comes from this.

But don't forget, as I mentioned earlier, about 90% of IT spending worldwide is still stuck in local areas. Many companies' data and infrastructure have not yet migrated to the cloud. But I believe that one day they will migrate to the cloud, and I believe that the arrival of AI functions will drive these companies to take the initiative to migrate to the cloud.

TD Cowen analyst John Blackledge: I have two questions. First, I see that the profit margin of AWS this quarter is very strong, at around 35%. What are the key driving factors behind management? What is the trend of AWS profit margin in the second half of this year? But at the same time, I see that the profit margin of Amazon's international business revenue has slightly decreased. What are the reasons behind management?

Brian Olsavsky: Regarding the profitability of AWS.

Indeed, the profit margin of AWS business in the second quarter is around 35%, and the data last year was about 25%. There are many reasons behind the increase in profits. First of all, we have taken a series of cost optimization measures, which have effectively improved business efficiency, and these have brought positive effects to our profit margin. In addition, in the first quarter, we also adjusted the service life of the server, and we mentioned this in the previous quarter. This will bring a YoY increase of about 200 basis points to our profit margin. In the future, we will continue to optimize the cost structure. Of course, you should pay attention to the fact that the business profit margin often fluctuates between different quarters, which is normal. Next, we will continue to develop new products and functions, attract more new users to choose AWS, and at the same time optimize costs and improve efficiency.

Dave Fildes: Regarding the international business issues you mentioned.

This quarter, our profit is about $0.3 billion. From the perspective of profit margin, our profit increased by about 390 basis points YoY.

We have shared with you before that Amazon's business in different countries is actually in different stages, and it may be mature in some countries, but not in others. Therefore, we will continue to focus on international business.

Compared with the same period last year, our international business has made some progress. Among them, the United Kingdom, Germany, and Japan make up a larger proportion of our international business. In these markets, we will continue to promote business scale, focus on improving operational efficiency, and optimizing user experience. Amazon has landed in about 10 countries and regions in the past seven years, and in these emerging market, our main task is to improve the user experience and gradually build an Amazon membership system that is suitable for local use.

In short, I believe that we are making progress every year in both mature and emerging markets, and we hope to continue to improve Amazon's international business performance.

JPMorgan analyst Doug Anmuth: In the guidance given by management for the second quarter, I saw that management has taken macroeconomic pressures into account, especially in the European market. The specific category affected includes computers, electronic products, and possibly televisions. My question is whether management thinks that macro trends will develop as you expected. Is the reason for giving the above guidance because you see weak performance trends? How will this affect management's outlook for the third quarter?

My second question is, Brian, can you quantify for us how much incremental investment Amazon has made in the Kuiper project?

Brian Olsavsky: First, about the macro environment issue.

In fact, Andy has shared his thoughts with everyone before. Over the past year, we have seen many similar consumer trends: consumers have become more cautious and more eager to buy inexpensive, discounted, and low-ASP products. This trend has also been reflected in the second quarter of this year, and we expect it to continue into the third quarter. However, the second quarter is slightly different in that we have achieved strong unit sales growth. Considering the impact of the leap year, the actual unit sales growth in the North American region in the second quarter is slightly accelerated.

However, the growth of revenue is closely related to the average unit price and existing consumer trends. Currently, our sales of high-priced products are basically stable, but they are far less than sales in a normal economic environment. In addition, due to the growth of daily necessities, low-ASP products occupy the majority of market share. We welcome this very much, because for this part of the product, we can achieve rapid delivery, and we hope that consumers can choose to buy these daily necessities on the Amazon platform.

Regarding the incremental investment in the Kuiper project, we will not give specific figures today, but I am very grateful for everyone's attention to the Kuiper project. (End)