Amazon CEO's Annual Open Letter: The company's self-developed chips will help reduce AI costs, and this year's capital expenditure is expected to reach $100 billion

On Thursday, Amazon CEO Andy Jassy emphasized the company's massive investments in the AI field in his annual letter to shareholders. Jassy stated: "If you are committed to making customers' lives better and more convenient every day, and you believe that every customer experience will be reshaped by AI, then you must invest extensively and deeply in AI." Jassy pointed out in the letter that Amazon will continue to invest heavily in AI chips, data centers, and AI infrastructure: "AI is a historic opportunity that changes everything we know. The customer demand it brings is unprecedented, and our accelerated investment now is for the long-term benefit of customers, shareholders, and the business." Notably, his remarks were highly consistent with those of Alphabet CEO Sundar Pichai at the Google Cloud Conference the day before. Pichai stated that Alphabet plans to invest about $75 billion this year in building data center infrastructure and noted: "The opportunities brought by AI are the biggest we can encounter." Additionally, in light of the current reality that the costs of AI development and deployment remain high, Jassy stated that as technology evolves, the unit cost of AI usage will significantly decrease in the future: "AI does not need to be as expensive as it is now forever, and it indeed will not be that expensive in the future." He expects several factors to drive down AI costs, including more cost-effective AI chips, more efficient model compression and prompt caching technologies, and optimized computing infrastructure and model architecture design. Jassy also compared the development of AI to the revolutionary changes that Amazon Web Services (AWS) brought to computing and storage costs: "Cloud computing reduced costs, spurred more innovation, better customer experiences, and more infrastructure investment. AI will follow a similar path." Jassy disclosed that Amazon plans capital expenditures of up to $100 billion this year, most of which will be allocated to AI-related projects, covering data centers, networking equipment, AI hardware, and the development of generative AI service capabilities. Currently, Amazon has launched several AI products, including its self-developed large model Nova series, self-developed AI chip Trainium, consumer shopping chatbot, service platform Bedrock for third-party models, and a newly upgraded version of Alexa as a generative AI assistant. Amazon has also invested about $8 billion in the well-known AI startup Anthropic and integrated its AI model Claude into the "Alexa+" platform. Since taking over as CEO in 2021, Jassy has continued to push for business streamlining and cost control at Amazon. From 2022 to 2023, the company laid off more than 27,000 employees. In 2024, it continued with small-scale layoffs and closed several unprofitable projects, including the "Try Before You Buy" clothing service and the offline rapid delivery store pilot. Jassy stated that Amazon must operate like the world's largest startup: "We must maintain a culture of quick response, efficiency, and willingness to take risks, avoiding bureaucratic pitfalls." He has set a goal to increase the ratio of frontline employees to managers by 15% by the end of the first quarter of 2024, reducing redundant management layers. At the same time, he has opened an internal channel called the "Anti-Bureaucracy Email" to collect employee feedback. To date, nearly 1,000 emails have been received, leading the company to complete over 375 organizational and process optimization adjustments: "Creators hate bureaucracy... Once we encounter 'red tape', we must cut it without hesitation." From significant AI investments to "de-bureaucratizing" the organizational structure, and to careful adjustments in culture and governance, Jassy outlines the core strategic logic in one sentence: "Today's investments are to equip the future Amazon with the ability to create the next era." Additionally, Jassy discussed the Trump tariff issue during an interview with CNBC on Thursday. He pointed out that the company is still assessing the overall impact of President Trump's new tariff policies, but the vast network of third-party sellers on the Amazon platform may choose to pass on cost pressures to consumers: "I can understand why this would happen; I mean, no matter which country you're in, you don't have a 50% extra profit margin to squander. I think merchants will try to pass on costs." To address potential impacts, Amazon is also taking several measures, including strategically stocking up in advance, renegotiating procurement contract terms with some suppliers, and striving to maintain price stability for online goods. Jassy also revealed that the company has observed some consumers starting to stock up in anticipation of possible price increases in the future, but it is currently unclear whether this phenomenon will persist: "In certain categories, we have indeed seen people purchasing in advance, but it's hard to know if this is just an anomaly in the data, as it has only been a few days, or how long this situation will last." Below is the full text of Amazon CEO's annual open letter, translated by AI: 2024 is a strong year for Amazon. Our total revenue grew by 11% year-on-year, from $575 billion to $638 billion. By business segment, North America revenue increased by 10%, from $353 billion to $387 billion; international business revenue grew by 9%, from $131 billion to $143 billion; AWS (Amazon Web Services) revenue rose by 19%, from $91 billion to $108 billion. For reference, just ten years ago, AWS revenue was only $4.6 billion, while Amazon's total revenue that year was $89 billion. In 2024, Amazon's operating income increased by 86% year-on-year, from $36.9 billion (operating profit margin of 6.4%) to $68.6 billion (operating profit margin of 10.8%). Free cash flow, adjusted for equipment financing leases, rose from $35.5 billion in 2023 to $36.2 billion. Besides financial performance, we have also genuinely improved and simplified customers' lives. In our retail business, we significantly expanded the range of product selections, continuously lowered product prices (independent research firm Profitero has named Amazon the platform with the lowest online retail prices in the U.S. for the eighth consecutive year), and for the second consecutive year, we have shipped to Prime members at a record pace AWS has launched a large number of new infrastructure and artificial intelligence services, making it easier to create an excellent customer experience, including our latest custom AI chip Trainium2, the Amazon Nova series of cutting-edge foundational models, and our flagship generative AI ("GenAI") services—Amazon SageMaker and Amazon Bedrock—with significant expansions in the number and capabilities of models. Prime Video continues to roll out exciting original series, including new seasons of "Fallout," "Jack Reacher," "The Boys," and "The Lord of the Rings: The Rings of Power," as well as movies like "The Boys in the Band," "I Want You Back," and "Red One," along with sports events such as "Thursday Night Football" and the UEFA Champions League (NBA and NASCAR will be available in 2025). Apple TV+ has also officially joined the Prime Video Channels, greatly enriching our content offerings. We have also launched a series of new Kindle devices, including color versions, larger Scribe options, and the fastest Paperwhite series to date (this series has achieved the highest sales for Kindle in over a decade within a single quarter). Additionally, we are continuously expanding product selection in Amazon Pharmacy, improving price transparency, and expanding same-day delivery services. These achievements are part of the projects launched by the Amazon team in 2024, but they reflect our significant innovations, hard work, and thoughtful execution across various business areas. I am deeply grateful for the team's efforts and results over the past year (some of which are reflected in our 2024 performance, while others may take a few more years to materialize). A Culture of "Why" Every year in my annual shareholder letter, I try to share some core principles that keep Amazon running. At the highest level, our goal is to be "the most customer-centric company on Earth," striving every day to make customers' lives better and easier. This is not easy in itself, let alone maintaining it year after year. In fact, it is very difficult, especially in the context of rapid technological advancements, changing customer habits, and the continuous introduction of new products by companies of all sizes. If we want to have a chance to achieve this mission, we must constantly question everything around us. Amazon has always adhered to a long-term philosophy: distinguishing between "two-way doors" and "one-way doors" in decision-making. A "two-way door" decision means that if the decision is wrong, you can still backtrack to the original state with almost no serious consequences. Such decisions can be made quickly and locally. In contrast, a "one-way door" decision means that once you go through, it is difficult (or even impossible) to go back, so these decisions need to be made more cautiously. However, the premise for both types of decisions is that the door is open. Many innovations are actually attempts to open those doors that seem to have been tightly locked for a long time Over the past 30 years, we have found that one of the most important keys to unlocking these doors is actually a simple question: "Why?" "Why does this customer experience have to be this way?" "Why can't it be better?" "What are the constraints we face—why must we accept them?" "Why can't we innovate around them?" "Why does it take so long for customers to benefit from it?" Why? My dad often says that I was the kind of kid who always loved to ask "why" when I was young—sometimes perhaps to the point of being annoying. He often reminded me that shortly after I joined Amazon in 1997, he tried to persuade me to work for a more traditional company (preferably on the East Coast, closer to home)—until he realized that I had found the place that suited me best. That's because Amazon itself is a "why" company. We constantly ask questions like "Why?" and "Why not?" This helps us break down problems, trace back to the root causes, understand obstacles, and open doors that once seemed impossible to breach. Amazon has an exceptionally high "Why Quotient" (let's call it YQ), and this YQ also forms the framework through which we view everything. Since 1995, we have been asking: "Why can't we provide customers with all printed books?" Then we asked: "Why limit it to printed books? Why not also offer all out-of-print books?" "Then why not go further and put all books—regardless of the language they are written in—on a lightweight, handheld device that can be accessed in 60 seconds?" (This is Kindle.) When we provide reviews, we ask: "Why can reviews only come from so-called professional 'experts'? Customers themselves are a great resource and will be very honest. Why not include customer reviews, even if they sometimes discourage purchases?" "Why only sell books?" "What about music, video, electronics, tools, kitchen supplies, clothing, home goods?" "Why not sell almost everything?" "Why can only we sell these products? There are millions of third-party merchants and small sellers who also have similar or unique products. Why not let customers choose the products, prices, and delivery speeds they want from these sellers?" After spending years working to enhance customer awareness of these sellers' products, we asked ourselves: "Why not display these third-party products directly on the product detail pages, alongside our own products? (After all, that's where all the traffic is concentrated.)" So why not allow our sellers to store their products in our fulfillment network? This way, these products can enjoy fast Prime delivery services and be shipped by us (this is the "Fulfillment by Amazon" program) Why not try displaying relevant ads in our stores to expose customers to new sellers and products (rather than just content recommended by algorithms based on their past purchase history)? Why does every company need to build its own capital-intensive data centers and infrastructure? Why does every development team have to repeatedly create services such as computing, storage, databases, and analytics? Why do builders spend 80% of their time on these "non-differentiated heavy lifting" tasks instead of focusing on creating unique customer experiences? Why don't we build a set of services (AWS) to help builders, both internal and external, solve these problems? Why do I have to buy physical films to watch movies? Why do I have to use cable TV or traditional channels to see great TV shows (Prime Video)? Why can my Prime delivery benefits only be used on the Amazon website? Why can't they be used on other websites as well (Buy with Prime)? I could go on and on, but you get the point. Each of these "whys" has sparked significant innovation, each has made customers' lives better and easier. Some may seem obvious now, but at the time, these were challenging questions that required curiosity, the courage to take risks, a spirit of continuous experimentation, and relentless effort to turn them into success stories. How to Build a Culture of "Why" If you believe that a high "Why Quotient" (YQ) is crucial for creating innovation for customers, how can you inspire and cultivate it? In my view, this cannot be solved by a single approach. It must be deeply embedded in your culture and leadership team, and once you are fortunate enough to achieve success, you must steadfastly protect this culture over the long term. Here are some strategies we employ. Establish Leadership Principles that Set the Tone We have established 16 leadership principles to guide our behavior. They are all core elements supporting our "Why Culture" (YQ), but I want to particularly emphasize three of them: Are Right a Lot "Leaders are often right. They have excellent judgment and keen intuition. They actively seek diverse perspectives and work hard to disprove their assumptions." When we first introduced this leadership principle, some people mistakenly thought it meant "the best leaders are those whose ideas are always adopted" (i.e., "they are often right"). This led to some people stubbornly holding onto their views and unwilling to compromise. There is nothing wrong with holding beliefs, but in my experience, the best leaders are willing to listen to others' opinions. They do not shy away or get angry when challenged; instead, they become intrigued. When new, compelling information emerges, effective leaders change their views (which is why it is somewhat ironic that politicians are mocked as "flip-floppers" when they change positions) Ultimately, a leader's responsibility is to find the optimal solution for customers, rather than who proposed the original idea. Learn and Be Curious "Leaders never stop learning and are always seeking self-improvement. They remain curious about new possibilities and actively explore." In my 28 years at Amazon, the biggest determining factor in the difference between corporate and personal growth is the ability to learn. At some point, some leaders seem to lose their desire to learn. The reasons vary; some may feel that learning is too exhausting or time-consuming, or they may feel uneasy or threatened when they no longer have all the answers. However, if one day Amazon stops learning, that will be the day we begin to weaken our future potential. Those with high YQ are always curious—they want to know how they can do better, become wiser, and translate new knowledge into better customer experiences. Stand Your Ground; Express and Commit to Execute When in Disagreement "Leaders have the obligation to respectfully and clearly challenge a decision when they disagree, even if doing so is uncomfortable or exhausting. Leaders have conviction and resilience. They do not compromise for the sake of superficial harmony. Once a decision is made, they commit fully to its execution." We not only empower employees to challenge each other but also require them to do so when there is disagreement. If you disagree, you should ask questions, raise sharp issues, and drive deeper discussions—rather than silently assume the team is making a mistake. This approach is what brings better solutions for customers. At Amazon, the phrase "I told you so" holds no value. At the same time, the second part of this principle is equally important: "Express disagreement, and once a decision is made, commit fully." While constructive debate is very valuable, the team ultimately needs to make a decision and take action at some point. From that moment on, everyone—even those who advocated for other options—must commit to making the chosen solution successful. In other words, the team must be fully engaged, with no room for passive resistance, and should not waver between options. This is the only way we maintain decision-making efficiency and team confidence, especially after thorough debate, ensuring everyone can move forward in unison. Establish Behavioral Norms that Support a "Why Culture" Just as leadership principles guide our behavior, over the years we have also established some "team norms" that guide the way we work. Here are a few examples: Narratives As early as 2004, we stopped using PowerPoint for internal information reporting. We found that while PPT is easy for presenters to prepare, the content is often too high-level and abstract, making it difficult for the audience to understand the core issues. Therefore, we switched to using narrative documents with a maximum of six pages of text for communication Writing a narrative is more challenging for presenters (writing a concise, clear, in-depth, and focused six-page document is no easy task), but it makes it easier for the audience to grasp the key points, think deeply, and ask the right "why" questions. Working Backwards Documents When we develop a service or feature, we first write a "press release" and a "Frequently Asked Questions" (FAQ) document before writing the first line of code. The purpose of the press release is to ensure that what we intend to develop is genuinely valuable and exciting for customers (rather than waiting until the release to ask, "Wait, why did we think customers would be interested in this?"). The FAQ is meant to force us to think about some tough questions, such as: Who will use this feature? What do they like most? What are they most likely to be disappointed about? Why do we draw the boundaries of the feature here? How is it better than existing alternatives? How should we think about pricing? What pricing dimensions do we recommend? Why did we make these architectural design choices? The press release and FAQ are part of our "customer-first, working backwards" approach, which also prompts us to raise those questions that customers would definitely ask in a meeting. Be Together Whenever Possible There are many paths to breakthrough innovation. Sometimes, a genius does come up with a brilliant idea alone, and others just need to execute it. But this is not the norm at Amazon. Innovation at Amazon is highly collaborative; it starts with a spark of an idea, and then a group of smart, mission-driven people work together to refine, challenge, and expand it. We find that this process is far more effective in "face-to-face" settings than in remote collaboration. Of course, innovation can happen remotely (some company cultures even prefer it that way), but in my experience, it is nowhere near as effective as when everyone is in the same room: that sense of energy, rhythm, spontaneous brainstorming, active participation, ideas evolving on the spot, and the efficiency of quick iterations afterward all lead to better outcomes for customers and the team. Given the rapid development of AI today, and the fact that almost every customer experience could be redefined, I believe now is the time we need to optimize for "efficient innovation" more than ever. Tolerating Messy Meetings Innovation cannot be "scheduled." You cannot book a 60-minute slot to invent Amazon Prime, AWS, Alexa+, Amazon Logistics (FBA), regionalization of the fulfillment network, or the Kuiper project. These innovations often stem from someone asking the question, "Why can't we redefine possibilities for customers?" They then take on a life of their own, often going through many twists and turns, and even dead ends, before finally finding a successful path This may be troubling for those who prefer process standardization. But when we are in an innovative state, we are willing to accept the imperfections of this process—and appreciate its "beauty in chaos." Operate like a startup (in our case, the largest startup in the world) We always strive to operate like the "world's largest startup." So, what does that mean? First, any product or service we consider developing must focus on solving a real customer problem or significantly enhancing the customer experience. Some companies stray off course due to excitement about a certain technology, prioritizing "the technology itself." In contrast, excellent startups always have the mission of "redefining possibilities for customers." Second, our need for "builders" far exceeds that of conventional enterprises. These individuals are innovators—they constantly dissect the customer experience, and even when the current experience is quite good, they still ask, "Why can't it be better?" They are "innately dissatisfied" with the status quo (which may annoy team members who take pride in existing achievements), and they never feel that things are complete. Third, we need "owners." One of Amazon's key advantages over the past thirty years has been our ability to attract a large number of smart, motivated, creative, and ambitious talents who possess a strong sense of "ownership." This means our team members constantly ask themselves, "If this money were my own, what would I do?" "If this were my company and I were the largest shareholder, how would I make decisions?" "Although I am only assigned to a part of the project, I am not sure if the other parts are doing well—should I step in to check and ensure everything is okay, or should I trust that others will handle it?" True "owners" hold themselves accountable. They care deeply about the quality and effectiveness of what they are responsible for, viewing the company's mission as their own (we want "missionaries," not "mercenaries"). This is also one of the reasons we strive to increase the ratio of "individual contributors" to "managers." We want a flatter organizational structure that allows those who are truly hands-on "owners" to lead "two-way door" decisions (which are the vast majority), act quickly, and take full responsibility for every "why" in enhancing the customer experience. Fourth, speed is disproportionately important for all businesses in any industry and at any time. The notion of "you can either be fast or have high standards" is a false dichotomy. If you want to be fast, you can absolutely do so while ensuring high quality. We have been doing this for many years (although we still have room for improvement). Speed is a leadership decision. The leadership team must genuinely believe in its importance, continuously reinforce this culture, optimize the organizational structure, remove structural barriers that hinder processes, and build systems in a modular way to support rapid advancement But if the entire company and culture do not collectively embrace "speed," then speed cannot be achieved. At Amazon, regardless of the business area, we always have a sense of urgency—every window of opportunity around us is closing rapidly. We are in an extremely competitive market environment, surrounded by talented, well-funded, and ambitious companies. Customers are always looking for better options. We spend a lot of time thinking: How can we unlock these experiences for customers faster? We are very clear: If we don't do it, someone else will. Another way to enhance speed is to eliminate bureaucracy. There is a difference between processes and bureaucracy. When you are operating a business at scale, you do need some mechanisms to ensure the correctness and continuous optimization of the customer experience. However, as the company grows and management layers increase, some unnecessary processes also accumulate, which have little actual value. Last fall, I invited colleagues from various teams in the company to send me examples of bureaucracy they encountered. I received nearly 1,000 emails, each of which I read carefully. Builders hate bureaucracy—it slows down the pace, is frustrating, and hinders them from doing what they truly love. As leaders, we may not always see the red tape hidden deep within the organization, but once we do, we must decisively eliminate it. So far, we have made over 375 changes based on this feedback. We must accelerate our actions and commit to continuously rooting out the bureaucracy that wastes time and undermines morale. Fifth, maintain a "grassroots spirit." As companies succeed and scale, many forget how they started. We initially developed Amazon S3 (Simple Storage Service) with 13 people and Amazon EC2 (Elastic Compute Cloud) with 11 people. Some managers mistakenly believe that growing the team is the only way to grow the business. We have historically had a period dominated by this mindset. But this is not the fundamental way Amazon builds teams and products; we have readjusted and returned to the essence. Our best leaders are often those who accomplish the most work with the least resources. They take pride in being "efficient and lean." Sixth, you must be willing to take risks. It's easier said than done. You need smart people to identify opportunities worth betting on. And when you have these high-standard, creative, and ambitious builders, they often are not accustomed to failure. They worry that if the innovations they attempt fail, they will be questioned and ridiculed by the outside world (and even internally). Therefore, many choose to "play it safe." But if you are only playing not to lose, you cannot create truly extraordinary value for customers. If the "why" you propose leads you down an innovative path that no one has ever taken, let "customer obsession" be your guide. After all, it is difficult to change the world by merely following in others' footsteps Ultimately, what you must care about the most is delivering attractive results for your clients. It's not about how charming you are, nor is it about how good you are at managing upwards or laterally. What truly matters is what we have genuinely accomplished for our clients. This is what we hope to reward. The Next Generation of "Why" While my team and I are very optimistic about the progress and potential of our existing business, we still have many new "whys" that we are contemplating. Here are some of the questions, along with our initial thoughts that we are exploring. Why is artificial intelligence so important? Will it really have the huge impact that some people claim? When will it be realized? Generative AI will redefine almost every customer experience we know and bring about entirely new experiences—experiences that previously only existed in imagination. The currently deployed AI applications mainly focus on enhancing efficiency and reducing costs, such as customer service, business process automation, workflows, translation, etc., which have already saved businesses a significant amount of money. In the future, AI will gradually change conventional patterns in various aspects such as programming, search, shopping, personal assistants, basic healthcare, cancer and drug research, biology, robotics, aerospace, financial services, neighborhood networks, and more. Some areas have already begun to evolve rapidly, while others are still in their infancy. But one thing is certain: If your customer experience has not yet planned to leverage these intelligent models—capable of querying vast amounts of data, quickly identifying key points, and continuously becoming smarter through feedback and data, even developing the ability to act autonomously—then your products and services will not be competitive. How long will it take to realize? It won't all happen in a year or two, but it won't take as long as ten years either. The pace of AI development is nearly surpassing any technological advancement in history. Okay, I believe AI is important; but why invest so quickly and on such a large scale? Fundamentally, if your mission is to "make customers' lives better and easier every day," and you believe that AI will redefine every customer experience, then you must make deep and broad investments in AI. This is precisely why Amazon is currently building over 1,000 generative AI applications across various fields, including shopping, programming, personal assistants, video and music streaming, advertising, healthcare, reading, and home devices, all aimed at fundamentally improving customer experience. This is also why AWS is rapidly developing key "primitives" (basic building blocks) that support AI, including: Amazon Trainium: Custom AI chips that provide better cost-effectiveness for training and inference; Amazon SageMaker and Amazon Bedrock: Platform services for flexibly building and inferring AI models; Amazon Nova: Our self-developed cutting-edge models that help customers reduce costs and latency; Intelligent Agent creation and management capabilities: driving AI systems to automatically complete complex tasks. This wave of AI also requires significant capital investment. At AWS, as demand grows rapidly, we need to procure more data centers, chips, and hardware (and the cost of AI chips is much higher than traditional CPU chips). These capital expenditures need to be made in advance, even though these assets will continue to function for many years (for example, data centers have a lifecycle of at least 15-20 years). We only begin to monetize our capital investments "months or even longer" after the investment, and it takes years to recover, which brings highly attractive long-term free cash flow and return on investment—as demonstrated by AWS in recent years. However, during this exceptionally strong demand period, we are investing a significant amount of capital. Our AI revenue is currently expanding at a triple-digit year-over-year growth rate, with annualized revenue reaching billions of dollars. We firmly believe that AI is a historic opportunity that will change everything we know. The customer demand it brings is unprecedented, and our accelerated investment now is for the long-term benefit of customers, shareholders, and the business. Why are chips and AI so expensive for customers? AI doesn't necessarily have to be as expensive as it is now, and it won't always be this way in the future. The biggest culprit for the current high costs is chips. So far, most AI has been built on the foundation of a single chip manufacturer, and its prices are not cheap. The Trainium chips we launched are expected to bring change—especially our Trainium2, which offers a 30-40% cost-performance improvement compared to the currently widely used GPU computing instances. Although model training still accounts for a significant portion of current AI costs, in the future, inference—i.e., model predictions and outputs—will become the absolute mainstay of AI costs. This is because customer model training is periodic, while inference occurs continuously, especially in large-scale AI applications. Inference will become another foundational service module, just like computing, storage, and databases. We feel a strong sense of urgency about this—we must reduce inference costs for customers as soon as possible. More cost-effective chips will help. At the same time, model distillation, prompt caching, computing infrastructure optimization, and advancements in model architecture will significantly improve inference efficiency in the coming years. Reducing the unit cost of AI will unleash greater demand for AI applications from customers and drive overall growth in AI spending. This trend is similar to AWS's development path: we achieved lower unit costs by revolutionarily reducing computing and storage costs, which in turn drove more innovation, better customer experiences, and higher overall infrastructure investment. Why have personal assistants not really become mainstream yet? How does Alexa work? An excellent personal assistant should be able to answer almost any question and also represent you in completing various tasks. So far, no digital solution has been able to do both simultaneously—until the emergence of Alexa+. Alexa+ is not only comparable to leading chatbots in terms of intelligence, but more importantly, she can perform a large number of real operational tasks for you: Play music and videos Switch media content between devices Set alarms and timers Control smart home devices Place orders among hundreds of millions of e-commerce products Make restaurant reservations and call ride services Order concert tickets Remind you when your favorite artists announce tours Find a plumber to fix your sink Remember everything you have done on Amazon This is a revolutionary change for consumers, and this is just the beginning for Alexa+. Currently, there are over 600 million Alexa devices worldwide, and we expect that in the future, Alexa+ will play an even more central role in the lives of these hundreds of millions of customers. Why can't we deliver products faster? Does it really matter? Every year, people ask us whether delivery speed has reached the limit of "diminishing marginal returns." Our data shows: not at all. When we promise faster delivery times, the proportion of customers completing their orders significantly increases, and the shopping frequency on Amazon also rises. Amazon Prime's initial promise was: unlimited, free, two-day delivery on a million products; now, it has expanded to over 300 million products, with tens of millions available for delivery within one day or even shorter. More and more orders can now achieve same-day delivery. This speed increase is mainly due to our regional restructuring of the fulfillment network, new inventory allocation algorithms, and the introduction of innovative Same-Day Fulfillment Centers. Although we have set delivery speed records for two consecutive years, these innovations are still being optimized, and more plans are underway. Don't forget our Prime Air drone project, which will deliver products to customers within an hour. We are far from done when it comes to delivery speed. Why can't small-town users enjoy the same fast delivery speed as urban users? While some companies abandon small-town customers due to high service costs, Amazon chooses to do the opposite—we are increasing our investment to better serve customers in remote areas. We have expanded same-day and next-day delivery to dozens of small cities and towns across the United States and are continuing to expand. This expansion will bring faster Amazon delivery services to millions of customers, especially in areas with lower population density In the future, we will be able to deliver over 1 billion packages annually to customers living in 13,000 zip codes, covering an area of 1.2 million square miles. Relatedly, why can't we help the hundreds of millions who lack broadband connectivity? There are approximately 400 to 500 million households globally, mostly located in remote, small towns, that cannot access broadband networks. They cannot use the internet as conveniently as people in big cities to learn, shop, do business, access entertainment content, or communicate with others. This is precisely the problem we aim to solve with Project Kuiper—a low Earth orbit satellite network. We have just launched the first batch of mass-produced satellites, and in the coming years, we will have more than 3,200 satellites in orbit. Although the launch costs are high, we believe that Kuiper will ultimately become a business with substantial operating profits and return on investment. Why must healthcare services be so anxiety-inducing? In the United States, the healthcare experience is frustrating—especially primary care. It is often very difficult to schedule an appointment with a doctor, and getting an appointment with a specialist is even harder. A lot of time is spent "waiting," while the actual time doctors spend communicating with patients is minimal. Afterward, patients often have to drive to another location (usually not close) to pick up their medications. And when they arrive at the pharmacy, they are often troubled by medication prices, insurance coverage, and certain "medications locked in cabinets." Customers deserve a better experience. This is why Amazon Pharmacy and Amazon One Medical have received positive customer feedback and rapid growth. We will continue to accelerate the optimization of pharmacy product selection and price transparency while expanding One Medical's physical clinic service capabilities. These are some of the "whys" we are currently asking ourselves. I am excited about future innovations—we are destined not to be bored. When I first started working, I could hardly imagine that my father worked for the same company for 45 years. I thought, "How is that possible? That's too long." I often told my friends, "I would never do that." And now, after nearly 28 years at Amazon, I have to face my friends' reverse "whys." After all these years, why are you still at Amazon? I am certainly a "super fan," but there are indeed many compelling reasons to work at Amazon. First, I am not sure if there is another company that is as obsessively customer-centric as we are. Many companies say this, but very few actually do it. Second, in this world, it is indeed not easy to find a company that allows you to make a greater impact Third, we will make long-term and in-depth investments in innovation and talent. This allows our team to continuously refine ideas and make truly long-term decisions for our customers and the company. Moreover, I have never seen a group of people like our Amazon team—they are smart, creative, ambitious, eager to break through, hardworking, and mission-driven. I believe this is the most remarkable set of qualities a company can possess. For those builders who want to change the world and are passionate, there is no better place than Amazon. The reason we can operate like the "world's largest startup" is largely due to our "Why culture." We do not always get it perfectly right, but we learn and iterate like crazy, continuously choosing to put customers first, pursue excellence in delivery, embrace an innovative spirit, take ownership, act with speed, embody grassroots spirit, and maintain curiosity, all committed to building a company that transcends each of us. For us, it is still Day One. Sincerely, Andy Jassy President and CEO Amazon.com, Inc. Note: As always, we have attached the original 1997 shareholder letter at the end of this letter. The content remains as true today as it was then