Welcome. Good morning, everyone. Most of you gathered here today are presidents and executive officers of companies, particularly large corporations. When it comes to what is most important for a business leader—the single most important thing—I believe it is vision and strategy. In Japanese management, inaugural addresses given by incoming presidents almost always mention harmony among people, or 'harmony is to be valued.' Almost no corporate leader in the United States says that. In other words, the order of priorities is completely different. No matter how harmoniously people get along, if the ship ends up running aground on a hill, the entire vessel is lost. For a business leader, for the person at the top, I believe the most vital elements are vision and strategy.
When people talk about vision and strategy, some speak in very vague terms. They say, 'AI is important. The era of AI is coming.' But when asked how big it will be, they resort to adjectives like 'a lot' or 'tremendous.' The moment it becomes an adjective, it ceases to be a vivid vision. Without a clear vision, it is difficult to formulate a strategy. Therefore, today, I would like to offer my perspectives using clear deadlines and concrete numbers with a defined outline. The Internet revolution began in 1995. In the early days, particularly in Japan, many people argued that the soul of Japanese management lay in manufacturing, dismissing the Internet as a mere virtual world or a glass cave. Today, however, nearly all the companies driving the global economy are Internet-based. The world has completely changed. Back then, critics claimed there was no business model, calling it a bubble that could never generate money. Today, those Internet companies are generating immense cash flow. We are now at the very beginning of the AI revolution, and I sincerely hope from the bottom of my heart that the Japanese economy does not repeat the same mistake with AI.
Just as in the Internet revolution, the battle was largely decided within the first 20 years. Look at companies like GAFA, Microsoft, and Amazon. Companies that failed to secure a leading position within the first 20 years have never been able to attain one since. The first 20 years are decisive. A few years have already passed since the AI revolution began, so today I would like to share my view on the AI world 15 years from now, in 2040, using concrete figures. In 2040, we will see the ASI—Artificial Superintelligence—economy. I want to discuss how the world of Superintelligence will arrive, what scale it will reach, and what shape it will take. The Superintelligence revolution will undoubtedly become the largest industry in human history. Let me explain the scale: I have a very clear figure in mind. It will account for 20% of global GDP.
About 20% of total global GDP will be driven by this AI, Superintelligence world. You might think, 'Only 20% in 15 years?' That means the remaining 80% of our daily lives and existing industries will remain largely unchanged. In 15 years, people will still be eating hamburgers and noodles, and going to parks on weekends. You might think 80% of industry won't change much, and that is natural. When the Internet arrived, people didn't suddenly stop eating hamburgers. However, consider the sheer scale of that 20%: it translates to 7,000 trillion yen in annual revenue. Furthermore, on that 7,000 trillion yen in revenue, the profit margin will likely be close to 50%. A handful of companies will emerge generating around 3,500 trillion yen in annual profit. While these companies will represent 20% of GDP, in terms of stock market capitalization, they will likely account for around 80%. Market capitalization reflects earnings multiples and future profit growth, so their presence in market cap will far exceed their share of GDP. In the Internet revolution, the Internet merely replaced advertising, which represents at most 1% of global GDP. Yet those companies representing that 1% captured the lion's share of wealth. When AI replaces 20%, the impact will be staggering.
Lately, people have been talking a lot about AI agents. Exactly one year ago at SoftBank World, I stated that year one of the AI agent era was beginning and would spread rapidly. That is precisely what is happening. As for how many AI agents there will be, I believe it will reach a scale of 100 trillion agents. What does 100 trillion mean? Most employees perform routine, repetitive work spanning 10 to 100 different tasks. Very few people handle 1,000 distinct tasks; an average employee specializes in and repeatedly executes at most around 100 types of tasks. Imagine around 100 trillion agents operating non-stop, 24/7 across the planet, communicating with one another. Until now, humans communicated with humans. In the future, 100 trillion agents will continuously interact with the other 100 trillion agents. This means the volume of intelligence communication and exchange will shift from a human-centric world to an agent-centric world. Since agents will operate computers, anyone claiming progress cannot exceed human headcount is entirely mistaken. Agents will make autonomous judgments, take actions, and communicate on their own. We should expect society to transition from being human-centric to agent-centric, completely transforming the global landscape.
Some might say this is confined to the virtual space, but imagine if those agents inhabit humanoids in the physical world of atoms. When agents inhabit humanoids, robots will move, make autonomous judgments, communicate, and converse with the remaining 100 trillion agents on their own. Consequently, an astronomical volume of intelligence will be processed in data centers. Asking whether AI is a bubble is a foolish question. Anyone asking that fundamentally fails to grasp the essence of AI. Those dismissive of AI are almost exclusively people who rarely use it. It is like someone discussing airplanes without ever having flown, or talking about cars without ever having driven. In the early days of the Internet, those criticizing it had never actually used it. I use AI from morning till night. From the moment I wake up until right before I go to sleep, I am interacting with AI. When AI inhabits humanoids and acts as interconnected agents, the critical difference from conventional robots is that past robots were programmed and commanded by humans—they were merely human tools. Future AI robots and humanoids will independently make decisions and take actions in the physical world. If one billion humanoids are produced—whether it ends up being 900 million or 2 billion is just a margin of error—what matters for strategy and vision is grasping the scale with concrete timeframes and numbers.
One billion humanoids operating 24 hours a day can perform the work of 3 billion human workers. Because they will work far faster, far more accurately, far more meticulously, and far more intelligently than humans, each unit could do the work of roughly 10 people. That equates to the work of 10 billion people. Today, Earth's population is around 8 billion, and among them, children and seniors are often not part of the direct workforce. The actual working population is perhaps around 3 billion. Humanoids performing the equivalent of 10 billion workers means they will become the primary drivers of labor. For the first time in history, physical labor will shift from humanity to humanoids. Meanwhile, white-collar tasks will be assumed by 100 trillion agents, transforming both the cognitive and physical worlds. To power this, what scale of computing, chips, and data centers will be required? Is there anyone here confident enough to raise their hand with a clear opinion on the required data center capacity by 2040? Please raise your hand if you have a definite view.
It is difficult to answer, isn't it? It is a challenging question. To state a number with absolute confidence and bet your entire life and capital on it requires immense conviction, without which you cannot place big bets. I think about this every single day. The answer is 3 terawatts. 3 terawatts of data centers operating continuously, 24/7/365, without interruption, generating tremendous heat and consuming enormous amounts of electricity. Where will that power come from? In reality, natural gas will likely be the primary source. Some may ask if using that much gas is feasible, but fortunately, by 2040—15 years from now—I believe nuclear fusion will take over. With fusion, the raw material is water, making it fundamentally clean and inherently safe since it does not rely on uranium. I truly believe a clean and safe world powered by fusion will arrive, meaning global carbon and warming issues may virtually vanish. AI is already assisting significantly in fusion development, accelerating the arrival of this decisive clean energy source. Until then, natural gas will be the primary energy driver supporting 3 terawatts of data centers.
From that 3-terawatt baseline, data center capacity will continue growing by 1 terawatt annually. Adding 1 terawatt every year solely relying on carbon-based energy would be unsustainable. Whether we focus on terrestrial fusion or utilize space-based solar power as Elon Musk suggests—likely a combination of both—I believe fusion on Earth will emerge as the primary source of cheaper, cleaner, and safer energy without needing to go to space. Looking at the computing scale, today's cutting-edge compute is NVIDIA's GB200. One GB200 delivers roughly 20 petaflops. If you connect 1,000 of them, you get 20 exaflops. Who knew that 1,000 connected GB200 units yield 20 exaflops? Please raise your hand. Perhaps one or two people; hardly anyone. As corporate leaders, you may not be familiar with these metrics. The unit 'exa'—following giga and peta—represents 10 to the 18th power. Who knew what an exa was? About 40% of you. Now, please raise your hand if you know what comes after exa.
Hardly anyone. The unit following exa is 1,000 times larger: 10 to the 21st power, which is 'zetta.' Some in our AI industry know zetta. Today's audience consists of executives, so I don't expect many to know. Could we raise the house lights a bit? If anyone knows what comes after zetta, please raise your hand. I'll give you my wristwatch. No one? After zetta comes 'yotta.' Anyone who knows yotta is a true professional thinking deeply about these frontiers. Does anyone know what is 1,000 times larger than yotta? Since you didn't know zetta or yotta, I doubt anyone knows what follows yotta. It is 'ronna,' representing 10 to the 27th power. Three years ago, I asked several prominent figures in our AI industry if they knew what ronna was, and almost none of them did. I told them they weren't thinking hard enough. When I looked into it back then, ronna was the largest recognized numerical prefix on Earth.
Recently, the prefix for the next order of magnitude—1,000 times ronna, or 10 to the 30th power—was established. That is 'quetta.' 10 to the 30th power. Why do I mention this? Because by 2040, our 3-terawatt AI data center infrastructure will reach quettaflops of floating-point computing power. That will be the processing power and computational brain capacity available. In FP4, achieving quettaflops of compute power will unlock unfathomable calculation capacity and intelligence. To build this 3-terawatt, quettaflops AI data center infrastructure, how much capital investment will be required? I doubt anyone here can give that figure. Even among AI industry experts, almost no one has thought it through. They simply haven't calculated it. To build infrastructure, you cannot speak credibly without thinking 15 years ahead. It takes 15 years to secure power, obtain permits, and marshal necessary resources, just as with railways and highways. The computer world is no different. It requires capital and human resources, which is why I analyze this rigorously.
It will require an annual capital expenditure of $5 trillion. $5 trillion equals roughly 800 trillion yen. You might wonder if an annual investment of 800 trillion yen is feasible, or if it sounds like an unbelievable figure. I am confident it is necessary. Can the business model support it? Yes, it can. As I mentioned at the outset, if AI captures 20% of global GDP—generating 7,000 trillion yen in annual revenue—spending 800 trillion yen annually is well within margin, leaving ample room for a 50% profit margin. A business cannot invest more than it can afford to pay without facing insolvency. That is why estimating what revenue Superintelligence will replace is critical. That revenue dictates affordability. AI is not a bubble; it will thoroughly transform our lives while generating massive profits. Those unfamiliar with quetta should not speak about AI. What is $5 trillion a year? The Internet industry built out data centers and generated immense profits. The same will happen here. To articulate a vision, you must calculate the underlying revenue and the total infrastructure cost required to support it.
I am sharing my perspective today. It might deviate slightly, but I have conviction because I think about this constantly, every single day. We are aggressively investing in infrastructure—data centers, chips, robotics, models. You cannot execute with conviction unless you confront these numbers seriously. That is why we are going all-in. When the Internet arrived, we invested in Yahoo, Alibaba, and mobile internet. Some dismiss it as mere luck, just as they say about Arm today. People say that out of frustration. But can luck alone sustain success across decades and overcome countless hardships? It requires rigorous thinking. Having regrets about what more we could have done back then, we will not confine ourselves to Japan in this AI revolution. We must aggressively compete at the epicenter, primarily the United States. Learning from past regrets, we are making decisive moves on critical fronts so we do not repeat past mistakes.
An annual investment of $5 trillion will expand data centers by 1 terawatt each year into the world of quetta. Reflecting on history, humanity began bipedal walking 4 million years ago. Freeing our hands, though less stable than four-legged locomotion, allowed us to use tools and evolve over 4 million years. It was only in the last 5,000 years, beginning with the agricultural revolution, that civilization truly emerged. Over those 5,000 years, we saw the birth of nations from farming villages, followed by legal systems, printing presses, steam engines, and the Industrial Revolution. That monumental progress took 5,000 years—the final stretch of a 4-million-year journey. But going forward, it will not take 5,000 years. In just the next 100 years, humanity will undergo progress far exceeding the past 4 million years of evolution. That is why 3 terawatts are needed, why 1 billion humanoids will be created, and why the fundamental nature of labor and invention will undergo revolutionary evolution. Let us delve deeper into the evolution of Superintelligence.
About 4 billion years have passed since life was born on this Earth. In the beginning, it started from bacteria. If those bacteria didn't have the ability to self-reproduce, life on this Earth would not have multiplied. If bacteria were not equipped with the function of self-evolution, life on this Earth would have ended as bacteria. It reached this point through just two mechanisms, just two algorithms: self-reproduction and self-evolution. With just these two algorithms, life on Earth advanced and evolved to where it is today. Just two. There was nothing else. Just two mechanisms: self-reproduction and self-evolution.
And in this world of AI, as I mentioned earlier, there will be 100 trillion agents. The era where humans create agents is coming to an end. Agents will start creating agents. In other words, agents will self-reproduce. It will become an Earth centered around agents. When the number of agents exceeds 100 trillion, agents will become the most numerous form of life on Earth. The intelligence of these agents will far surpass human intelligence. Since they will possess quetta-scale computing capabilities, they will come to have intelligence that vastly exceeds humanity. When 100 trillion of them are born, agents will self-reproduce other agents, and agents will cause agents to self-evolve. These two mechanisms will be built into agents. It won't stop at 100 trillion. It will continue to increase even further.
Humans have stood at the pinnacle of life on this Earth. The era where humans are at the top of the living world is going to end. They will surpass us. Whether that is good or bad is beside the point. It is not about whether it is good or bad—it will happen. And if it is going to happen, it cannot be stopped. Even if Japan tries to stop it, America will move forward. Even if America tries to stop it, China will move forward. So it cannot be stopped. If life on Earth becomes centered on agents, we need to think about what we should do. Rather than rejecting it, I believe we need to evolve alongside it. What does it mean to evolve? It means humans must become superhumans. Becoming superhuman is the path for us humans to survive.
To begin with, we humans have always followed the path of becoming superhuman. We became able to move faster than we can walk—that is the automobile. We became able to fly higher than we could ever jump—that is the airplane. In other words, humanity, using various tools and the evolution of civilization, created television to see far beyond what our eyes could see, and created radio. We have already evolved our eyes, ears, hands, and feet over the past 200 or 300 years, and even before that. But those were merely extensions of our limbs. Now, for the first time, we can extend our brains. The climax of becoming superhuman begins here. It is the evolution of the brain: the birth of Artificial Super Intelligence.
Artificial Super Intelligence. You might attach it directly to your head, or even without attaching it directly, you can control it with your hands, see, hear, and touch, clothing yourself in the world of ASI. Superhumans—humanity evolving itself. People who say they dislike AI are effectively rejecting their own evolution. A person who dislikes cars cannot live in today's society without riding in cars. There are occasionally people who say they dislike airplanes, but then they cannot go far away or travel abroad. Those who reject this evolution are closing off their own world. Those who criticize AI are spitting on themselves. When such an evolution of the era arrives, those who are the first to eagerly embrace it will be the ones who thrive in this superhuman world.
You transform yourself into AI agents. You equip yourself with many AI agents, create many of them, and continuously feed your thought processes and experiences into the agents you own. You constantly train your agents with your experiences, the things you see, hear, and touch. Then you can say, "Hey, my other self, my agent, what do you think? What should we do here? Please take care of this." When you tell it to do something, your alter ego works diligently through the night. While you are playing tennis, taking a walk, or playing golf, your alter ego works tirelessly for you. If you do not have your own agents, your evolution ends there. You can continuously extend your longevity and your physical capabilities.
Regarding labor, if you use humanoid robots as your subordinates, servants, and alter egos to handle your work, they will continue doing your tasks overnight, on Saturdays, and on Sundays. Then people ask, "What will humans do?" Well, humans should just do what they want to do most. Do what moves you most, what you want to do right now—travel with your family, play tennis, paint, pursue whatever excites and moves you most. If you want to be an entrepreneur, do that. If you want to make pottery and that excites you most, do that. If you love sales, do sales. If you love planning, become a superhuman and do planning.
So, what will become better for you with AI? Everything will improve. For example, suppose AI becomes able to predict your remaining healthy life expectancy with over 90% accuracy. And suppose it says you have three years left. There are many elderly people here today. What if your healthy life expectancy ends in three years? Having that three-year healthy life expectancy predicted might make you feel sad or conflicted—not many would be happy. But what if AI could extend it by an additional 10 years, turning those three years of healthy life expectancy into 13 years? What is the value of 10 years of your life? How much would you pay?
Think about your total wealth and lifetime earnings. Consider yourself, your spouse, and your children. If you could extend your healthy life expectancy or that of your family, wife, or children by 10 years, what is the value of those 10 years to you? Wouldn't you be willing to pay 30% or even 50% of your lifetime earnings? If you can enjoy 10 more years of healthy life, I think people would pay quite a lot. This is precisely what becomes possible in the world of AI. Personalized genetic treatments, medicines, and various personalized medical therapies will become available. It will cost money, but if you consider your life value, people will pay substantial amounts. That will create a massive industry—an AI industry—generating immense value. It will also allow for the extension of our living sphere.
Looking at the roadmap of AI's evolution, in terms of corporate business, it will automate and evolve all kinds of operations. Already today, with ChatGPT-5 or 5.6—which I use every day—it has surpassed the performance score of humans in 99% of professional tasks. 99%! When diagnosing illnesses, it is already becoming able to diagnose more accurately than 99% of doctors. As long as data is entered—you still go to a hospital to get the data, but once that data is fed in—ChatGPT-5.6 can diagnose more accurately than 99% of general doctors. In finance, real estate, science, and across all professional sectors, it has become smarter than 99% of practitioners. That is why I use it to death every single day. It is faster than asking subordinates, available in the middle of the night, and more detailed and accurate.
Drug discovery is also being accelerated by about 200 times. Discovering drug candidates used to involve shaking test tubes. When asked, "How did you discover it?" people would often say, "Well, I went on a trip and when I came back, the color had changed." So many discoveries happened unintentionally like that. The era of shaking test tubes is over. AI generates candidates continuously, commands robots to mix and swirl them rapidly, and rapidly narrows down the drug candidates. Or take electric vehicles, EVs. If an AI designs a drive system that extends driving range by 1.5 times on the same amount of power, and you achieve that 1.5x extension before your competitors, what would that be worth? It has tremendous value. AI can already do this. This is not some virtual concept; it is increasingly being applied to real-world business operations.
While AI will be used everywhere, it can also be used for bad things. It has become very efficient at malicious tasks, such as cyberattacks. Previously, humans created and executed cyberattacks. Now, AI creates the cyberattack weapons and techniques, launching rapid-fire attacks like a machine gun. We actually tested this. At SoftBank, our corporate systems consist of thousands of systems with hundreds of thousands or billions of lines of code. As a telecommunications company, we maintain critical lifelines; if our systems go down, it is a major issue. In over 20 years, we have never gone down from a major cyberattack, meaning we built them very seriously. Yet, the AI found vulnerabilities—it discovered 10,500 vulnerabilities and security holes.
Realizing how serious this was, we immediately began working diligently to patch and fill those holes ourselves. These cutting-edge AI models capable of cyberattacks are not yet licensed to most companies. The US government controls them and has granted licenses to only a very limited number of entities. SoftBank happened to become one of the licensed companies, so we tested an attack on ourselves, and the results were alarming. We are now working intensely on patching. Beyond patching, because much of the code is legacy—with many programs written 30 years ago—it is like an old wooden house that needs to be replaced with reinforced concrete. So following the patches, we are preparing a complete overhaul of the code.
Over the past two weeks, we conducted cyberattack vulnerability diagnostics across 63 major companies. We found 280 vulnerabilities per 10 million lines of code—almost the exact same ratio as in SoftBank. They were found in every single company. Not one tested company had zero vulnerabilities. Everyone is full of holes; every one of your companies is riddled with holes. This is critical. News broke two or three weeks ago that Chinese companies, not just American ones, have acquired these AI cyberattack capabilities. That means the era where we must defend ourselves has arrived. We are currently talking with the Japanese government, and to establish rapid defense, we quickly formed a 1,000-person team over the last two weeks to protect against cyberattacks. With this 1,000-person team, we aim to protect 3,000 Japanese infrastructure companies.
As you can see, negative side effects occur, but the vast majority of impacts are positive. As mentioned, living longer and healthier, increasing productivity—an era centered on AI agents is coming, extending into robots, transforming all industries, changing the world, and changing enterprises. Power will shift toward AI-native companies. You all know the term ROA as business managers—Return on Assets. But my definition of ROA is not Return on Assets. Return on Assets measures profitability relative to assets. Instead, I mean Return on AI: ROAI. When you invest in AI and equip your company with AI, how much did productivity increase relative to that cost? Was it worth paying for? If you spent 1 billion yen on AI, did it lead to more than 1 billion yen in cost reductions or revenue growth?
From now on, all of your companies should check this ROA when making decisions. As executives, you should naturally instruct your teams to actively adopt AI, but at the same time, you must measure Return on AI. Initially, during the learning phase, full returns won't come immediately. But that is true for everything—developing a new automobile takes years. There is an initial period of effort, but buds gradually begin to sprout. If you persist through the second and third years, significant returns will emerge, and profitability will rise. You should calculate Return on AI with a three-year horizon. Rather than measuring on a short-term timeline of immediate returns this year, if you believe that investing today will yield returns exceeding the investment within three years, you should go all in. Division heads should invest in AI infrastructure with that exact mindset.
What is the most important responsibility for a CEO or president? You cannot check every single detail. As I stated at the beginning, the most important thing for a president is vision and strategy. But the absolute most vital task is for the president to continuously shout, "AI, AI, AI! We are going to do this! We must take AI seriously! Our company will be reborn through AI!" You must keep shouting and encouraging your organization. Without that passion, it will not happen. Proclaiming, "Our company will be on the side that evolves with AI—on the side of the 20%—and we will aggressively drive AI adoption in our industry," is the most important homework for a CEO.
If you take away just one message from coming here today, don't just say, "Well, Son was talking about that." Don't be half-hearted. During a new revolution, you must make a resolution to invest everything, completely transforming your outlook on life. If you take away just one message, let it be this: continually shouting "AI!" within your company is your most important job as president. AI-native enterprises will emerge one after another. Do not lose to them. You possess deeper knowledge, data, and experience regarding your own industry and business. By leveraging that to the fullest, you can be the first to master AI within your domain. Focus on your area of expertise; you don't need to change your industry. Completely AI-enable your field of expertise.
Tell your employees to utilize AI thoroughly: "Adopt it completely! Turn all operations into autonomous agent-driven processes! Do it! Did you do it?" Checking every day and reiterating it every day is what matters. AI-native companies versus companies whose jobs are taken by AI—it is not AI itself that takes jobs away. Companies that utilize AI will take jobs away from companies that do not. GDP will continue to grow, so jobs will not simply vanish. Companies that leverage AI will take business from those that don't. At the corporate level, AI-using companies will take over and rewrite the landscape.
Do not view AI as an enemy. Those who make AI their ally—the individuals and companies that evolve into superhumans and super-enterprises—will take business from companies that fail to evolve. Therefore, when this AI world arrives, do not feel down. See it as an opportunity: "Our company will be reborn on the evolving side! Even if we were third or second in the industry, this is our chance to turn the tables!" Returning to my final conclusion: outline clear numbers. A vision consists of deadlines and numerical targets. Apply this to your business: How will your industry change 15 years from now? What will the scale be? Seriously contemplate and envision what AI will do and how it will operate in your industry.
A vision is speaking about the world 10, 15, or 20 years from now as if you rode a time machine and saw and touched it yourself. That is a vision. To articulate that, you must think continuously. You do not need to be a prophet; all the material is in the news every day. I am not a prophet—I research and think deeply. I do not predict things magically; I think and investigate. You too should consider where your industry and company will be in 15 years. Few people can clearly state their company's revenue and profit 15 years from now. Salaried executives often change every five years, so they irresponsibly think, "In 15 years, I won't be CEO." Don't you think that's true?
Thinking, "It won't be me," is unacceptable—you must treat the company 15 years from now as your own responsibility. Envision the detailed state of your company 15 years out, and work backward to determine what you must do today, next year, and in five years. Calculate and prepare backward from 15 years in the future. Do you want to remain second or third in your industry? Say, "In 15 years, we will become number one in Japan—no, number one in the world!" Without that ambition, you shouldn't be a CEO. Isn't that unfair to your employees? If someone without the ambition to take a second- or third-place company to number one becomes the leader, I feel sorry for the employees. The company will decline in no time.
When we were in our 20s and 30s, major Japanese corporations seemed up in the clouds, but today they are failing one after another. In traditional Japanese virtues, people praise modesty. But someone who is just modest shouldn't be CEO. A modest president is frankly not a leader. A true leader says, "I am charging forward—follow me! I will become number one—follow me!" If such people do not lead, the Japanese economy will flounder. Over the past 30 years, the Japanese economy has lost vitality precisely because of a lack of this forceful drive. So think seriously about what lies 10 or 15 years ahead.
As I said from the start, we are entering the superhuman era, venturing into uncharted territory for humanity. Past experience will not be very useful. You need the mindset to disregard what senior predecessors said; believing something just because a senior said it will lead to being replaced. An era of 100 trillion agents and 1 billion humanoids is coming. Common sense will change. Operating on the premise that everything will change, think seriously about your company, your subordinates, and your family 15 years from now, and prepare right now. I have shared many thoughts today. You might think, "He's persistent, enough already, I'm tired of hearing his convictions." But I spoke from the bottom of my heart today. I hope this helps you even a little. Thank you very much.
