Utopia Is Already Under Construction — So Why Are People Still Struggling?

Utopia is already under construction — so why are people still struggling? August 2026. On one side, they are trying to build the sun on the ground. To be precise, it is not that they are making one more star called the sun. It is the attempt to reproduce, inside a machine on Earth, the principle at the sun's core — fusing hydrogen nuclei to release enormous energy. Fusion, which sat in laboratories for decades, is now crossing bit by bit into the language of power plants and industry. The U.S. Department of Energy, in the national fusion roadmap it released in June 2026, set the goal of supporting the expansion of a commercial fusion industry by the mid-2030s. What is interesting is that in the process it explicitly named high-performance computing and artificial intelligence as core technologies for accelerating fusion development. Companies are not waiting. Microsoft signed a contract to buy electricity from the first fusion power plant Helion intends to build. The target is operation in 2028, with an initial output goal of 50MW or more. Helion broke ground on an actual plant site in 2025. Google, too, signed with Commonwealth Fusion Systems — agreeing to purchase 200MW of fusion power from the first commercial plant CFS intends to build in Virginia, and investing further in the company. Why are the world's largest technology companies suddenly talking about power plants? Because of AI. Follow AI all the way down and you arrive at electricity. Bigger models need more computation, more computation needs more semiconductors and data centers, and data centers need staggering amounts of power. The International Energy Agency projects global data-center electricity consumption to more than double from about 460TWh in 2024 to about 945TWh in 2030. The most important driver of that growth is AI. In the United States, data centers are projected to account for nearly half of all electricity demand growth through 2030. So at some point the AI race stops being a software race. You need companies that build good models. You need chips to run the models. You need data centers to install the chips by the hundreds of thousands. You need power plants to feed those data centers electricity twenty-four hours a day. And as AI grows stronger, that AI in turn controls fusion plasma, searches for new materials, designs power plants, and optimizes the grid. Energy makes AI, and AI makes better energy systems. "Energy → computation → intelligence → science → better energy → more computation → stronger intelligence." A self-amplifying structure rare in human history is being built. OpenAI likewise defines the core foundations of AI competition as chips, data, energy, talent. For an AI company, electricity is no longer a server-room utility bill. It has become something close to a strategic resource that determines national competitiveness. Stare only at this picture and it is easy to fall into enormous optimism. If cheap energy overflows and AI and robots take over most labor, how abundantly will humans live? When generation costs fall, the cost of purifying water falls. Heating and cooling get cheaper. Running factories gets cheaper. Transport and logistics get cheaper. If AI can mass-replicate the expertise that once lived only in doctors, lawyers, accountants, developers, and teachers, the price of intellectual services can fall too. When AI and robots and abundant energy move together, the price structure of countless things we now consider expensive can collapse. Perhaps we stand, for the first time in history, at the entrance of an era where we can truly ask this question: must human beings sell most of their time for survival? But turn off the computer screen and step into the streets of August 2026 and an entirely different world appears. Far too many people are struggling. Korea's household credit reached 2,019.8 trillion won at the end of the second quarter of 2026 — up 25.9 trillion won in a single quarter. The household survey for the second quarter of 2026, released just yesterday, shows real household income up 1.5% from a year earlier. By the numbers alone, things are improving. But descend a little and a different reality appears. The bottom 20% of households had average monthly disposable income of about 1.09 million won against consumption expenditure of about 1.37 million won. Their average propensity to consume is 125.3% — meaning that, on average, staying alive requires spending more than what they have to spend. In the same survey, interest costs rose 12.1% year over year. On one side, humans reproduce the energy principle of the sun. On one side, AI replicates human intellectual labor. On one side, hundreds of billions of dollars flow into data centers and semiconductors and power plants. And on the other side there are people calculating this month's living costs, calculating jeonse-loan interest, calculating rent, worrying about finding work, worrying about closing their shop, worrying about next month's card bill. These two scenes look like a contradiction. In fact they are not. Because technologically we are entering the age of abundance while socially we still live in the age of scarcity. The speed of technological change and the speed of social change differ. AI models change every few months. Semiconductors change generations every few years. Data centers are built in years. Power technology advances fast as well. But housing systems, labor markets, education, taxes, social security, corporate organization, and the structure of asset ownership do not move that fast. We are building twenty-first-century production technology while still largely using twentieth-century modes of distribution, labor contracts, and social institutions. So strange things happen. One person using AI can now do the work ten people once did — and instead of working two hours a day, that person may find themselves competing against ten people. A company doubles productivity through automation — and the gains may go first to shareholder value and market share rather than returning as halved working hours. AI could push the cost of education toward zero — but if the status competition of universities remains, parents still pour enormous money into education. Technological scarcity is disappearing while the institutions and competitions that manufacture social scarcity may persist. So rising productivity and human happiness are not the same proposition. Here the future forks. In the first future, AI and robots and cheap energy actually return time to human beings. People need less and less to cling for life to work they hate in order to eat. Anyone can access quite good education, medicine, and expertise. A small team — even one person — can build what once took a giant company. The production costs of housing, cities, transport, and energy fall. The greatest luxury humans gain then may be neither cars nor designer goods. It is time. Time to eat with the people you love. Time to play with your child. Time to read. Time to walk. Time to make something. Time to do nothing. And time to think about what your life is. This is the most important face of the technological utopia as I see it: not a society that gives people more things, but a society that gives people back their own lives. But a second future is also possible. Energy is abundant. AI is immensely powerful. Factories run themselves. Goods are cheap. And every one of those means of production is owned by a tiny few. If a handful of companies or states control energy and computing and models and robots and data all at once, the most abundant society in history could become the most centralized society in history. A society where what you need is nearly free but what you can decide for yourself is nearly nothing. A society where no one starves but no one chooses. That future is entirely possible too. So the real political question of the AI era is not simply "will the jobs disappear." To whom does the productivity accrue? Whose is the time AI creates? Whose are the profits automation creates? Who can access computing? Who can use energy? Who can control AI? Who sets the goals of the system? These questions may be as important as money or elections, and in the long run as important as an ideology or a constitution. Because future power may not come simply from having the most money. It may come from being able to determine the objective function of the system connecting energy, computation, intelligence, and production. Therefore the question we must ask now cannot stop at "will AI replace humans." A larger question is needed: what will we use this staggering productive power for? At this very moment, fusion laboratories are trying to hold plasma above one hundred million degrees. In a data center somewhere, tens of thousands of GPUs spin at once. AI searches for new proteins, new materials, writes code, optimizes factories, computes the grid. Humanity is building productive capacity at a level never before imagined. And at the same time, someone tonight will again lose sleep over rent and loans and living costs. Between those two scenes — that is 2026. So I do not think of now as the completed AI era. I think of it as the greatest transition period. Technology has arrived in the future first, and people still remain in the present. And the most important task of the coming decades will not only be making technology faster. It will be delivering the abundance already being created all the way into a human being's day. Even if fusion succeeds, that alone does not bring utopia. Even if AGI is built, that alone does not bring utopia. Even if robots run every factory, that alone does not bring utopia. Those things only create utopia's conditions of production. The final step must be done by humans. Making the technology usable by anyone, making the growth of productivity flow into actual ease of life, making it so that people do not exist for the system but the system exists for people's lives. In the end, the measure of the most advanced civilization will be neither how enormous an AI it built nor how much electricity it generates. I believe it will be a simpler measure: how freely an ordinary person in that society can live their own life as they wish. To wake in the morning and talk with the person you love, do the work you like, learn something new when needed, make something, travel, start again after failure, and at some moment sit under the sunlight with no purpose at all. Perhaps the reason we are building the sun on the ground, making machines intelligent, automating factories, and endlessly seeking more energy must, in the end, lie there too. So that people can live more humanly. As of August 2026 we have not yet arrived. If anything, the struggling are everywhere. To speak only of the future utopia while looking away from that fact is hollow. But to stare only at present pain while ignoring the possibility of the enormous abundance to come is wrong as well. Both scenes must be seen at once. Humanity is growing unbelievably powerful. And yet human life has not yet improved to match. The history ahead will likely depend on who closes that gap, and how. And perhaps the most important thing we are building right now is neither artificial intelligence nor an artificial sun. It is the blueprint of a new society — for what kind of life humans will live once they are freed, little by little, from survival.