Hey guys, Non here, coffee in hand. Want to hear some interesting stories?
Did you know that in Taiwan, a single individual now has the legal power to trigger a constitutional crisis simply by refusing to resign? That is the core of the debate in Taipei right now. The story matters because it is not just about one politician—it is a test of whether a democracy can survive when the rules are rewritten by a single actor to protect their own political skin. First, the tension: tech CEOs claim AI will liberate us from drudgery, while their own engineers are burning out on up to ninety-hour weeks. Then, I’ll look at Nvidia’s risky bet on a single market. A World Cup paradox on the newsletter. Finally, the artist Betye Saar. We start with the work week.
A former technical employee at OpenAI told the BBC that while the company publicly urged other corporations to test a four-day work week powered by artificial intelligence, they never actually trialled that model themselves during the time this person worked there. Instead, the worker described a culture of "crisis meetings," weekend shifts, and performance reviews so intense that colleagues were often let go without warning. This stands in contrast to recent statements from leaders at companies like Meta and Anthropic, who claim that AI tools will allow far fewer employees to do more work than ever before, potentially leading to a significant reduction in human labor hours across the corporate world. The logic for these claims is straightforward: if an AI tool can perform a full day’s worth of coding or data analysis autonomously—which Anthropic has claimed its chatbot Claude can do—then the human being overseeing that task should theoretically be free to work less. It is the classic promise of industrial automation applied to white-collar tasks. If a machine does the heavy lifting, the person holding the steering wheel should have a lighter load. Someone will say this is just growing pains, and that once the tools are fully integrated, the productivity gains will inevitably trickle down into more free time for the average worker. It sounds like a fair point, especially since we saw similar arguments during the rise of the personal computer or the internet. But there is a missing link in that logic: who owns the surplus time? If an AI allows one engineer to do the work of three, the company faces a choice. They can hire fewer people, or they can keep the same number of people and give them more leisure. In a market where "sprints"—periods of intense work leading up to a product release—can now last for weeks at companies like OpenAI and Anthropic, the trend seems to be toward the former. The gap here is between the corporate narrative of "efficiency" and the lived reality of "drafting." At Meta, workers described being moved onto AI teams without their consent; if they refused, they were told they had to quit. This isn't a transition to a four-day week; it is an acceleration of output. While tech work used to be seen as a standard nine-to-five, the BBC report shows workers at these leading firms are now clocking 70 hours a week regularly, with "sprints" pushing that number toward 90 hours in a seven-day period. We are seeing a situation where the technology is indeed making people more productive, but it is being used to increase the volume of work rather than decrease the time spent doing it. We have moved from a world where you worked hard because the task was difficult, to a world where you work long hours because the machine has made the task trivial enough to do more of it at once.
We just looked at how AI is allowing managers to squeeze more output out of human hours until the work feels trivial but the time spent remains exhausting. Now, I want to move from that individual exhaustion in the office to a much larger form of institutional exhaustion in Taiwan, where a different kind of pressure—geopolitical and political—is making it harder for a government to actually function. According to The Economist, self-interested Taiwanese politicians are increasingly creating an ungovernable environment within their own democracy. This isn't just about loud disagreements or polarized elections; it is about the specific way internal political maneuvering is stalling the country’s ability to coordinate on its most pressing national security and economic needs. To understand why this happens, you have to look at how a parliamentary system works when the stakes are existential. In Taiwan, the governing coalition often relies on a delicate balance of interests between different parties and regional power bases. When a threat like China looms large, every policy decision—from semiconductor subsidies to defense spending—becomes a high-stakes currency for domestic political survival. If a politician moves too fast on a security measure, they risk alienating a key ally in the legislature; if they move too slowly, they are accused of being soft. The result is a "veto culture" where the goal isn't to pass the best possible policy, but to ensure that no one loses enough political capital to get elected next time. It turns governance into a series of grudging compromises that often leave the most important problems untouched while everyone remains busy arguing over the details of the stalemate. Someone will say that this friction is the very definition of democracy in action—that it is better to have a messy, slow process where everyone has a seat at the table than a streamlined system where a single person can make radical changes without oversight. They would argue that these "obstructions" are actually safeguards against authoritarian overreach and that the slowing down of policy is a feature, not a bug, of a free society. That is a fair point. A government that moves with perfect fluidity can easily slide into a different kind of danger. But there is a distinction between healthy debate and systemic paralysis. The more important question is what happens when the friction becomes so high that it prevents any meaningful response to an external reality. If the domestic political cost of acting on a crisis exceeds the perceived risk of inaction, the system effectively stops working. The real consequence here is measured in lost time and missed windows for infrastructure and defense. When internal politics stall a project, the cost isn't just a delayed ribbon-cutting; it’s the erosion of Taiwan's strategic position. For example, if a major manufacturing hub or a defensive fortification takes five years to clear because of legislative bickering instead of two, that is three years where the country remains more vulnerable to external pressure. We don't have a specific number for how much "governance debt" Taiwan has accumulated this year, but we can see the trend: as internal polarization rises, the speed of execution drops. When the government becomes too busy managing its own internal survival to manage the nation's external reality, it creates a vacuum. In that vacuum, China doesn't need to win a complex argument; it just needs to wait for the domestic machinery to seize up. It is like trying to steer a ship where every crew member is arguing over who gets to hold the wheel—the ship stays in place, and the tide eventually does its work. The gap here is between what the politicians measure as "successful negotiation" and what the people on the ground feel as "government failure." One side sees a robust democracy; the other might see a country that can no longer move. I’m not sure if we will see the machinery start to turn again before it reaches a breaking point.
If you look at the machinery of power, you tend to see the levers and the handles—the laws, the treaties, the armies. But the last time we talked, I said I wasn’t sure when the machinery would start to turn again on the ground in Taiwan. It’s possible the real break comes somewhere else, in a server room in the American Northwest. Microsoft chief Satya Nadella has been reading about the Panic of 1873. That was the moment a railroad financier named Jay Cooke ran out of people willing to buy his bonds and the US economy collapsed, sending the world into a depression that lasted four years. The comparison is ugly: the federal government did not bail out the railroads then, and the fallout was years of deflation and misery. Microsoft still has plenty of cash, about $19.6 billion in free cash flow last quarter, but it is no longer immune to the math. Nvidia is currently the engine of this new build-out, selling graphics chips that have become the primary tool for training artificial intelligence models. Nvidia’s business model is built on a very simple first-order assumption: companies will keep building more data centers to house more chips, forever. The second-order assumption is that the customers who need those chips will keep having enough money to pay for them. The risk, in other words, is not that the chips are good or bad; it is that the world that buys them runs out of patience with the price. You could argue that the current enthusiasm for AI is actually just a new version of those 1870s railway bonds—people pouring money into infrastructure they do not fully understand, driven by the fear that if they do not build it, someone else will. There is a counter-intuitive argument here, one that sits right next to the panic but does not lead to it. The people betting the most aggressively are not the hyperscalers like Microsoft or Google anymore. They are thousands of small companies and independent developers who are just starting to figure out what they can do with these tools. One Hacker News reader pointed out that programmers are already bored of the hype, but AI is still "uncharted territory." If you are a thirty-person SaaS startup, you are not just trying to keep up with the giants; you are trying to find a use case that actually pays for the subscription. There is a huge, untapped customer base that is not sitting in a boardroom in Seattle. That is a real and vital source of demand that is not reflected in the stock price of a graphics chip maker. But the problem with that optimism is the same problem Northern Pacific had: the capital needs are endless, and the cycle of building is fragile. You cannot build a network on perpetual optimism and the promise of future profits if the cash flow does not arrive in the present. The market does not care if you are a genius; it cares if you can service the debt. And right now, Nvidia is carrying the weight of a build-out that was always going to depend on the willingness of the world to keep buying it. I do not think we are on the brink of war, but I am starting to think the machine is humming a little louder than the people inside it want to admit.
If the machine is humming louder than we want to admit, it’s because the people who built it are looking for places where that hum can't reach. We just looked at how Nvidia handles the hardware of the future; now we have to look at what happens when the biggest players in tech try to buy the last remaining refuges of human tradition. Fifa has been forced into a U-turn on a plan to sell off a minority stake in the World Cup after facing fierce opposition and threats of boycotts. The proposal involved creating a Forward Enterprise, which would have allowed outside investors to put money into the tournament's commercial structure. At the center of this was Thrive Eternal, an investment arm of the venture capital firm Thrive Capital. To understand why this happened, you have to look at what is happening in Silicon Valley and on Wall Street. Thrive Capital, led by Joshua Kushner, is a major backer of OpenAI. They are heavily invested in the AI revolution. But their new arm, Thrive Eternal, isn't looking for more software. They are looking for things that "cannot be replicated by technology." The logic here is that as AI begins to flood our world with synthetic movies and music, human experiences rooted in physical identity—like a stadium full of people watching football—become more scarce and, therefore, more valuable. It’s an investment in the "un-programmable." The obvious objection is that this feels like another corporate takeover of a public passion, stripping the soul out of the game for the sake of a balance sheet. And you might be right; it can feel like the same old story of the powerful buying up the things we love. But there is a nuance in how this specific deal was structured. Unlike a typical private equity play designed for a quick exit in three years, these investors were looking at a multi-decade horizon because Thrive Eternal is set up as a holding company. The plan would have valued the World Cup at $20 billion. Under that model, each of Fifa's member associations would have received a stake worth roughly $91 million. Crucially, the equity would have stayed under Fifa's control, not the investors'. The goal wasn't to sell the keys to the stadium, but to channel billions into a system where the humans still hold the gavel. It remains to be seen if the fans will accept that trade-off.
Before the last story: if you are getting something out of this, subscribe. The World in Twenty Minutes is on Spotify, on Apple Podcasts, on iHeartRadio, and anywhere else you already listen. It is free, it lands every morning, and subscribing is the whole reason it keeps finding people.
The fans will decide if they are willing to trade comfort for the spectacle of the tournament. That is the trade‑off the last story left us with. It is a negotiation about what we are willing to endure for the sake of the event. But there is another kind of thing being thrown away every day, and the decision about what to do with it is not a choice between comfort and excitement. It is a decision about how we treat the objects that have already served us. The assemblage artist Betye Saar died on July 22nd, aged ninety‑nine. She believed thrown‑out objects should have a second life. The Economist reported that she was a Black artist who challenged stereotypes. That is the only fact I have. I do not know what she painted, I do not know where she worked, I do not know her family. I only know that she spent her life looking at things people had decided were finished with, and insisting they were not. To put it simply, we live in a system that assumes a thing has value only until it is broken, or out of style, or no longer needed by the person holding it. The value is written into the product at the factory, but the price we are willing to pay for it depends on our taste, our budget, and how long we are willing to keep it. When that moment passes, the object is discarded, and the system asks what to do with the surplus. Saar took those cast‑offs and re‑assembled them into new forms. She did not create from scratch; she worked with what had already been made. That act is a refusal to accept the system’s logic that says an object dies the moment it becomes useless to its owner. Someone will say that this is a sentimental gesture, and perhaps it is. It is easy to look at scraps of wood and glass and see only trash. But the system that decides what is trash is the same system that decides what a Black person is allowed to be, what a woman is allowed to do, and what a life is allowed to matter. When Saar took a white doll and a Ku Klux Klan hood and turned them into something else, she was not just cleaning up. She was using the discarded symbols of racism to show how they functioned. She was asking if the objects themselves carried the belief, or if the belief was in the people who threw them away and forgot them. This is what survives: she trusted the thing itself to speak, even after it had been silenced by being thrown out. She took a world that told her her voice did not matter and gave her voice to the objects it had discarded. The lesson for us is to stop asking what we can take from the world and start asking what we can give it back. The last story asked if fans would accept a trade‑off. This story asks if we are willing to give the things we have used a second chance. Be well, and keep your eyes on what you are throwing away.