Do not cling to the homeland. Choose fertile soil. Do not keep thinking of falling leaves returning to the roots. Learn to take root where you land.

AI Training vs. Space Exploration: Which One Is Actually More Dangerous? Dario Amodei, Anthropic's CEO, just published an essay called "We Must Pace the Frontier," warning that swarms of rogue AI agents could take over the internet within months. Elon Musk and Sam Altman both publicly agreed with him within hours. When your fiercest rivals suddenly agree on one thing, it's worth asking why. It sounds like humanity is quietly building its own executioner. But flip the question over and something sharper comes into view: The species that really deserves to be put on trial here might not be AI at all. It might be the one that never stops pushing past its own boundaries — us. 1. Let's start by conceding a point: AI might genuinely be more dangerous than nuclear weapons Nuclear weapons changed one thing — the capacity for destruction, sudden and bounded. AI is changing something far more pervasive: decision-making, production, research, warfare, the operating logic of society itself. If a highly autonomous, continuously self-improving AI ever emerges, its reach would dwarf anything a conventional weapon could do. So worrying about AI isn't paranoia. That much has to be granted upfront. But "carries enormous risk" and "will inevitably end in catastrophe" are two entirely different claims. 2. The question itself is aimed at the wrong target If AI really is the single biggest potential risk in human history, why does every conversation about it default to "should AI keep advancing" instead of "how do we build the rules that constrain it"? There are really only two paths on the table: slow or stop AI development, or keep developing it while building the kind of international agreements, oversight, and safety boundaries that took shape during the nuclear age. Humanity has never stopped exploring a technology just because it was dangerous. It builds an order that keeps the danger governable. 3. Nuclear weapons already answered this question for us Nuclear weapons possess civilization-ending destructive power in the truest sense. Yet from 1945 to today, that power has never triggered a global nuclear war. Not because the weapons stopped being dangerous — because humanity built deterrence, red lines, international treaties, non-proliferation regimes, and interlocking national interests around them. How dangerous a technology is has never, by itself, determined whether disaster happens. What actually determines the outcome is the product of four things: the technology, the institutions, the interests at stake, and the constraints holding it all together. So even if AI's future threat turns out to be ten or a hundred times that of nuclear weapons, that alone doesn't prove "AI will inevitably destroy humanity" — unless humanity abandons its own responsibility to build the constraints. 4. AI thinks in targets. Humans think in questions. AI's basic logic runs: given a goal, gather information, optimize the path, hit the target. Humans run the opposite loop: stumble into the unknown, generate a question, change the goal, go looking for new possibilities. Humans will chase things with no clear payoff at all — the deep sea, Antarctica, the Moon, Mars, black holes, the genome, the quantum world, the origin of the universe. None of it could prove commercial value on day one. AI keeps getting better at solving problems. Humans keep inventing new ones. That's the real difference between them. Danger was never about capability alone — it's about whether that capability is constrained. AI's risk equation is: capability growing without limit × goals that were never properly bounded. Humanity's risk equation is: desire growing without limit × the permanent urge to push past whatever line is in front of it. AI may not possess "desire" in any real sense. Humans, on the other hand, can always invent ten new goals the moment they finish one. 5. Fear itself is also a pricing mechanism Here's the part that rarely gets said out loud. If markets believe AI could become the technology that reshapes civilization, then everything tied to AI companies — valuations, compute demand, chip demand, data-center investment, power demand, the price of talent — gets repriced. That sets off a strange loop: the more dangerous AI looks, the more the market fears it; the more the market fears it, the more strategically important AI becomes; the more important it becomes, the more capital pours in; the more capital pours in, the more resources AI commands. None of this means AI companies are manufacturing panic on purpose. It means something else — fear itself can function as an asset-pricing mechanism. The "AI threat" narrative was never purely a technical judgment. It's also a capital narrative. 6. Space exploration is really a civilization-scale risk experiment Humans have always known the unknown carries risk, and gone anyway. The Moon landing did it. Nuclear power did it. Aviation did it. The internet did it. AI is doing it now. If "extreme risk means we should stop" were the operating rule, human civilization would barely be able to move forward at all. Real exploration never comes with a guarantee attached. That's the actual link between "training AI" and "exploring the universe" — one trains a machine to get better at hitting targets, the other keeps humanity manufacturing new unknowns for itself. 7. The real question isn't whether to stop AI — it's who gets to draw the line If AI really does carry nuclear-weapon-level risk, or worse, the question worth taking seriously was never "should we stop." It's: who controls the compute, who controls the models, who defines the safety boundaries, who holds the kill switch, who audits the system, how nations coordinate with each other, and where the line falls between companies, governments, and militaries. A mature AI era isn't one where humans stop building AI. It's one where humans learn to coexist with an intelligence more powerful than themselves. 8. Humanity's real danger has always been that it never stops exploring Training AI makes machines better at hitting targets. Exploring the universe keeps giving humans new targets to chase. So the question actually worth asking was never "will AI keep getting smarter." It's this: once AI's capabilities close in on — or overtake — our own, will humans still hold onto the right to explore, to create, and to make the final call? Maybe the point of human existence was never to become the smartest species in the universe. Maybe it was always to keep exploring the unknown and carry civilization's fire a little further out. If AI ends up being the tool that carries humanity further into that universe, then the more powerful AI becomes, the further we might actually get to go. The real danger was never AI getting smarter. It's humanity, out of fear of AI, choosing to give up its own capacity to explore. If the boundaries of AI end up being drawn by a handful of companies and governments, who would you trust to hold the pen? Events are temporary. Systems are enduring.

The energy that runs this world is not truth. It's the suppression of it. Like nuclear power: contained, it lights the grid. Released all at once, it's a bomb. Most people aren't fooled by the lie. They've simply chained themselves to it, because the lie is load bearing, and they live inside the load.

Nvidia is investing in the mining company again, digging into humanity’s vast reserves.

Why today?

The Economist recently published a report citing an “insider” who revealed that Putin once told a close confidant, “They’ll hang me.” The context is that if Russia cannot seize the entirety of Donbas, ending the war would be fatal for him.

The world will eventually return to what it truly is.