thesis.

Nation states are feeling accelerationism more than they'd care to admit. The military might is real, but it corrodes from within. The veil is thinning, new powers and great houses are forming, and the prima authority is already being challenged.

China keeps expanding its energy output and is winning where the US is failing, often by taking from the US with ease, and likely the other way around too. Both powers consolidate and the world wraps around them, but they aren't what makes the world move.

Nvidia is worth over $4 trillion, and it invests in the labs that buy its chips: up to $100 billion announced for OpenAI, up to $10 billion for Anthropic, with Google as a partner too. Your average normie sees money cycling inside one sector and calls for the bubble to pop, as if it's all just something being sold.

But the demand for compute is abundantly clear and still growing. It isn't falling off the way they wish. It isn't a fad. It's reshaping everything we operate on.

The entire US economy, seemingly overnight, became the whim of giants like Nvidia. They need the US less every year; the US needs them more. Slowing down in this climate means certain death, so the government is obligated, and its path is sealed.

The Pentagon's 2026 blacklisting of Anthropic, still being fought in federal court, was the first public test of that arrangement. Every player knows the game, and none of them can stop playing. All empires end.

Power consolidates where it is demanded, not where it is projected. Fear still works, but it can only be propped up as long as the populace allows.

Many Americans smell the corruption. They feel treated like statistics, not people. They can't afford their dailies, while elites who've had their fill keep taking from the workers they're now replacing. It sits under everything, and it's building.

Scarcity doesn't set Bitcoin's price; the currency it's measured in does. The dollar hasn't been backed by gold since 1971, and gold itself has more than doubled since 2020 as people look for a way out. We're past the point of going back.

People can buy gold, metals, Bitcoin, anything, yet most stay in dollars out of fear. They don't see the loss, because a dollar has no price ticker. One dollar is always one dollar. It just buys less every month.

What used to be split across your team, your manager, and the guy down the hall nobody was sure about but who'd been there eleven years consolidates into margin. Margin goes up, headcount goes down, and nobody on the earnings call says the word "human," because it isn't relevant to the slide.

BlackRock, Vanguard and State Street don't really invest in companies. They invest in the structure of accumulation itself. The index fund is sold as access but engineered as concentration: retail buys in after CNBC tells them to, and the institutions were there at the foundation.

Bitcoin crossed $100 in 2013 and $100,000 in 2024. Whoever got in at $100 isn't having the same conversation as whoever got in at $100,000. AI is the same: the biggest gains happen in private rounds before a company goes public, and most people only read about them afterwards.

The tools that make wealth now are things most people can't access, can't understand, and aren't invited into. By the time they're invited, the early ones have already eaten. Nobody is coming to fix it, because the people who could are the ones profiting from the gap.

Protocol creators, compute owners, founders who built the rails everything else runs on: wealth is reforming around entirely different axes. The old gatekeepers are buying into the new structures as fast as they can, which only accelerates the concentration they claim to regulate. The game rewards it, and the game doesn't care who's playing.

The human brain runs on about 20 watts, tuned by billions of years of evolution. Frontier models train on clusters that draw tens to hundreds of megawatts, on whatever Nvidia shipped last quarter. Both hit ceilings. Both want more.

The race for GPUs isn't a tech trend. It's the arms race of this century, and every major government now treats the chip supply chain as the most strategic infrastructure since oil. Intelligence wants to expand, the physical world pushes back, and everything in the next decade routes through this bottleneck.

Training a frontier model costs hundreds of millions of dollars and is heading toward billions, so only a few labs can build them. Everyone else rents access. That's stratification applied to intelligence itself. If your AI runs on someone else's compute, you don't own your intelligence. You lease it, and leases can be revoked.

Bittensor, Akash and other distributed GPU networks bet that compute doesn't have to stay centralized. Whether they can compete with the datacenter is an open question. The attempt matters, because the alternative is a world where three companies decide who gets to think at scale.

What's embedded in human DNA pulls toward being useful and doing things; when we become passive, we become depressed. So the hard question ahead is whether we keep our humanity as we've always known it, or modify it to become more. What that "more" means is still being worked out.

Whether this transhumanist kind of performance arrives isn't a question. It's already here, invisible to the many who don't feel any need to name it, and it exists all the same.

The ones who modify have always outperformed the ones who don't: the first tool, the first written word, the first prosthetic. The difference now is the rate. The gap between those who adopt and those who resist no longer takes a generation to open.

They know this is coming, and they can't make major changes all at once, so they boil the frog. Acceleration makes slow boiling impossible. So they boil faster, the people say what the fuck, stop boiling me, and they've had enough of the boiling.

This isn't an investment class. It's the direction.

Land, then labor, then information: each era priced what it couldn't get enough of. As AI takes on more of the routine labor and output, the average American worker needs something else to bring, and the need for human input still has to be met.

What machines can't generate on their own is a person actually being there: their presence, awareness and attention.

Every person is a unique, branching life: individual spirit and difference, both deterministic and entirely chaotic. None of that reduces to attention. But attention is the part an economy can measure and trade, so it's where the transition starts.

It's dark. It's dystopic. It's current. It doesn't have to operate within dystopic mediums forever. But the transition and acceleration are undeniable, and it is not waiting for permission.