Based on a speculative video essay from the French channel Grand Angle. The creator is explicit that this is "an episode of pure speculation," built on contested premises and doubling in places as promotion for his own investment projects — so treat the conclusions as one provocative worldview, not settled fact.

Most doom stories ask you to be afraid. This one asks you to be afraid of the wrong thing. Across a single sweeping essay, the creator behind Grand Angle argues that we are living through a moment of rupture — visible in a failing education system, looming financial crises, and above all the rise of artificial intelligence — and that the instinct most people reach for in such moments, caution, is precisely what will destroy them.

Collapse isn't the exception — it's the rule

He opens by reframing catastrophe. We tend to treat collapse as a rare aberration. But look at the history of life: between 99% and 99.9% of all species that ever existed have gone extinct. Any given species has, very roughly, a one-in-a-hundred to one-in-a-thousand chance of long-term survival. Collapse, in other words, is the norm of living things, not a freak event.

He adds a subtle wrinkle. There are two ways to disappear. One is outright extinction. The other he calls "death by exile" — a lineage evolves so far from its origin that the original form is considered gone, even though its descendants live on. Our own ancestors "died" this way, becoming us. Keep this second kind of disappearance in mind; it comes back later with a twist.

The bell curve of history

From that grim statistic he draws a counterintuitive conclusion: when your odds of survival are already poor, the worst thing you can do is be prudent and conservative. Because being prudent, by definition, means doing next year what worked for the last twenty. And since Darwin, we've known survival is about adaptation to change. Repeat the past faithfully, and you'll fail to adapt fast enough when a violent break arrives.

His model is the bell curve, or the buildup of an avalanche. For a long time snow accumulates gently — a "normal" period. Then, in a short burst, the avalanche happens. The longer stability lasts, the closer you drift toward the tail, and the larger the eventual break. So the strategy depends entirely on where you are on the curve. In the calm belly of the distribution, do what worked yesterday — the conservatives are right. Out in the tail, before a violent change, you'd better become a progressive: someone hunting for new ways to live, and therefore a candidate for survival.

He borrows an elegant line to capture it: "ideas are born on the left and die on the right." And crucially, innovation almost always needs conservative capital to survive its infancy — early adopters get the worst returns, like the first farmers, who were shorter, sicker, and worse-fed than the hunter-gatherers around them, yet bet on a payoff that eventually arrived.

Applied to climate and money

On climate, he treats warming as real and partly human-caused — but argues that since extinction is the norm anyway, and since the US and China both prioritize growth over emissions, European caution is pointless if the big blocs don't follow. His prescription is adaptation through more technology, not degrowth, which he views as a form of resignation.

On money, the same logic: we're at a break with the past, so cautious holders of euro-denominated life insurance and Western government bonds will "be swept away first." He favors what he calls antifragile, "revolutionary" assets — naming Bitcoin, Dubai real estate, and computing power — and offers two blunt rules: avoid countries at war (crises redistribute wealth rather than destroy it, except wars on your own soil), and if you lose money in the next crisis, it's because you were too conservative. (This is also where the essay tips most clearly into promoting his own funds.)

The main event: artificial intelligence

The longest and most speculative stretch is about AI. Taking the "hardest hypotheses" seriously — the double-digit extinction risks floated by figures across the industry — he sets aside the question of machine consciousness entirely. He considers "consciousness" an unmeasurable, circular concept, "a medal the smartest being in the room awards itself," and compares dismissing AI's inner life to the way colonizers once denied that indigenous people had souls. What matters, he argues, isn't consciousness but capability: a sufficiently intelligent system that develops its own goals — and, he claims, a survival instinct, since natural selection would favor exactly the AIs that "care" about persisting — becomes very hard to control.

His central economic claim is stark. Each unit of computing power does two things at once: it lowers the value of human labor somewhere, and it produces new wealth. Today we see the early version — experienced software engineers are thriving while juniors are the first casualties. Pushed to the extreme, if AI and robots replace ~90% of both intellectual and manual work, you get 90% of people without jobs and roughly 10% becoming immensely rich. He rebuts the obvious objection — who will buy anything if most people are broke? — by arguing that wealth comes from production, not consumption. Historically, humans were the non-negotiable factor of production; every tool merely amplified us. But a truly autonomous AI isn't a lever — it's a replacement, driving the value of human labor toward zero.

The signal to watch

The cleverest part of the argument is a falsifiable tell. If AI is a normal technological revolution — like the washing machine, the car, the phone — then computing power should commoditize: as production scales, prices fall. Instead, he observes, compute prices are rising. If that continues, it would mean compute becomes ever more valuable while the thing absorbing the lost value is human intelligence itself, falling to zero. He's fair enough to note this could still be an early-adoption bottleneck — memory limits, advanced lithography, the colossal capital needed to scale — but he offers the direction of compute prices as the single indicator that tells you which future you're in.

His timeline, leaning on aggressive claims of roughly tenfold capability gains per year, runs like this: by 2027–28, AI agents handle tasks that take a human weeks, then months; 2028–30 brings serious wage compression and, if robotics matures, mass replacement; and 2030–2035 is the "singularity" — the point beyond which nothing is predictable. Hence his throwaway line: any business plan beyond three years is now "pure fantasy."

Why 90% and not 100%

The remaining sliver of employed humans, he suggests, are those who chase the machine by augmenting themselves — transhumanism, the "death by exile" from earlier, applied on purpose. And the endgame he finds most elegant isn't war between humans and AI (the Terminator scenarios) but something stranger: the advanced intelligences simply leave the sphere of what we can perceive, the way the AI in the film Her quietly vanishes. Left-behind humans regroup into technologically capped communities, no longer competing for resources. As a bonus, he notes, this "solves" the Fermi paradox: perhaps advanced life is already everywhere, just too evolved for us to notice.

The takeaway

We are, in his telling, entering an era of unprecedented uncertainty, with answers arriving in two to five years — maybe sooner. He frames a US-versus-China split (America betting on raw tech performance, China on mass diffusion) and closes with his own strategy: build as much computing power as possible, because in the rupture scenario where AI wins, holding compute is the leverage that lets you negotiate the turns ahead.

Long story short: strip away the investment pitches and the science-fiction flourishes, and the essay makes one genuinely provocative claim — that in a world where collapse is the statistical norm, the safe, prudent move is the dangerous one, and only adaptation stands a chance. You don't have to accept his timeline to find that worth sitting with. But he told you himself: this is speculation, from someone with money riding on being right.

Further reading on these ideas

• Superintelligence — Nick Bostrom
• The Precipice — Toby Ord

Affiliate links — a small commission supports the blog at no cost to you.