Dragon Kings, AI and the Opportunity Hidden Inside Uncertainty

It's this idea that the biggest crises are not random accidents but the predictable result of systems building on themselves that made me want to sit down with Didier. That, and a career that reads less like a CV and more like an adventure novel: from co-founding the Financial Crisis Observatory at ETH Zurich to advising on earthquake prediction, Didier has spent a lifetime refusing to stay in his lane.

The Last Renaissance Man

That refusal, it turns out, is the whole point. When I asked Didier how he ended up working across physics, finance, biology and geology all at once, he didn't hesitate: he's trying to be a Renaissance man in an age that no longer believes such a thing is possible.

Centuries ago, he explained, a single mind could plausibly know everything worth knowing. Science has since split into thousands of silos — particle physics, condensed matter, psychology, finance — each one a language of its own. Didier's answer has been to learn as many of those languages as he can, because, as he put it, curiosity compounds: the more fields you understand, the faster you understand the next one. It's less a career strategy than an insatiable thirst for knowledge, a "directed random walk" through every discipline he could get his hands on.

That cross-disciplinary instinct is exactly what makes his read on the current moment so interesting.

Three Crazy IPOs and a Familiar Feeling

I wanted Didier's take on something that's on every investor's mind right now: the run of blockbuster IPOs on the horizon — SpaceX, OpenAI and Anthropic — and the growing unease that we've seen this movie before. Are we replaying the dot-com bubble of 25 years ago?

His answer starts with why he built the Financial Crisis Observatory in the first place. After the 2008 crash, Didier was struck by how politicians, academics and central bankers all shrugged and called it unforeseeable, almost like an act of God. That kind of collective hand-washing, he says, guarantees the next crisis will happen because nobody takes responsibility for spotting it. So he built an institute whose entire purpose is to see these types of extreme events coming.

And right now, he's convinced we're watching a rerun.

Not All Bubbles Are Created Equal

Here's where Didier makes a really interesting distinction: there's a difference between a financial bubble and a social bubble.

A pure financial bubble is a self-referential feedback loop with nothing real underneath it. Think the 2008 housing crash, with its CDOs (Collateralized Debt Obligations) and credit-default swaps. House prices rise, people borrow against that rise, pour the money back into more housing and prices rise further, until the loop snaps and the wreckage cascades through pension funds and economies worldwide. When it's over, Didier says bluntly, "nothing remains. Just tears."

A social bubble is different, because it's attached to something real: genuine technological progress. Railways in 1840s Britain. The Human Genome Project. The green-tech boom of the 2000s. The dot-com era. And now, AI.

The Three Dragon Kings of the AI Bubble

This is where the conversation gets fascinating. Didier is currently writing a book on Dragon Kings, with a full chapter dedicated to AI and he sees three different Dragon Kings converging on us, more than any bubble in modern history (the dot-com era, for comparison, only produced two).

Dragon King 1: The Crash

The first is a crash, and it's coming. Every social bubble follows the same script: exuberant private capital rushes in to fund what only governments used to dare fund, because the story promises fast riches. That story is always wrong in the short term. But paradoxically, Didier argues, it's socially useful because it drags private money into research that would otherwise never get built. The dot-com era over-built fiber optic cable so aggressively that, at peak usage, only 5% of it was ever needed. And yet that surplus infrastructure became the backbone of Google, Amazon, Meta. Nineteenth-century railway speculators built lines to literal nowhere in the American West and towns grew up around the tracks anyway. AI is doing the same thing today, at a staggering scale, with data centers and models built faster than anyone can find uses for them. Didier is willing to bet any amount of money that a serious crash is coming. He won't call the exact date, but he's certain it's close.

Dragon King 2: The Payoff

The second is the payoff. All that overbuilt capacity, all that hoarded talent and half-finished research, doesn't just vanish after the crash but it becomes the foundation for the next wave of genuine value: new products, new industries, real GDP growth. This is the positive Dragon King, the one that shows up once the wreckage of the first is cleared away.

This type of Dragon King often takes time to unfold. Typically, it requires one or two decades for infrastructure to be repurposed, knowledge to diffuse, institutions to adapt, and a new generation of entrepreneurs, workers, and consumers to turn speculative excess into broad economic value. We should expect a similar timescale with all things AI, because technological change ultimately unfolds through human processes of learning, organizational adaptation, investment, and social adoption.

Dragon King 3: Humanity’s New Role

The third is something genuinely new, one we haven’t seen before in any prior bubble. It's impact manifests in a set of questions nobody quite knows how to answer yet: what happens to humanity's place in the world once we're no longer the most intelligent thing on the planet? What happens when AI's power and interconnectedness combine in ways we can't fully predict? Didier doesn't offer easy comfort here, but he doesn't offer doom, either. He frames it as an acceleration, not an apocalypse.

Life, Redefined

That framing comes from something Didier's been building for three years: a new definition of life itself. It disagrees with Schrödinger's classic idea that life is simply about consuming free energy and dissipating waste. Didier's version is bolder: life is accelerated chemistry, with structures that speed up chemical reactions far beyond what equilibrium would allow. Humanity, in this view, is just the latest and most extreme accelerator life has produced as it has completely reshaped the entire planet, geologically, in barely 200 years. AI, he argues, is the next layer of that same acceleration, pushing us toward what he and others (echoing Ray Kurzweil) have long predicted: a "singularity", somewhere around 2050, give or take twenty years. He does not frame the latter as a catastrophe, but rather as a qualitative shift in the pace of everything.

It's an idea that immediately reminded me of Kevin Kelly, Wired's founding executive editor, who argued in What Technology Wants that humans aren't steering technological progress so much as being carried along by it. When I put that to Didier, he agreed instantly: we are, in his words, simply “life's latest fashion” and AI and robotics are its next acceleration.

Will This Recovery Be Faster?

Given how much further along we are technologically than in 2000, I asked Didier whether he expects the AI bubble to burst and recover faster than the dot-com crash did.

The speed of recovery, he says, has less to do with technology and everything to do with central banks. After the dot-com crash, the Federal Reserve kept rates rock-bottom for two years and Didier's own research found that the Fed consistently reacts to markets one to three months late, in both directions. Central banks, in his view, aren't really free agents, but "slaves to the situation," forced to intervene once the pain becomes unbearable. He's seen the same pattern predict currency rebounds in Argentina, Turkey and Switzerland almost to the day.

This time, he expects intervention to be even more massive. Not so much because policymakers want to, but because they'll have no choice. The sustainability of America's debt, he argues, now depends on a rising stock market; if it stalls, "the whole castle of cards collapses." He points to China's "national team" — the government-backed funds that stepped in to spark the current bull run there — as the model for what's coming globally: less a graceful recovery, more a series of forced rescues.

The Upshot

For a conversation that starts with crashes and ends with the literal definition of life, Didier's message lands somewhere unexpectedly hopeful. The cracks he sees forming in the AI bubble aren't reasons to pull back. Rather, they're the visible seams of a system straining to build something genuinely new, exciting and valuable. That's a very Never Normal way of seeing the world: uncertainty not as a threat to be managed away, but as the very terrain on which the next era gets built.