Are we in an AI bubble?

America’s Nvidia just became the world’s first firm to break the $5T valuation barrier, three months after it became the first to pass $4T. So the AI chipmaker is now worth…

• More than every GDP besides the US and China

• More than two Canadas, and

• Believe it or not, roughly $5T more than Intrigue.

Why? Markets were responding to some spicy remarks, first from Nvidia’s Jensen Huang, who told a pack of nerds at Nvidia’s DC conference that he’s now sitting on…

• $500B in AI chip orders (including for seven US government supercomputers), and

• New partnerships with Finland’s Nokia, plus US defence tech player Palantir, while

• Nvidia’s newest Blackwell chip is now in full production and “shipping like crazy”.

Second, President Trump told Air Force One reporters en route to meet China’s Xi in South Korea that his high-stakes Xi chat would cover Nvidia’s “super duper” Blackwell chips above, again hinting he might approve (downgraded?) Blackwell sales to China!

Anyway, while this fuels an epic AI boom, there’s another b-word quietly circulating, and we’re almost too scared to whisper it out-loud: it’s Ben Affleck’s Batman reboot bubble.

So… are we in an AI bubble? Some quick reasons why you might conclude no…

• i) Maybe this boom is different given it rests on real, foundational AI advances (though dot com, housing, and even tulip investors variously said the same thing)

• ii) Nvidia is actually profitable, posting Q3 revenue of $35B (though that rests on hyperscalers continuing their reported $600B p/year capex), and so…

• iii) Maybe we’re just seeing a smart, accelerating, capitalist flywheel.

But now some quick reasons you might conclude there’s a bubble:

• There’s sobering research from places like MIT, finding 95% of big corporate AI initiatives still show zero return

• There’s maybe circular financing rather than a flywheel at play (Nvidia invests in OpenAI, which buys cloud from Oracle, which buys chips from… Nvidia), and so…

• There are smart people, whether Ray Dalio or the Bank of England, now sounding the alarm (though smart people can be wrong — see Affleck above).

So stir in some hype, some sky-high valuations, and FOMO-fuelled capital chasing uncertain returns, and maybe you’ve got yourself a nice little bubble there, pal.

But then… would a bubble even matter to your favourite ex-diplomats? Well, yes.

First, there’s the concentration risk: US stocks now make up ~56% of the world’s public market cap; America’s Mag7 AI stocks in turn make up a third of the US market cap alone; and their AI infrastructure build-out alone is driving a ~third of all US economic growth.

So any ‘pop’ might rattle not just the US economy, but the world’s economy. It’d also hit a key pillar (US tech dominance) now balancing the world’s concerns about US political risk.

Second, an AI pop would have downstream impacts on other sectors like energy, where there are already countless billions in AI-related energy projects in the US alone.

And third, a pop would also shape US-China competition, which is already in turn shaping this century: Nvidia’s chips are now arguably America’s biggest source of leverage.

But the result of that leverage is the two rivals are now bifurcating into two parallel, algorithmic blocs built on parallel stacks and supply chains. So any AI pop — inevitably slowing US investments and therefore advances — could end up less like the global 2008 meltdown, and more specifically shifting our world’s centre of gravity even further east.