Hello discoverers!

Last year SpaceX made no profit, but lost several billion dollars. Earlier this year it was valued at close to two trillion. I am not an accountant, but even I know those two numbers are not supposed to sit next to each other looking so relaxed about it. A friend shared a short reel breaking down the ‘creative’ bookkeeping that makes this kind of valuation possible. I watched it twice. Still legal, apparently. Still feels like a crime.

By now most of us have some sense that the AI industry runs on a similar kind of circular logic – a handful of companies investing in each other and calling it ‘explosive growth’. In this opinion piece, an anonymous blogger lays out the hype-flated AI economy in an infuriating and also very entertaining way.

It opens with an old joke: a man sells ten dollar bills for five dollars each, and when asked how business is going, he says it’s booming, he just needs more volume. That’s the essay (and the AI economy) in a nutshell.

“For four years we have watched the smartest people in technology reinvent this joke, add a chatbot and agents to it, and raise money at a valuation that assumes the man will one day discover a margin hidden inside the ten dollar bill.”

The old rule of software was simple: as you scale, cost goes down, profit goes up:

“You write the code once. The ten thousandth customer costs you almost nothing to serve. Marginal cost trends toward zero, margins trend toward heaven, and the spreadsheet curves apart into that beautiful pair of diverging lines that venture capitalists have tattooed on their hearts.”

But AI doesn’t do this. Every query burns real electricity, real water, real chips. Revenue and cost haven’t diverged, they hold hands.

What we’re using right now is a heavily subsidised version of AI, priced to build a habit rather than to reflect what it actually costs to run. As the author notes, “when a company files to go public, it has to stop talking like a founder and start talking like an accountant”. At some point, everyone who’s bolted AI onto their workflow is going to discover the real price of the tokens.

That gap is one of five ways the author reckons ordinary people end up footing the bill, mostly without noticing: through the token costs that surface once the subsidies dry up; through the index funds and pensions that own a slice of these valuations whether you chose to or not; through electricity bills padded by data centres on your grid; through tax rebates communities hand over to data centres chasing jobs that don’t materialise; and through water, which unlike power doesn’t get cheaper once the infrastructure’s built.

Of course, none of which means the technology is fake, and the author is careful to say so:

“Electricity was real during the bubble in electric companies. Railroads were real, and the internet was extremely real, while fortunes evaporated in both. A technology can be world changing and a generationally bad investment at the same time, and history is mostly the gap between those two facts being violently closed.”

Some call it a bubble, some call it infrastructure, some just shrug and say, ‘Ask me in five years’. But we don’t need consensus to notice the scale of it. We’re all in the AI casino now, whether we walked in or got carried in on our pension or super funds. As ever, the house is not worried.

And now to this week’s discoveries. – Kai