GMO expects new share supply to end the AI bubble before AI revenue disappoints

Inker and Pease sort bubbles by what kills them. The British railway boom died of too much railway: excess investment sank returns on capital in the sector once demand fell behind ^[7]. The 2000 internet bubble, in their account, died of too many shares. Fiber capacity did outgrow internet traffic, but the market rolled over on a flood of internet-related share issuance before that was common knowledge and before disappointing news on internet growth arrived ^[8]. GMO's guess is that AI will end up with the railway problem too, with demand failing to keep up with supply and returns on the build-out disappointing ^[10]. Most of the letter is about the internet version ^[14].

GMO's SpaceX argument is about the stocks buyers sell to pay for it. Buyers have to offer the going market price, and since most investors do not hold much idle cash, GMO expects them to raise it by selling liquid equities ^[13]. What that does to the wider market depends on where the sellers' proceeds go. Cash reinvested in other listed shares offsets the selling; cash sent to lenders or spent by endowments does not. Each tenth of the roughly $2 trillion that leaves equities that way is about $200 billion of other stock sold with nothing reinvested to offset it ^[1].

The June date can go three ways: holders sell and spend the cash, as GMO expects; they sell and buy other stocks, and the index absorbs the trade; or they keep their shares. The letter does not say how much of the roughly $2 trillion is Musk's own stake ^[11].

The counter-case is that the date is public. Anyone who reads GMO's exhibit can sell ahead of it, and the pressure would then arrive spread over months, with no single day to call the top. I think the supply argument is the strongest part of the letter, because it names where buyers get the money and why sellers need cash. It is wrong if SpaceX's last unlock and the rest of the coming issuance pass with the broad market steady. That would mean buyers found the money without selling other stocks, against GMO's assumption about idle cash ^[13].

A date matters because of how managers are judged. The tracking error from avoiding bubble stocks grows the longer a bubble runs, and "nothing infuriates clients more than seeing everyone else get rich as the bubble inflates," Inker and Pease wrote ^[4]. On their own record, they wrote: "We generally consider ourselves much better at spotting bubbles than calling their tops" ^[5].