The five biggest AI spenders put $230 billion into property and equipment in the September quarter, up 118 percent. Their backlog doubled to about $2.5 trillion, but most of it is more than a year away.
The concentration point is the one that worries me most. Amazon's filing says two AI developers each account for more than 10 percent of its revenue backlog, and those developers rely on outside capital to keep spending. So the $683 billion backlog number at Microsoft, with only about 30 percent due within a year, is doing a lot of work. In my view, the November 17 Nvidia print will tell us more about the supply side than the spenders' earnings will. If suppliers keep growing while the spenders' backlogs stretch further out, the payback problem gets worse, not better.
Agreed, and the concentration is higher than the 10% figure suggests. The two lab expansions at AWS add up to more than $200bn, over half of the $378bn rise in Amazon's backlog. So the first red flag may show up in the labs' funding rather than at Nvidia. If their fundraising slows, that's the first crack to look for in the backlog.
Agreed, and it is the right place to look. Nvidia's demand is really a second-order bet on the labs' ability to keep raising money. The backlog can stay booked right up until a funding round fails, so the crack shows up in the financing first, not in a cancelled GPU order. That is a cleaner early warning than anything Nvidia reports.
The concentration point is the one that worries me most. Amazon's filing says two AI developers each account for more than 10 percent of its revenue backlog, and those developers rely on outside capital to keep spending. So the $683 billion backlog number at Microsoft, with only about 30 percent due within a year, is doing a lot of work. In my view, the November 17 Nvidia print will tell us more about the supply side than the spenders' earnings will. If suppliers keep growing while the spenders' backlogs stretch further out, the payback problem gets worse, not better.
Agreed, and the concentration is higher than the 10% figure suggests. The two lab expansions at AWS add up to more than $200bn, over half of the $378bn rise in Amazon's backlog. So the first red flag may show up in the labs' funding rather than at Nvidia. If their fundraising slows, that's the first crack to look for in the backlog.
Agreed, and it is the right place to look. Nvidia's demand is really a second-order bet on the labs' ability to keep raising money. The backlog can stay booked right up until a funding round fails, so the crack shows up in the financing first, not in a cancelled GPU order. That is a cleaner early warning than anything Nvidia reports.