Money stopped being the bottleneck

A company earning $62.6B in three months does not usually describe itself as constrained. But on Alphabet's late-April earnings call, CEO Sundar Pichai said Google Cloud's revenue 'would have been higher if you were able to meet the demand' — the business is leaving sales on the table because it cannot stand up data-center capacity quickly enough.

A $460B order backlog nearly doubled in a single quarter — most of the growth is already sold, and the limit is how fast it can be built.

The numbers around that admission are large. Google Cloud crossed $20B in quarterly revenue for the first time, up 63% from a year earlier — faster growth than either Amazon's or Microsoft's cloud arms, though both remain bigger in absolute terms. Cloud's signed-but-undelivered order backlog nearly doubled in a single quarter to more than $460B. Total revenue was $109.9B; profit jumped 81%. And Alphabet raised its 2026 capital-spending plan to $180-190B, most of it chips and buildings.

What the constraint reveals is a shift in the scarce input. For the firms building frontier AI, it is no longer cash — Alphabet has plenty, and is raising $80B more — but the physical supply of accelerators and the power and land to run them. A near-doubled backlog means most of this growth is already contracted; the limit on cashing it in is how fast concrete gets poured and racks get filled. When the richest companies in the world are gated by supply chains rather than balance sheets, the bottleneck for the whole AI build-out moves from spreadsheets to construction sites.