Loading market event…
Loading market event…
MajorMarket eventTue, Oct 201 min read
Hardware spending can run ahead of cloud revenue when electricity, rather than chips, limits usable AI capacity.
Amazon's cloud business, AWS, is reportedly storing some Nvidia AI systems elsewhere because its data centers lack the power to turn them on.[1]
The report says AWS expected to receive $5 billion of Nvidia GB300 rack systems, followed by another $3 billion each month; these systems package AI computing chips and supporting equipment into racks.[1] Those figures describe expected deliveries, not the value of equipment already sitting idle.[1]
After the equipment arrives, AWS still needs enough electricity before it can sell the additional AI computing capacity to customers. Systems in storage cannot add usable capacity, so a longer gap between purchase and activation delays the revenue that can justify the investment.[1]
That separates the business outlook for equipment suppliers from that of cloud operators: expected deliveries indicate hardware demand, but do not prove that the corresponding cloud capacity is in service. A persistent power gap could force operators to adjust future purchasing and deployment schedules; the report does not establish that AWS has cut orders.[1]
AWS must resolve the power constraint before stored equipment can become a billable service. For rival cloud providers, access to power could also shape delivery competitiveness, potentially favoring operators with powered capacity available; this report does not establish that rivals face the same shortage.[1]
AWS's expected continuing deliveries support demand for Nvidia and related equipment suppliers. But the gap between delivery and activation means supplier shipment growth is not the same as growth in customers' usable computing capacity. Persistent power constraints could affect later order timing, though no cuts are established here.[1]