Dell Books $60.9B in AI Server Orders, Backlog Hits $95B
3 min readDell Technologies booked $60.9 billion in AI server orders in a single quarter and still could not ship them fast enough. The company closed its fiscal second quarter with a record $95 billion backlog of unfilled AI server orders. That one figure says more about the state of AI infrastructure demand right now than most model launches do.
From PC Maker to AI Server Supplier
Dell spent decades known mainly for laptops and desktops. The AI buildout reshaped the company. Its infrastructure group now sells the rack-scale systems that hyperscalers, neoclouds and large enterprises use to house Nvidia and AMD accelerators, and that business has grown faster than anything else Dell owns.
The competition is real. Super Micro, HPE and a set of original design manufacturers chase the same orders, usually on thin margins. Dell’s argument has been that it wins on supply chain scale, financing and service rather than on price alone.
What Dell Reported
Dell posted record revenue of $47.0 billion for its fiscal second quarter of 2027, up 58 percent year over year, and record non-GAAP diluted earnings per share of $7.04, up 203 percent. AI server revenue doubled from a year earlier to $16.4 billion, according to Dell.
Orders outran shipments by a wide margin. The $60.9 billion in AI server orders booked during the quarter pushed backlog to a record $95.0 billion. Dell raised full-year revenue guidance by $25 billion to $192 billion and lifted its fiscal 2027 AI server revenue outlook to roughly $74 billion. The company also said it holds the top position in mainstream server revenue with 33 percent share.
Why the Backlog Matters
A backlog is not revenue. It is a promise Dell has to fund, build and deliver, and at $95 billion it represents well over a year of AI server sales at the pace the company expects. The gap between orders and shipments is the clearest public evidence that the binding constraint in AI infrastructure is no longer demand. Accelerators, high bandwidth memory, electrical power and data center space all ration how quickly those orders convert into shipped machines.
For anyone watching whether AI spending is cooling, this points the other way, with a caveat. Orders can be canceled, and a backlog concentrated in a handful of very large buyers carries real risk if any one of them pulls back. The number to watch next is conversion: how much of that $95 billion becomes shipped revenue in the coming quarters, and at what margin.
Record orders are the easy part to announce. Delivering them is the harder quarter.