Methodology

Primary sources, one link per number, a date on every row.

Where the data comes from

We read each maker’s own product page, datasheet, user guide or press release first, and trade press only where the maker prints nothing. Every record lists its sources with the date we accessed them, and every number in the table links to the source it came from.

How we verify

Each row shows the date we last checked it. A field is filled only when a source states it; when a maker does not publish a figure, the cell stays empty.

What a row is

One row is one named server model from one maker, as listed on that maker’s own product page, datasheet or user guide. Makers include OEMs (Dell, HPE, Lenovo, Cisco), ODMs (Wiwynn, Wistron, Inventec, Ingrasys, QCT, MiTAC) and system builders. The OEM and ODM labels are our editorial grouping by how each maker mostly sells, not a vendor-declared category. When a model supports several GPUs, the row names the option the page was read for and says so. Counts are per server unless the row is a rack-scale system, where they cover the whole rack.

Derived fields

GPU vendor and cooling type are read from the GPU model and cooling wording on the same source. Rack power and system power are the maker’s printed figures only; we never add up power supplies or estimate draw. Makers that publish no kW figure leave the cell empty.

Corrections

Vendors and readers can send corrections to hello@aiserverlist.com with a public source. We update the row and its verified date. Featured placement is paid and labelled; it never changes a row’s data or its position in the sorted table.

Reuse

The dataset is free to reuse under CC BY 4.0 with a link to aiserverlist.com. Download it as CSV or JSON.

Field definitions

26 fields
Maker
The OEM, ODM or system builder that sells the server model. Corporate names follow the vendor’s own pages.
GPU vendor
Maker of the accelerators named in the GPU model field: NVIDIA, AMD or Intel. Blank when the vendor page names no accelerator.
GPU or accelerator
Accelerators the vendor page names for this model. A model that supports several options lists the option that was read, marked as one supported option.
GPUs per server GPUs
Installed count or supported maximum per server, as the qualifier in the source says. Rack-scale systems show the GPUs in the whole rack (72 for an NVL72). Blank when the page gives no count.
Chassis height U
Rack height of the server as the vendor states it. Open Rack Units (OU) are not the same as EIA U and sit in the form factor field instead.
Form factor
Configuration-specific chassis or rack description where the vendor splits it by cooling or rack standard.
Cooling type
Air, liquid, or air or liquid by configuration, read from the vendor’s cooling statement. Blank when the page does not state cooling.
Cooling (vendor wording)
The vendor’s cooling statement with configuration qualifiers kept.
CPU support
Processor families or installed processors the named model supports.
CPU sockets sockets
Processors per server.
CPUs per rack CPUs
Total CPUs in a rack-scale configuration.
Memory per GPU GB
Memory on each accelerator in the cited configuration.
Total GPU memory GB
Vendor-stated total accelerator memory per system.
Max system memory TB
Vendor-stated maximum host RAM, not GPU memory.
Memory slots DIMMs
Host memory DIMM slots.
GPU TDP W
Thermal design power per GPU, not measured server draw.
System power kW
Maximum power the vendor states for one system, as printed on its own page.
Rack power kW
Power the vendor states for a full rack-scale system.
Power supplies
PSU ratings, quantity and redundancy where published. Rating, not draw.
Rack power delivery
Power-shelf rating in rack-scale and rack-sled systems. Delivery capacity, not rack demand.
GPU interconnect
GPU fabric and published link bandwidth.
Networking
Published ports, adapters and speeds.
Source date
Publication date printed on the cited page, where the vendor prints one.
Warranty
Base warranty or service statement the vendor publishes.
Operating temperature °C
Supported ambient range with the vendor’s caveats.
Maker headquarters
Headquarters or principal corporate office of the maker, not a factory location.