# Data-centre server: Working

Source page: https://www.materialinsights.org/products/p_data_centre_server/prod-working/

#### What it is

The product and the job it does, in a short paragraph

A data-centre server is a computer built to run all day in a rack, serving many users and programs at once. This node covers rack servers in two variants: general-purpose servers with processors only, and AI accelerator servers that add GPUs. The functional unit is one two-socket rack server. Racks, network switches, storage arrays and the building are left out. Reference machines are the Dell PowerEdge R740 at 26.3 kg ^[2], the Lenovo ThinkSystem SR650 at 18.47 kg ^[4] and the NVIDIA DGX H100 at 130.45 kg ^[8].

as of 2026-11-16

#### How it works

How it does its job, step by step in plain words

Two processor sockets hold the CPUs, which run programs and share a bank of DRAM memory modules. Drives keep data and network cards link the server to the rest of the rack. A motherboard joins these parts. Two power supplies turn mains electricity into the low voltages the chips need and give redundancy if one fails. Fans pull air through the chassis to carry heat away, and management firmware lets operators control the machine remotely. An AI accelerator server adds GPUs on a baseboard: the DGX H100 has eight H100 GPUs with 640 GB of GPU memory, two Intel Xeon CPUs and 2 TB of system memory in 8U of rack height ^[8]. Nearly all the electricity ends as heat. The physics of data centres is on the twin technology pages.

as of 2026-11-16

#### Key specifications

The numbers that describe it, each with unit, variant, year and source

| Spec | Value | Unit | Variant | Year | Source |
|---|---|---|---|---|---|
| Mass of a Dell PowerEdge R740 rack server (two CPUs, 32 GB DRAM, three 2.5-inch drives) | 26.3 | kg | General-purpose | 2019 | ^[2] |
| Mass of a Lenovo ThinkSystem SR650 rack server (Lenovo's representative rack server) | 18.47 | kg | General-purpose | 2019 | ^[4] |
| Typical yearly energy of a Dell PowerEdge R740 (typical energy consumption, EU) | 1760.3 | kWh per year | General-purpose | 2019 | ^[2] |
| Idle power, high-end configuration of a Lenovo SR650 V3 | 326.5 | W | General-purpose | 2023 | ^[7] |
| Maximum power, high-end configuration of a Lenovo SR650 V3 | 1335.4 | W | General-purpose | 2023 | ^[7] |
| Power supply efficiency at 50% load, Lenovo DPS-1800AB-11 in a ThinkSystem SR650 V3 | 96.08 | % | General-purpose | 2023 | ^[7] |
| Mass of an NVIDIA DGX H100 (eight GPUs, 8U), maximum | 130.45 | kg | AI accelerator | 2026 | ^[8] |
| Maximum power of an NVIDIA DGX H100 | 10.2 | kW | AI accelerator | 2026 | ^[8] |
| Average power measured while an 8-H100 node trained a model | 7.9 | kW | AI accelerator | 2024 | ^[11] |

as of 2026-11-16

#### Variants

The variants whose make-up or footprint differs a lot, what defines each and its share of units

| Variant | Defining spec | Share | Year | Source |
|---|---|---|---|---|
| General-purpose | Two CPU sockets, DRAM, a few drives and two power supplies, air cooled; mass about 18 to 26 kg and about 1.7 to 2.0 MWh a year (Dell's typical yearly energy for an R740 and four-year use of an R760) ^[4]^[2]^[1] | 83 | 2026 | Calculated as 100% minus the 17% share of AI servers that TrendForce forecasts for 2026 ^[14] |
| AI accelerator | Adds eight GPUs on a baseboard to two CPUs; the DGX H100 weighs up to 130.45 kg and draws up to 10.2 kW ^[8] | 17 | 2026 | TrendForce forecast of AI servers' share of overall server shipments ^[14] |

as of 2026-11-16

#### Limits

The physical limits set by science and the engineering limits of today's design, with today's value

| Limit | Kind | Limit | Today | Unit | Source |
|---|---|---|---|---|---|
| Power supply efficiency at 50% load (energy cannot be created, so 100% is the ceiling) | Physical | 100 | 96.23 | % | ^[6] |
| Idle power as a share of maximum power, high-end SR650 V3 (calculated: 326.5 W over 1335.4 W) | Engineering | – | 24.5 | % | ^[7] |
| Maximum power of an 8-GPU server in 8U, cooled by 1105 CFM of air front to back | Engineering | – | 10.2 | kW | ^[8] |

as of 2026-11-16

#### Science it relies on

The science principles it works by or is limited by
 Relies on

- [Boolean logic](https://www.materialinsights.org/science/boolean_logic/) The processors and memory work by switching signals on and off to carry out logic and store bits.
 Limited by

- [Second law of thermodynamics](https://www.materialinsights.org/science/second_law_thermodynamics/) Almost all electrical power ends as low-grade heat that fans and cooling must carry away, which caps the power of a rack.
- [Heat conduction (Fourier's law)](https://www.materialinsights.org/science/heat_conduction/) Heat has to flow from chips through heat sinks and heat pipes into moving air, which sets how hot a chip may run.

as of 2026-11-16

#### Sources

1. [Life Cycle Assessment of Dell Server PowerEdge R760 (report produced September 2025)](https://www.delltechnologies.com/asset/en-us/products/servers/briefs-summaries/poweredge-r760-life-cycle-assessment.pdf) Dell Technologies, 2025
2. [PowerEdge R740 product carbon footprint (report produced January 2019)](https://i.dell.com/sites/csdocuments/corpcomm_docs/en/carbon-footprint-poweredge-r740.pdf) Dell Technologies, 2019
3. [Lenovo Server Product Recyclability Assessment (file dated December 2019; the document shows no date)](https://static.lenovo.com/ww/docs/sustainability/server-recyclability-assessment-dec-2019.pdf) Lenovo
4. [Lenovo ErP Lot9 Information Sheet, Servers & Storage Products: ThinkSystem SR630 V3 (Annex B2, issued 10 January 2023, updated 1 October 2023)](https://cloud.ibm.com/media/docs/downloads/environment-info-datasheets/eco-thinksystem-sr630-v3.pdf) Lenovo, hosted by IBM Cloud, 2023
5. [Technical Documentation for EU Regulation 2019/424 laying down ecodesign requirements for servers and data storage product: ThinkSystem SR650 V3](https://cloud.ibm.com/media/docs/downloads/environment-info-datasheets/SR650v3-Lot9.pdf) IBM Corporation, hosted by IBM Cloud, 2023
6. [Introduction to NVIDIA DGX H100/H200 Systems (last updated 17 October 2026)](https://docs.nvidia.com/dgx/dgxh100-user-guide/introduction-to-dgxh100.html) NVIDIA, 2026
7. [2024 United States Data Center Energy Usage Report (LBNL-2001637)](https://eta-publications.lbl.gov/sites/default/files/2024-12/us_data_center_energy_usage_report_lbnl-2001637.pdf) Lawrence Berkeley National Laboratory, 2024
8. [Strong Demand from CSPs and Sovereign Cloud to Drive Over 20% Growth in AI Server Shipments by 2026, Says TrendForce (30 October 2025)](https://www.trendforce.com/presscenter/news/20251030-12762.html) TrendForce, 2025

## Other sections

- [Summary](https://www.materialinsights.org/products/p_data_centre_server/index.md)
- [Bill of materials](https://www.materialinsights.org/products/p_data_centre_server/prod-bom/index.md)
- [Use case](https://www.materialinsights.org/products/p_data_centre_server/prod-use/index.md)
- [Supply chain](https://www.materialinsights.org/products/p_data_centre_server/prod-supply/index.md)
- [Footprint](https://www.materialinsights.org/products/p_data_centre_server/prod-footprint/index.md)
- [Evolution](https://www.materialinsights.org/products/p_data_centre_server/prod-evolution/index.md)
- [Interventions](https://www.materialinsights.org/products/p_data_centre_server/prod-interventions/index.md)
