Computer hardware is the physical equipment that makes up a computer system and can be physically touched or otherwise physically measured. It includes processors, memory modules, storage devices, motherboards, power supplies, graphics processors, network adapters, displays, keyboards, mice and other components. Hardware contrasts with software, which consists of instructions and associated digital resources that direct hardware to perform tasks.
Firmware occupies an intermediate conceptual position because it is software stored in non-volatile hardware-associated memory and is used to control or initialise physical components. Hardware is governed by physical principles involving electronics, semiconductor physics, electromagnetism, materials science, thermodynamics and mechanical engineering. Its performance is constrained by physical characteristics such as transistor switching speed, memory latency, bandwidth, energy consumption, heat dissipation and manufacturing tolerances.
The CPU is one of the most important categories of computer hardware. It contains millions or billions of transistors arranged into circuits capable of executing machine instructions. Modern processors use techniques such as multiple cores, cache hierarchies, instruction pipelines and speculative execution to increase performance. Different processor families implement different instruction-set architectures, including x86-64 and ARM-based architectures. The instruction-set architecture defines the machine-level operations software can rely on, while the microarchitecture describes how a particular processor internally implements those operations. Hardware compatibility therefore involves more than physical connection. Software compiled for one architecture may require translation, recompilation or emulation to run efficiently on another architecture, illustrating the close relationship between hardware design and software execution.
Computer hardware also includes memory and storage technologies. RAM provides fast volatile working memory, while non-volatile storage retains information without continuous power. Hard disk drives use magnetic storage, while solid-state drives use semiconductor memory. Flash memory is used in SSDs, memory cards and many portable devices. Different memory technologies occupy different points in the hierarchy of capacity, speed, persistence and cost. Processor caches are extremely fast but small, while persistent storage offers much larger capacity but generally higher latency. Hardware designers exploit this hierarchy because no single memory technology simultaneously provides unlimited capacity, minimal latency, low cost and negligible energy consumption. The organisation of memory is therefore a fundamental aspect of computer architecture.
Hardware interfaces allow components to communicate. PCI Express provides high-speed communication between processors and expansion devices such as graphics cards and storage controllers. USB supports many external peripherals, while SATA has historically been widely used for storage devices. Memory modules connect through specialised memory interfaces, and network adapters connect computers to wired or wireless networks. These interfaces use electrical signalling, defined protocols and physical connectors or integrated pathways.
Standards allow components from different manufacturers to interoperate when they implement compatible specifications. Hardware compatibility can nevertheless depend on factors such as physical dimensions, electrical characteristics, firmware support, operating-system drivers and available bandwidth. A device fitting physically into a computer does not necessarily mean that it will function correctly in that system.
Graphics hardware is a particularly important class of modern computer hardware. GPUs contain large numbers of execution resources designed for parallel processing and are highly effective for rendering graphics and performing numerical workloads with substantial data parallelism. Dedicated graphics cards can contain GPU processors, high-speed graphics memory, power-delivery circuits and cooling systems. Integrated graphics share system resources or are incorporated into a processor package. The distinction between CPU and GPU workloads has become increasingly important as applications such as scientific computing, machine learning, image processing and video production demand large amounts of parallel computation. Hardware acceleration allows particular operations to be performed more efficiently than on a general-purpose CPU alone.
Hardware also includes devices that connect computers to the physical environment. Cameras and microphones convert physical phenomena into electrical or digital information. Sensors can measure temperature, pressure, acceleration, light, position and many other variables. Actuators can convert electrical control signals into physical movement or other effects. In industrial computers, these interfaces can connect computational systems to machinery and production processes. In consumer devices, sensors provide capabilities such as orientation detection, biometric recognition and environmental measurement. Embedded systems often integrate these functions tightly into a single device. This demonstrates that computer hardware is not limited to conventional PC components; it encompasses any physical computing apparatus required to sense, process, communicate or control information.
Hardware design involves significant engineering trade-offs. Increasing performance can increase energy consumption and heat generation, requiring improved cooling and power delivery. Smaller components can improve portability but make repair and thermal management more difficult. High-capacity storage can require greater physical density or more complex controllers. Redundancy can improve reliability but adds cost and complexity. Manufacturers also balance performance against manufacturing yield, component availability, material cost and expected product lifetime. Hardware reliability can be affected by heat, vibration, moisture, electrical instability, manufacturing defects and physical wear. Preventive maintenance, cooling, surge protection and appropriate operating conditions can extend useful life, although semiconductor components can also fail unexpectedly.
Computer hardware is therefore best understood as the physical foundation of a computing system rather than simply a collection of visible parts. Hardware provides computation, memory, storage, communication, input, output, power and physical interfaces, while software supplies the instructions that make these capabilities useful. Modern hardware is increasingly integrated, with processors combining CPU cores, graphics, memory controllers and specialised accelerators in system-on-chip designs. At the same time, large computing environments combine many separate servers, storage systems and networking devices. The distinction between hardware and software remains conceptually useful, but modern systems depend on their interaction. Hardware determines what physical operations are possible and how efficiently they can be performed, while software determines how those capabilities are organised into useful computational behaviour.
