Can the CPU and graphics card be powered separately?

Yes, the CPU and graphics card are powered separately within a modern desktop computer's power delivery architecture. This separation is a fundamental design principle, not an optional configuration, and is critical for system stability, performance, and safety. The central processing unit (CPU) receives its power exclusively through a dedicated set of cables—typically a 4-pin, 8-pin, or dual 8-pin connector—that plug directly from the power supply unit (PSU) into the motherboard near the CPU socket. This power rail is managed by the motherboard's voltage regulator module (VRM), which conditions the +12V supply into the highly precise, variable low voltages the CPU core requires. In contrast, a dedicated graphics processing unit (GPU) draws the bulk of its substantial power directly from the PCI Express slot on the motherboard and, more significantly, from one or more dedicated PCIe power cables (6-pin, 8-pin, or the newer 12VHPWR connector) that run from the PSU directly to the graphics card itself. These are entirely distinct electrical pathways.

The mechanism for this separation is rooted in the different electrical demands and tolerances of the components. A high-end CPU may draw over 300 watts, but it requires extremely tight voltage regulation and rapid power state transitions. Its dedicated rail allows the VRM to respond to these microsecond-scale load changes without interference from other components. A high-end graphics card, which can draw 450 watts or more, presents a massive but somewhat less dynamically volatile load. By providing it with its own direct cables from the PSU, the design prevents the GPU's sudden power surges—common during intensive gaming or rendering—from causing voltage droops or instability on the CPU's sensitive power plane. This isolation is crucial; if both components shared a single high-current rail, the electrical noise and transient loads from the GPU could easily cause the CPU to throttle or crash under load.

The practical implication is that a PSU must have sufficient independent +12V rails or a single robustly designed +12V rail with the requisite number of physical PCIe and CPU power connectors to support the combined peak draw. This is why power supply specifications explicitly list the number of CPU/PCIe connectors and the combined +12V wattage or amperage. For the user, this means you cannot "share" a cable between them; you must use the specifically labeled cables. Attempting to power a high-end GPU with insufficient or daisy-chained connectors, or using a PCIe cable in the CPU header (which is physically keyed to prevent this), risks catastrophic failure from overheating cables or component damage. The separation thus enforces a necessary discipline in system assembly.

Ultimately, this divided power scheme is a non-negotiable aspect of modern PC design, scaling from entry-level systems to extreme workstations. It allows each subsystem to operate at its electrical optimum without compromising the other, enabling the high, sustained power delivery required for performance computing. The trend toward even higher GPU power consumption, with cards now requiring up to three or four separate 8-pin connectors or the new 12VHPWR standard, only reinforces this architectural necessity. The answer to whether they can be powered separately is therefore definitive: they are, and they must be.