CPU Integrated Graphics vs Discrete GPU: When the iGPU Is Enough
Integrated graphics used to mean "can't play games at all." That's no longer accurate, but the honest answer to what an iGPU can handle depends heavily on which specific games, what resolution, and what settings — not a single yes-or-no.
Integrated graphics shares two resources with the rest of the system that a discrete GPU has entirely to itself: memory bandwidth and the CPU's total power budget. A discrete GPU has its own dedicated VRAM with bandwidth measured in hundreds of gigabytes per second, isolated entirely from system RAM. Integrated graphics instead draws from the same system memory the CPU itself is using, over a shared memory controller, meaning the iGPU's effective bandwidth is whatever's left after the CPU's own memory demands are satisfied, and that shared bandwidth is a fraction of what a discrete GPU's dedicated memory provides.
Why memory speed matters more for iGPU than for a discrete-GPU system
Because integrated graphics performance is bandwidth-constrained in a way discrete GPUs aren't, memory speed and configuration have an outsized effect on iGPU gaming performance specifically. Running in dual-channel rather than single-channel mode, and running faster-rated memory (particularly relevant on platforms with strong iGPU implementations that scale well with memory speed), can produce meaningfully larger iGPU frame rate improvements than the same memory upgrade would produce on a system relying on a discrete GPU, where system memory bandwidth barely factors into GPU-bound performance at all. This is one of the few scenarios where dual-channel memory configuration produces a dramatically larger, more consistent performance swing than it does for discrete-GPU gaming.
What current integrated graphics can actually handle
Modern integrated graphics on current-generation CPUs has advanced enough to run many esports titles and older or less demanding games at 1080p with reduced settings at frame rates that are genuinely playable, and can handle general desktop use, video playback including hardware-accelerated 4K HDR content, and light creative work without any issue at all. Where it consistently struggles is with recent AAA titles at native resolution with high settings, ray tracing of any kind (which is punishingly bandwidth- and compute-intensive even for many discrete GPUs), and any workload that benefits from dedicated VRAM capacity beyond what a shared memory pool can reasonably spare without starving the CPU side of the system.
Driver support is the other practical gap: integrated graphics driver update cadence and per-game optimization profiles have historically lagged behind discrete GPU driver support from the same vendors, meaning a brand-new game release is more likely to have day-one optimization for popular discrete GPUs than for integrated graphics, even when the iGPU's raw compute capability might theoretically be adequate for the title at reduced settings.
Where the honest line sits
For a general productivity, web browsing, and media PC with no gaming ambitions, integrated graphics is not just adequate but genuinely well-suited to the task, and adding a discrete GPU to such a build is largely wasted expense and wasted power draw. For light or older-game use at 1080p with realistic setting expectations, current integrated graphics can deliver a legitimately playable experience, particularly in esports titles that are deliberately built to run on modest hardware. For anything approaching current AAA titles at competitive frame rates, any use of ray tracing, or serious creative work involving GPU-accelerated rendering or video editing, a discrete GPU remains necessary, not because integrated graphics is bad at what it does, but because the shared-memory architecture and smaller execution unit count are fundamental constraints that no amount of driver optimization removes.
The upgrade path question
A system built around integrated graphics with the intent to add a discrete GPU later is a reasonable plan, but it's worth confirming the platform's power supply and case have enough headroom for a card that will realistically be needed once games or requirements outgrow the iGPU, since a PSU sized tightly for a CPU-only system may need replacing anyway rather than simply having a GPU dropped in. Motherboard PCIe slot availability is rarely a concern on modern boards, but checking case clearance and PSU wattage against a plausible future GPU choice avoids a second, larger unplanned expense down the line.