More people get this wrong than any other part of a PC build. They spend carefully on a motherboard, deliberate for hours over CPU options, then either massively overspend or underspend on the graphics card. Since the GPU is what actually determines your gaming experience, that mismatch costs you performance you already paid for.

The Rule That Matters Most

For a gaming PC, around 40% of your total budget should go toward the GPU. This isn't arbitrary. It reflects where the performance actually comes from in gaming workloads, and it's been tested across thousands of real builds.

On a $1,000 USD build, that puts your GPU budget around $400. On a $1,500 build, around $600. On a $2,500 build, around $1,000.

This doesn't mean you must hit exactly 40%. It means the GPU allocation is the anchor. Everything else in the build gets sized relative to it. The CPU takes about 17%. The motherboard around 11%. RAM and PSU each take about 8%. Storage 7%. Case 5%. Cooler 4%. These percentages exist to keep the whole build proportional so nothing is a weak link.

The most common mistake is underspending on the GPU to afford a "better" motherboard or more RAM. You end up with a $300 GPU and a $250 motherboard when that same $550 should have gone toward a $450 GPU and a $100 board. The GPU would have performed better and the experience would have been noticeably different.

What Your GPU Budget Gets You

GPU pricing moves constantly, but the tiers are fairly stable in terms of what you get for the money.

Under $300 USD

This range covers last-gen cards and the lower end of current-gen. You're looking at solid 1080p gaming at medium to high settings. Competitive titles at high frame rates are doable. Demanding open-world games at maximum settings are not. Good if 1080p is your target and you're not chasing ultra settings.

$300 to $500 USD

The sweet spot for most builders. You get strong 1080p at high to ultra settings and solid 1440p at medium to high settings. Cards in this range typically have 8GB to 12GB of VRAM, which covers most games at these resolutions. RTX 4070, RX 7800 XT, and their AMD/Nvidia equivalents typically land here depending on sales and availability.

$500 to $800 USD

1440p at high to ultra, with 4K becoming viable at medium settings. VRAM starts at 12GB here. If 1440p at high refresh rates is your goal, this range delivers it cleanly. The performance gains over the $300-500 range are real, not marginal.

Above $800 USD

4K territory. High-end RTX 4080, RTX 4090, RX 7900 XT class. These are cards where you're paying for 4K at high refresh rates or pushing very high frame rates with ray tracing at 1440p. The performance is real. So is the diminishing return per dollar compared to lower tiers.

Resolution Changes Everything

Your monitor resolution determines which GPU tier makes sense. Buying a $700 GPU for a 1080p monitor is wasted money. Buying a $250 GPU and pairing it with a 4K monitor means you'll be gaming at low settings to hit playable frame rates.

If you don't have a monitor yet, pick your target resolution first, then buy a GPU that matches it, then build the rest of the system to support that GPU.

VRAM: The Number People Ignore Until It's Too Late

VRAM is the memory on the GPU itself. Games store textures and assets there during rendering. When you run out of VRAM at a given resolution and settings, frame rates drop hard and often stutter badly.

At 1080p, 8GB is workable but tight in newer titles. At 1440p, 8GB becomes a limitation in more demanding games. 12GB is the comfortable minimum for 1440p. At 4K, 16GB or more is where you want to be.

This matters when comparing cards at similar prices. A $350 card with 8GB and a $380 card with 12GB are not equivalent for a 1440p build. The VRAM difference will matter more as games get heavier.

New vs. Used GPUs

Used GPUs are worth considering if you know what you're buying. The main risk is that the card was used for cryptocurrency mining, which puts sustained load on the GPU for long periods and degrades it faster. If a used card is priced more than 30-40% below the new equivalent, the reason might be condition.

Buying used from a private seller with a receipt or through a platform with buyer protection is far safer than buying from an unknown source. A card that was used for normal gaming for two to three years is generally fine. A mining card is harder to assess.

If budget is tight, a previous-generation used GPU from a known source at a good discount is a reasonable choice. Previous-gen cards at lower prices can outperform current-gen cards at higher prices depending on the comparison.

AMD vs. Nvidia: Does It Matter?

At equivalent prices, AMD and Nvidia trade blows. For raw rasterization performance (how games actually look most of the time), AMD has been competitive with Nvidia at the mid-range for several generations now.

Nvidia has advantages in ray tracing performance, DLSS (its upscaling tech), and software ecosystem maturity. AMD's FSR upscaling works on any GPU but doesn't match DLSS quality at the same settings.

If you're buying at the same price tier, benchmark the specific models you're comparing. Don't default to Nvidia out of brand loyalty if an AMD card at the same price performs better in the games you play.

The Simple Version

Allocate 40% of your total PC budget to the GPU. Match the GPU tier to your target resolution. Don't buy more VRAM than you need, and don't buy less than your resolution requires. Everything else in the build exists to support the GPU decision you make first.

If you're not sure where to start, use our budget calculator to see the full breakdown based on your total budget.