PSU Wattage Guide: How Many Watts Does Your PC Need?

Most gaming PCs need somewhere between 550 and 1,000 watts of power supply capacity, and the single biggest factor in that number is your graphics card: a budget GPU like the RTX 5060 asks for a 550W system, while a flagship RTX 5090 asks for 1,000W, according to Nvidia's own published specifications. The right way to size a power supply unit (PSU) isn't to guess or round up blindly — it's to add up what your actual components draw, build in headroom for spikes and aging, and then pick a unit with the efficiency certification and connector standard that match a modern build.

This guide walks through all of it: how to estimate your system's total wattage, why GPUs from Nvidia's RTX 50-series and AMD's RX 9000-series have pushed recommended PSU wattage noticeably higher than a few years ago, what the 80 Plus efficiency tiers actually guarantee, and why ATX 3.0/3.1 and the 12V-2x6 connector matter if you're running (or buying) a high-wattage graphics card.

Quick answer: PSU wattage by build tier

If you just want a number to work from, here's a build-tier breakdown using the official "required system power" figures GPU makers publish for their own cards, plus a practical PSU size that leaves headroom above that bare minimum.

Those "required system power" figures aren't the card's own power draw — they're Nvidia and AMD's recommendation for the whole PC, including a high-end CPU, drives, and fans, which is why they already run well above each GPU's own rated board power. The "practical PSU to buy" column simply adds a further margin, which the next section explains.

How to estimate your system's total wattage

The most reliable way to size a PSU is a rough bottom-up add: list your major components, note their power draw, sum it, then add headroom. You can check exactly which GPU you're running with our guide on how to check what graphics card you have on Windows, Mac and Linux if you're not certain which card (and therefore which power figure) applies to your machine.

• GPU: By far the largest single draw in a gaming PC, and the component whose published figure should anchor your whole calculation. Use the "Total Graphics Power" or "TGP"/"TBP" number from the manufacturer's own spec page — not a retailer's estimate.
• CPU: A high-end desktop CPU under full load adds a meaningful chunk on top of the GPU, which is one reason Nvidia's own recommended-PSU figures are based on test systems built around a high-end Ryzen 9 CPU rather than the GPU in isolation — the published "required system power" numbers already have CPU headroom baked in.
• Storage and RAM: Individually small (a NVMe SSD typically draws only a few watts), but a build with several drives and a full set of RAM slots populated adds up. If you're also deciding how much memory to install, our RAM sizing guide covers the trade-offs — more RAM has only a minor effect on power draw compared to the GPU and CPU.
• Fans, pumps and RGB: Case fans, an AIO liquid cooler pump, and lighting are each only a few watts apiece, but a heavily accessorized build can add 20–40W collectively.
• Motherboard and peripherals: The board itself, plus anything powered through USB (and overhead for VRM inefficiency), typically adds another modest draw.

In practice, this is exactly why GPU makers' own "required system power" figures are the easiest starting point rather than a from-scratch spreadsheet: they're built around a realistic high-end system, not just the card. Start there, then add your own margin if you're running multiple drives, a heavily overclocked CPU, or several case fans beyond a typical build.

Why modern GPUs have pushed PSU wattage up

If you built a gaming PC five or six years ago, a 650W unit was plenty for a high-end build. That's no longer true at the top end, and the reason is squarely the graphics card. Nvidia's current RTX 50-series lineup illustrates the jump clearly, based on the company's own published specifications:

AMD's current flagship follows the same pattern. The Radeon RX 9070 XT's own product page lists a 304W typical board power and a 750W minimum PSU recommendation — roughly in the same bracket as Nvidia's RTX 5070 Ti, and well above what a 300W-class card would have required a generation or two ago. If you're deciding between the two camps at this tier, our RTX 5080 vs. RX 9070 XT comparison breaks down the rest of the specs and pricing.

Two things are driving this upward march in GPU power, beyond simply "faster chips use more electricity":

• Transient power spikes. High-end GPUs don't draw a smooth, constant wattage — their draw can spike to roughly double the card's rated power for fractions of a millisecond under certain loads. A PSU that's sized exactly to a card's average draw, with no margin, can trip its own protection circuitry during these spikes even though the average load is well within spec. This is part of why the ATX 3.0 and 3.1 specifications explicitly require PSUs to tolerate larger, faster transients than older standards did.
• Diminishing efficiency gains per generation at the high end. As flagship GPUs push for maximum raw performance, the top of the lineup trades some power efficiency for throughput, which is why the gap between a maker's entry card and its flagship (145W vs. 575W TGP, in Nvidia's current lineup) has widened rather than narrowed. If you're pricing out a flagship card to pair with a new PSU, see our breakdown of what the RTX 5090 actually costs today.

Why you need headroom above the bare minimum

A PSU rated for exactly your system's peak draw is a bad plan for three reasons that have nothing to do with marketing.

Transient spikes. As noted above, momentary spikes in GPU draw can exceed the card's average rating by a wide margin for microseconds at a time. A PSU with no margin above your calculated total is more likely to hit its over-power or over-current protection and shut down mid-game, even though your system's average draw never got close to the unit's rated capacity.

Efficiency curves aren't flat. Every PSU is most efficient somewhere in the middle of its load range — typically around 50% load — and measurably less efficient very close to its maximum rated output. Running a PSU near 100% of its capacity as a matter of routine wastes more power as heat, runs the unit hotter and louder, and accelerates component aging.

Capacitors degrade over years, not days. A PSU's true maximum output capacity drifts downward slightly as its internal capacitors age, especially if it regularly runs hot. A unit bought with 10–20% of headroom on day one still has comfortable margin three or four years later; a unit bought at the bare minimum may not.

The practical rule that follows from all three: take whichever "required system power" figure your GPU maker publishes, and treat it as a floor rather than a shopping target. If Nvidia says 850W for your card, a quality 1000W unit — not an 850W one — is the one that will still feel comfortable in year three.

80 Plus certification: Bronze, Gold, Platinum, Titanium

Efficiency certification is a completely separate decision from wattage, and it's worth not conflating the two. The 80 Plus certification program grades a PSU on how much of the AC power it draws from the wall is actually delivered to your components as usable DC power, versus lost as heat, at several load levels. A more efficient PSU doesn't let you run a bigger GPU — it just wastes less electricity (and generates less heat) doing the job a given wattage rating already implies.

In practice, for most gaming builds the calculus is simple: Gold is the sweet spot most reviewers and system builders land on — it's efficient enough that heat and noise aren't a real concern, without the price premium of Platinum or Titanium, which tend to matter more for systems that run near-constant heavy loads (workstations, servers, 24/7 rigs) than for a PC that's gaming a few hours an evening. Bronze is an acceptable floor for a budget build, but at the 850W-plus tier that a high-end GPU now demands, the efficiency difference between Bronze and Gold translates into a real amount of waste heat inside your case, which is worth avoiding even before you factor in the electricity cost over the PSU's lifespan.

ATX 3.0/3.1 and the 12V-2x6 connector

Wattage and certification tell you how much power a PSU can deliver and how efficiently — but for any GPU pulling meaningfully more than 300W, the connector standard matters just as much. Modern high-wattage GPUs are increasingly built around a single 16-pin power connector rather than two or three separate 8-pin PCIe cables, and the specification governing that connector has gone through a notable revision.

Per Wikipedia's ATX specification reference, the ATX 3.0 standard (2022) introduced the 16-pin 12VHPWR connector, rated to deliver up to 600W over a single cable, and required PSUs to tolerate power transients of up to twice their rated output for very short durations — a direct response to the spiky power behavior of high-end GPUs described earlier in this guide. ATX 3.1 then replaced 12VHPWR with a revised connector called 12V-2x6, which keeps the same 600W ceiling and backward mechanical compatibility, but reshapes the sense pins so a loose or partially seated cable is far less likely to go undetected — a problem that drew real scrutiny on early 12VHPWR implementations.

What this means practically when you're shopping for a PSU to pair with a current high-end GPU:

• If your GPU's board power is comfortably under roughly 250–300W, a traditional PSU with standard 8-pin PCIe connectors (via the adapter many cards still ship with) is generally fine.
• If you're buying a card in the RTX 5080/5090 or RX 9070 XT class, look specifically for a PSU advertised as ATX 3.0 or ATX 3.1 compliant with a native 12V-2x6 (or 12VHPWR) cable from the factory, rather than relying solely on a multi-8-pin-to-16-pin adapter, which is a more common point of failure than a cable built for the connector from the start.
• ATX 3.0/3.1 compliance is specifically about transient tolerance and connector design — it isn't a replacement for an adequate wattage rating or a reputable 80 Plus certification. You want all three: enough watts, a solid efficiency tier, and the right connector standard.

Signs your current PSU may be undersized

If you're not building new but wondering whether an existing PSU is holding you back, a few symptoms are worth taking seriously:

• Random shutdowns or reboots under heavy GPU load — particularly during demanding games or benchmarks, rather than at idle — are the clearest sign of a PSU hitting its protection limits.
• Instability that goes away when you cap your frame rate or power limit suggests the issue is peak draw, not an unrelated software problem.
• A PSU that's several years old paired with a GPU upgrade is worth double-checking against the new card's official required-system-power figure, even if the old unit handled your previous GPU without issue — both the card's demands and the PSU's real-world capacity have likely changed since you bought it.
• Coil whine or unusual fan ramp-up under load isn't proof of an undersized PSU on its own, but combined with either symptom above, it's worth checking the unit's rated wattage against your GPU's current official recommendation.

None of these symptoms are exclusively a PSU problem — driver issues, an unstable overclock, or a failing GPU can look similar — but a PSU rated below (or barely at) your GPU's official minimum is the first thing worth ruling out before chasing more exotic causes.

Frequently asked questions

Do I need to match my PSU wattage exactly to my GPU's rating? No — and you shouldn't try to. Buy above the manufacturer's stated "required system power" figure, not exactly at it, to cover transient spikes and long-term headroom.

Does a higher-wattage PSU use more electricity even when idle? A PSU only draws as much power as your components actually demand at any moment; a bigger unit doesn't waste power sitting idle. What matters for your electricity bill is efficiency (the 80 Plus tier), not the unit's maximum rated capacity.

Is 80 Plus Gold enough for a high-end gaming PC? Yes, for the vast majority of builds. Platinum and Titanium mainly pay off for systems under near-constant heavy load, like workstations or always-on servers.

Can I use an 8-pin-to-12VHPWR adapter instead of buying a new PSU? It will often work, but a PSU with a native 12V-2x6/12VHPWR cable from an ATX 3.0/3.1-compliant unit is the safer long-term choice for a high-wattage GPU, since adapters are a more common point of marginal or loose connections than a cable built for the connector.

Why do official PSU recommendations seem higher than my GPU's actual power draw? Because they're sized for the whole system, not just the card — typically modeled around a high-end CPU alongside the GPU — and they build in margin for transient spikes rather than just average draw.

Written byMara Lindqvist

Mara Lindqvist edits the hardware desk. She covers graphics cards, processors, memory and storage, the foundries and chip designers behind them, and what the numbers on a spec sheet mean for people choosing a PC. Specifications in her stories come from manufacturer spec pages and datasheets.

More from Mara LindqvistAll Hardware coverage