850W. That is NVIDIA’s required system power for the RTX 5080, a card rated at 360W, and for a normal build it is the number to buy rather than a floor to argue with. The variable that catches people out is not the processor — it is which 5080 you bought. A factory-overclocked board draws roughly 60W more than a Founders Edition, and at the top of the CPU range that difference is worth an entire power-supply tier.

The card’s actual power specification

SpecificationRTX 5080
Total graphics power (TGP)360W
NVIDIA required system power850W
Power connectors3x PCIe 8-pin (adapter in box), or one 450W-or-greater PCIe Gen 5 cable

Both connector figures moved up from the RTX 5070, which asks for two 8-pins or a 300W Gen 5 cable. That matters on an upgrade: an older supply with two spare 8-pin cables may not have a third, and a Gen 5 cable rated for a 5070 is not necessarily rated for this card. Check the cable, not just the wattage on the box.

Which 5080 did you buy?

This is the question that decides your answer, and it is specific to this tier. Summing component draw the way our PSU calculator does — 45W for an ATX board, 5W per RAM stick, 8W for a Gen4 M.2, 2W per fan, 12W for an AIO, plus a 10W miscellaneous allowance, so 93W for a reference build of two sticks, one drive, four fans and an AIO — the two variants land like this:

Your processor360W Founders-class card~420W factory-OC card
65W518W peak → 650W578W peak → 750W
120W573W peak → 750W633W peak → 850W
170W623W peak → 750W683W peak → 850W
250W703W peak → 850W763W peak → 1000W

Sizes are peak draw plus a 20% margin, rounded up to a wattage actually sold — the smallest unit that works, in the calculator’s language, not the most comfortable one. As on the 5070 page, one tier above each figure is the quieter, longer-lived choice. Read across any row: the overclocked board costs you a tier on its own — same silicon, different power target.

Our own reference card is the overclocked kind, and we have logged it. Across a 25-minute gaming session on the ASUS ROG Astral RTX 5080 OC — 308 HWiNFO64 samples on the hand-tuned profile — board power averaged 261W and peaked at 326W. That is comfortably under the ~420W the calculator uses for an overclocked board, which is the point: you size a supply for the worst case it must survive, not for the average it will usually see.

Notice that our arithmetic for a Founders-class card with a mainstream CPU lands at 750W, below NVIDIA’s 850W. That is not a gap to exploit. Vendor minimums tend to be worst-case, and NVIDIA’s note on the RTX 5090 is the one place they show their assumption — that figure is based on a system built around a Ryzen 9 9950X, a flagship chip. Read their 850W as covering the bottom of that table rather than the middle, and buy it.

When 850W becomes 1000W

Three things push you up, and they stack:

  • A factory-overclocked board — roughly 60W over a Founders-class card.
  • A processor at or above 200W. A Ryzen 9 9950X sits right around 200W by our calculator’s figure; an unrestricted Core i9 goes well past it and lands you in the bottom row.
  • Manual overclocking or a raised power limit, which moves the card past its stock target entirely.

Any two together make 1000W the sensible buy. Whether a given processor limits your frame rate rather than your power budget is a different question — the PC bottleneck calculator covers that one.

Transient spikes, honestly

A 360W card draws in bursts rather than a flat line, and this is where PSU advice usually becomes “buy double.” It should not.

Intel’s ATX 3.0 design guide requires a compliant unit to survive excursions to 200% of its rated wattage for 100 microseconds, with lower brackets for longer events — but only where both conditions hold: the unit is above 450W and ships a native 12VHPWR / 12V-2x6 connector. An 850W ATX 3.x supply with that cable meets the bar; one without the native cable does not, whatever its wattage.

So on a compliant modern unit, size for sustained peak plus the headroom in the table and stop padding for spikes. On an older ATX 2.x supply paired with a 5080, leave more room — it was never designed against those brackets. The full methodology is published on the PSU calculator.

The undervolt that can buy a tier back

The 5080 sits right on a tier boundary, which makes undervolting worth more here than the raw watts suggest. Curve undervolting a 5080 is widely reported to cut 30–70W depending on where your silicon lands, and our own RTX 5080 undervolt guide uses that published range rather than a single number, because the spread between chips is the whole point.

Against the table above, even the middle of that range matters. A ~420W overclocked card with a 120W processor peaks near 633W and wants 850W; take 50W off and it peaks near 583W, which a 750W unit covers.

The honest caveat is that you cannot bank it in advance. Silicon varies, the range is wide, and a PSU has to cover the case where your card undervolts poorly. Buy for the stock figure, undervolt afterwards, and treat the saving as headroom you gained rather than money you saved. Our undervolt results database lists what specific cards achieved, with the source behind each row.

If you are coming from a smaller card

Sizing for an RTX 5070 is a different exercise — 250W, 650W of system power, two 8-pins — and it is covered in what PSU an RTX 5070 needs. Moving from a 5070 to a 5080 is genuinely a power-supply purchase as well as a card purchase, worth budgeting before the card arrives rather than after. For where both land on raw speed, see the GPU hierarchy.

Sources

  • NVIDIA GeForce RTX 5080 official specifications — total graphics power, required system power, and supplementary power connectors — RTX 5080 specifications, verified live 2026-07-25.
  • NVIDIA GeForce RTX 5070 specifications, for the upgrade comparison — RTX 5070 family, verified live 2026-07-25.
  • Intel ATX 3.0 desktop power supply design guide — the excursion brackets and the wattage/connector conditions that qualify a unit (Intel ATX 3.0 PSU power excursion).
  • NVIDIA GeForce RTX 5090 specifications — the note stating its minimum-system-power figure is based on a system configured with a Ryzen 9 9950X — RTX 5090, verified live 2026-07-25.
  • The ~420W figure for a factory-overclocked 5080 is this site’s own sizing estimate, taken from the GPU list in the PSU calculator; it is a conservative peak for sizing, not a measured gaming draw.
  • The 261W average / 326W peak board-power figures are our own first-party telemetry: 308 HWiNFO64 samples across a 25-minute gaming session on the hand-tuned profile, published in the ROG Astral RTX 5080 OC review with the raw dataset.
  • The 30–70W curve-undervolting range is the published community range cited in our RTX 5080 undervolt guide.
  • Component draw coefficients and sizing rules are this site’s own published figures, listed in full in the PSU calculator methodology.

Frequently asked questions

What are the RTX 5080's PSU requirements?
NVIDIA lists 850W of system power for the RTX 5080, which is a 360W card, and the board needs either three PCIe 8-pin cables — an adapter ships in the box — or a single PCIe Gen 5 cable rated 450W or higher. That is a step up from the RTX 5070 on both counts, so a supply that comfortably ran a 5070 is not automatically enough here.
What power supply do I need for an RTX 5080?
850W is the number to buy for a normal build. The variable most people miss is the board rather than the processor: a Founders-Edition-class card is 360W, while factory-overclocked models pull nearer 420W under sustained load, and that 60W is enough to move a flagship-CPU build up to 1000W. Overclocked card plus a 200W-plus processor means size for 1000W.